source concept and its utilization for the analysis of ... · pdf filesource concept and its...

68
Source Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav ˇ Capek Department of Electromagnetic Field Faculty of Electrical Engineering Czech Technical University in Prague Czech Republic [email protected] Habilitation – Public Lecture Prague April 12, 2017 ˇ Capek, M., Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 1 / 23

Upload: buikhue

Post on 06-Feb-2018

214 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Source Concept and Its Utilization for the Analysisof Electrically Small Antennas

Miloslav Capek

Department of Electromagnetic FieldFaculty of Electrical Engineering

Czech Technical University in PragueCzech Republic

miloslavcapekfelcvutcz

Habilitation ndash Public LecturePrague

April 12 2017

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 1 23

1 Why Are We Here Today2 Source Concept3 Mastering Sources4 From Current to Antenna Optimization5 Antenna Toolbox for Matlab (AToM)6 Afterword

This talk concerns

I time-harmonic quantities ie A (r t) = Re A (r) exp (jωt)I surface regions

I and electric currents J in vacuum

Document available at capekelmagorgTo see the graphics in motion open this document in Adobe Reader

Electrically small antenna inside acircumscribing sphere of a radius a

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 2 23

Why Are We Here Today

Ubiquity of Wireless Communication Devices

2016 2017 2018 2019 2020 20210

10

20

30

40

50

711

17

24

35

49

year

mob

ile

dat

atr

affic

per

mon

th[E

B]

yotta Y 1024

zetta Z 1021

exa E 1018

peta P 1015

tera T 1012

giga G 109

CISCO predictions for mobile data traffic1

CISCO reported1

I 429million mobile devices wereadded in 2016

I by 2021 there will be nearly 12billion mobile-connected devices

I by 2021 the average mobileconnection will surpass 204Mbps

We are rapidly approaching

I the era of IoT and 5G

I a world that demands on smallersize higher rates more devicesoperations in complex environments

1CISCO (Feb 9 2017) Cisco visual networking index ID 1454457600805266 (visited on 02122017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 3 23

Why Are We Here Today

Ubiquity of Wireless Communication Devices

2016 2017 2018 2019 2020 20210

10

20

30

40

50

711

17

24

35

49

year

mob

ile

dat

atr

affic

per

mon

th[E

B]

yotta Y 1024

zetta Z 1021

exa E 1018

peta P 1015

tera T 1012

giga G 109

CISCO predictions for mobile data traffic1

CISCO reported1

I 429million mobile devices wereadded in 2016

I by 2021 there will be nearly 12billion mobile-connected devices

I by 2021 the average mobileconnection will surpass 204Mbps

We are rapidly approaching

I the era of IoT and 5G

I a world that demands on smallersize higher rates more devicesoperations in complex environments

1CISCO (Feb 9 2017) Cisco visual networking index ID 1454457600805266 (visited on 02122017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 3 23

Why Are We Here Today

Ubiquity of Wireless Communication Devices

2016 2017 2018 2019 2020 20210

10

20

30

40

50

711

17

24

35

49

year

mob

ile

dat

atr

affic

per

mon

th[E

B]

yotta Y 1024

zetta Z 1021

exa E 1018

peta P 1015

tera T 1012

giga G 109

CISCO predictions for mobile data traffic1

CISCO reported1

I 429million mobile devices wereadded in 2016

I by 2021 there will be nearly 12billion mobile-connected devices

I by 2021 the average mobileconnection will surpass 204Mbps

We are rapidly approaching

I the era of IoT and 5G

I a world that demands on smallersize higher rates more devicesoperations in complex environments

1CISCO (Feb 9 2017) Cisco visual networking index ID 1454457600805266 (visited on 02122017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 3 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω

I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton0)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton1)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton2)ocgs[i]state=false

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 2: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

1 Why Are We Here Today2 Source Concept3 Mastering Sources4 From Current to Antenna Optimization5 Antenna Toolbox for Matlab (AToM)6 Afterword

This talk concerns

I time-harmonic quantities ie A (r t) = Re A (r) exp (jωt)I surface regions

I and electric currents J in vacuum

Document available at capekelmagorgTo see the graphics in motion open this document in Adobe Reader

Electrically small antenna inside acircumscribing sphere of a radius a

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 2 23

Why Are We Here Today

Ubiquity of Wireless Communication Devices

2016 2017 2018 2019 2020 20210

10

20

30

40

50

711

17

24

35

49

year

mob

ile

dat

atr

affic

per

mon

th[E

B]

yotta Y 1024

zetta Z 1021

exa E 1018

peta P 1015

tera T 1012

giga G 109

CISCO predictions for mobile data traffic1

CISCO reported1

I 429million mobile devices wereadded in 2016

I by 2021 there will be nearly 12billion mobile-connected devices

I by 2021 the average mobileconnection will surpass 204Mbps

We are rapidly approaching

I the era of IoT and 5G

I a world that demands on smallersize higher rates more devicesoperations in complex environments

1CISCO (Feb 9 2017) Cisco visual networking index ID 1454457600805266 (visited on 02122017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 3 23

Why Are We Here Today

Ubiquity of Wireless Communication Devices

2016 2017 2018 2019 2020 20210

10

20

30

40

50

711

17

24

35

49

year

mob

ile

dat

atr

affic

per

mon

th[E

B]

yotta Y 1024

zetta Z 1021

exa E 1018

peta P 1015

tera T 1012

giga G 109

CISCO predictions for mobile data traffic1

CISCO reported1

I 429million mobile devices wereadded in 2016

I by 2021 there will be nearly 12billion mobile-connected devices

I by 2021 the average mobileconnection will surpass 204Mbps

We are rapidly approaching

I the era of IoT and 5G

I a world that demands on smallersize higher rates more devicesoperations in complex environments

1CISCO (Feb 9 2017) Cisco visual networking index ID 1454457600805266 (visited on 02122017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 3 23

Why Are We Here Today

Ubiquity of Wireless Communication Devices

2016 2017 2018 2019 2020 20210

10

20

30

40

50

711

17

24

35

49

year

mob

ile

dat

atr

affic

per

mon

th[E

B]

yotta Y 1024

zetta Z 1021

exa E 1018

peta P 1015

tera T 1012

giga G 109

CISCO predictions for mobile data traffic1

CISCO reported1

I 429million mobile devices wereadded in 2016

I by 2021 there will be nearly 12billion mobile-connected devices

I by 2021 the average mobileconnection will surpass 204Mbps

We are rapidly approaching

I the era of IoT and 5G

I a world that demands on smallersize higher rates more devicesoperations in complex environments

1CISCO (Feb 9 2017) Cisco visual networking index ID 1454457600805266 (visited on 02122017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 3 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω

I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton0)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton1)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton2)ocgs[i]state=false

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 3: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Why Are We Here Today

Ubiquity of Wireless Communication Devices

2016 2017 2018 2019 2020 20210

10

20

30

40

50

711

17

24

35

49

year

mob

ile

dat

atr

affic

per

mon

th[E

B]

yotta Y 1024

zetta Z 1021

exa E 1018

peta P 1015

tera T 1012

giga G 109

CISCO predictions for mobile data traffic1

CISCO reported1

I 429million mobile devices wereadded in 2016

I by 2021 there will be nearly 12billion mobile-connected devices

I by 2021 the average mobileconnection will surpass 204Mbps

We are rapidly approaching

I the era of IoT and 5G

I a world that demands on smallersize higher rates more devicesoperations in complex environments

1CISCO (Feb 9 2017) Cisco visual networking index ID 1454457600805266 (visited on 02122017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 3 23

Why Are We Here Today

Ubiquity of Wireless Communication Devices

2016 2017 2018 2019 2020 20210

10

20

30

40

50

711

17

24

35

49

year

mob

ile

dat

atr

affic

per

mon

th[E

B]

yotta Y 1024

zetta Z 1021

exa E 1018

peta P 1015

tera T 1012

giga G 109

CISCO predictions for mobile data traffic1

CISCO reported1

I 429million mobile devices wereadded in 2016

I by 2021 there will be nearly 12billion mobile-connected devices

I by 2021 the average mobileconnection will surpass 204Mbps

We are rapidly approaching

I the era of IoT and 5G

I a world that demands on smallersize higher rates more devicesoperations in complex environments

1CISCO (Feb 9 2017) Cisco visual networking index ID 1454457600805266 (visited on 02122017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 3 23

Why Are We Here Today

Ubiquity of Wireless Communication Devices

2016 2017 2018 2019 2020 20210

10

20

30

40

50

711

17

24

35

49

year

mob

ile

dat

atr

affic

per

mon

th[E

B]

yotta Y 1024

zetta Z 1021

exa E 1018

peta P 1015

tera T 1012

giga G 109

CISCO predictions for mobile data traffic1

CISCO reported1

I 429million mobile devices wereadded in 2016

I by 2021 there will be nearly 12billion mobile-connected devices

I by 2021 the average mobileconnection will surpass 204Mbps

We are rapidly approaching

I the era of IoT and 5G

I a world that demands on smallersize higher rates more devicesoperations in complex environments

1CISCO (Feb 9 2017) Cisco visual networking index ID 1454457600805266 (visited on 02122017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 3 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω

I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton0)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton1)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton2)ocgs[i]state=false

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 4: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Why Are We Here Today

Ubiquity of Wireless Communication Devices

2016 2017 2018 2019 2020 20210

10

20

30

40

50

711

17

24

35

49

year

mob

ile

dat

atr

affic

per

mon

th[E

B]

yotta Y 1024

zetta Z 1021

exa E 1018

peta P 1015

tera T 1012

giga G 109

CISCO predictions for mobile data traffic1

CISCO reported1

I 429million mobile devices wereadded in 2016

I by 2021 there will be nearly 12billion mobile-connected devices

I by 2021 the average mobileconnection will surpass 204Mbps

We are rapidly approaching

I the era of IoT and 5G

I a world that demands on smallersize higher rates more devicesoperations in complex environments

1CISCO (Feb 9 2017) Cisco visual networking index ID 1454457600805266 (visited on 02122017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 3 23

Why Are We Here Today

Ubiquity of Wireless Communication Devices

2016 2017 2018 2019 2020 20210

10

20

30

40

50

711

17

24

35

49

year

mob

ile

dat

atr

affic

per

mon

th[E

B]

yotta Y 1024

zetta Z 1021

exa E 1018

peta P 1015

tera T 1012

giga G 109

CISCO predictions for mobile data traffic1

CISCO reported1

I 429million mobile devices wereadded in 2016

I by 2021 there will be nearly 12billion mobile-connected devices

I by 2021 the average mobileconnection will surpass 204Mbps

We are rapidly approaching

I the era of IoT and 5G

I a world that demands on smallersize higher rates more devicesoperations in complex environments

1CISCO (Feb 9 2017) Cisco visual networking index ID 1454457600805266 (visited on 02122017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 3 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω

I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton0)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton1)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton2)ocgs[i]state=false

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 5: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Why Are We Here Today

Ubiquity of Wireless Communication Devices

2016 2017 2018 2019 2020 20210

10

20

30

40

50

711

17

24

35

49

year

mob

ile

dat

atr

affic

per

mon

th[E

B]

yotta Y 1024

zetta Z 1021

exa E 1018

peta P 1015

tera T 1012

giga G 109

CISCO predictions for mobile data traffic1

CISCO reported1

I 429million mobile devices wereadded in 2016

I by 2021 there will be nearly 12billion mobile-connected devices

I by 2021 the average mobileconnection will surpass 204Mbps

We are rapidly approaching

I the era of IoT and 5G

I a world that demands on smallersize higher rates more devicesoperations in complex environments

1CISCO (Feb 9 2017) Cisco visual networking index ID 1454457600805266 (visited on 02122017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 3 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω

I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton0)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton1)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton2)ocgs[i]state=false

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 6: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω

I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton0)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton1)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton2)ocgs[i]state=false

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 7: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω

I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton0)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton1)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton2)ocgs[i]state=false

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 8: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω

I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton0)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton1)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton2)ocgs[i]state=false

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 9: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω

I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton0)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton1)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton2)ocgs[i]state=false

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 10: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Why Are We Here Today

Antennas ndash The Inevitable Components

The antenna predominates the overall performance of the communication system

I existence of fundamental bounds2

DirectivityD

Efficiencyη

BandwidthFBW

Electricalsize

ka lt 12

Trade-off between crucial parameters of anelectrically small antenna and its electrical size

I IoT RFID 5G etc electrically smallantennas (ESAs)

I ESAs encumbered with mutuallyconflicting parameters

Antenna design

I tries to find the optimal combination ofshape materials and feeding from a setof an infinite number of candidates

2J L Volakis C Chen and K Fujimoto Small Antennas Miniaturization Techniques amp Applications McGraw-Hill 2010

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 4 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω

I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton0)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton1)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton2)ocgs[i]state=false

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 11: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω

I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton0)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton1)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton2)ocgs[i]state=false

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 12: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω

I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton0)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton1)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton2)ocgs[i]state=false

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 13: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton0)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton1)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton2)ocgs[i]state=false

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 14: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton0)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton1)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton2)ocgs[i]state=false

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 15: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton1)ocgs[i]state=false

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton2)ocgs[i]state=false

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 16: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Why Are We Here Today

Antenna Analysis times Synthesis

Ω1

Folded loop(handsets)

Ω2

E-shaped patch(GPS WLAN)

Ω3

ldquoMag monopolesrdquo(PGB HIS)

Ω4

Meandered dipole(RFID)

Ω5

MonopolesPIFAs(LTE)

sim

Ω I

X X

X

Antenna analysis Antenna synthesis

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 5 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton2)ocgs[i]state=false

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 17: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 18: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 19: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 20: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Why Are We Here Today

Previous Approaches ndash Historical Overview

Former approaches to antenna design made use of

I circuit quantities3 (Vin Zin Γ ) rarr equivalent circuits

I field quantities4 (E H)

However

I any antenna parameter p can be inferred directly from acurrent as well ie

p = 〈J L (J)〉 J = J (r) r isin Ω

〈f L (g)〉 =intΩ

flowast (r) middot L (g (r)) dV

ε0amicro0a Z0

Equivalent circuit for TM1q

spherical mode

E

R3

TM10 mode y-z cut

3L J Chu ldquoPhysical limitations of omni-directional antennasrdquo J Appl Phys 19 no pp 1163ndash1175 1948 doi10106311715038

4R E Collin and S Rothschild ldquoEvaluation of antenna Qrdquo IEEE Trans Antennas Propag 12 no pp 23ndash27 1964doi 101109TAP19641138151

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 6 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 21: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 22: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 23: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Source Concept

Sorcery with Sources

p = 〈J L (J)〉 rarr IHLI (1)

Observations

I only the properties of the operators L are important

I all physics is imprinted in their structure

I as compared to fields the support Ω of current J is spatially limited

I can be represented in many ways eg L = [〈ψmL (ψn)〉]

Example Consider region Ω being made of PEC then the Electric Field Integral Equation (EFIE) reads5

Z (J) = ntimes ntimesEs (J) (2)

Let us represent it in a basis ψn (r) n isin 1 N ie J asympsumn Inψn as Z = [Zmn] in which

Zmn = 〈ψmZ (ψn)〉 =jZ0

intΩ

intΩprime

(kψlowastm middot ψn minus

1

knabla middot ψlowastmnablaprime middot ψn

)eminusjk|rminusrprime|

|r minus rprime|dΩprime dΩ (3)

5R F Harrington Field Computation by Moment Methods Wiley ndash IEEE Press 1993

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 7 23

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 24: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Source Concept

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 8 23

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 25: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Source Concept

Operators to Rule Them All

Source Concept(J M)

Integral andvariationalmethods

Modal de-compositions

Perspectivetopol-

ogy andgeometry

HPCalgorithmefficiency

Heuristicor convex

optimization

Sketch of main fields of the source concept

Complex power Pr + 2jωWA rarr1

2IHZI

Stored energy Wsto rarr1

4IHXprimeI

Electric energy We rarr1

4ωIHXeI

Magnetic energy Wm rarr1

4ωIHXmI

Far-field Fι (r)rarr Fι (r)I

Near-field E (r)rarr Ne (r)IH (r)rarr Nm (r)I

Directivity Dι (r)rarr IHUι (r)I

IHRI

Ohmic losses PL rarr1

2IHΣI

El and mag moment prarr PI mrarrMI

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 9 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 26: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 27: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Mastering Sources

Modal Decomposition Characteristic Modes

First two modes on PEC plate

Diagonalization of crucial operator Z = R + jX as6

XIm = λmRIm (4)

I a useful set of entire-domain basis functions

I =summ

αmIm (5)

I only a few modes needed to represent ESAs

(1 + jλm) δmn =1

2IHmZIn (6)

I many theoretical and practical problems solved7

6R F Harrington and J R Mautz ldquoTheory of characteristic modes for conducting bodiesrdquo IEEE Trans AntennasPropag 19 no pp 622ndash628 1971 doi 101109TAP19711139999

7M Capek P Hazdra P Hamouz et al ldquoA method for tracking characteristic numbers and vectorsrdquo ProgElectromagn Res B 33 no pp 115ndash134 2011 doi 102528PIERB11060209

M Capek P Hazdra M Masek et al ldquoAnalytical representation of characteristic modes decompositionrdquo IEEE TransAntennas Propag 65 no pp 713ndash720 2 2017 doi 101109TAP20162632725

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 10 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 28: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

IFS fractal discretized to 366 triangles (491 basis functions)

P+n

Pminusn

ρ+nρminusn

A+n

Aminusn

lnT+n

Tminusn

O

r

y

z

xψn (r) =

ln

2Aplusmnnρplusmnn

Rao-Wilton-Glisson basis functions ψn (r)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 29: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Mastering Sources

Characteristic Modes Provide Physical InsightExample Characteristic Modes of a PEC Plate (ka = 05)

I CMs are given only by PEC shape Ω and frequency ω (no excitation)

I eigenvalues λn indicate resonance or capacitiveinductive behavior

I CMs are orthonormal with respect to their far-fields

Current density of first characteristic mode λ1 = minus877Radiation pattern for first characteristic mode ()

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 11 23

var ocgs=hostgetOCGs(hostpageNum)for(var i=0iltocgslengthi++)if(ocgs[i]name==MediaPlayButton3)ocgs[i]state=false

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 30: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 31: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 32: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Mastering Sources

Optimal Currents ndash What Are TheyCase Study Minimization of Quality Factor Q

Antenna predefines capacity of a channel bandwidth of a system

I An extremely important parameter of any electrically small radiator

Antenna quality factor Q

I inversely proportional to bandwidth8

Q (I) =2ωmax WmWe

Prrarr max

IHXmI IHXeI

IHRI

(7)

Current Jopt minimizing quality factor Q of a given shape Ω

Q (Jopt) = minJQ (J) (8)

I optimal current Jopt constitutes the fundamental bounds on quality factor Q

8M Capek L Jelinek and P Hazdra ldquoOn the functional relation between quality factor and fractional bandwidthrdquoIEEE Trans Antennas Propag 63 no pp 2787ndash2790 2015 doi 101109TAP20152414472

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 12 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 33: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor QTake a pen and try to draw a current possessing minimum quality factor Q

PEC plate Ltimes L2 ka = 05 (Lλ asymp 0142)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 34: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Mastering Sources

Quiz Optimal Current for Minimal Quality Factor Q here is the correct answer

Optimal current with respect to minimum quality factor Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 13 23

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 35: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Mastering Sources

Modal Composition of the Optimal Current Jopt

Dominant (dipole-like) characteristic mode J1

+

First inductive (loop-like) mode J2 α2 = 04553

I composition of characteristic modes offers additional insight9

I superposition of two characteristic modes10 as Q (Jopt) asymp Q (J1 + α2J2)

9M Capek and L Jelinek ldquoOptimal composition of modal currents for minimal quality factor Qrdquo IEEE TransAntennas Propag 64 no pp 5230ndash5242 2016 doi 101109TAP20162617779

10M Capek P Hazdra and J Eichler ldquoA method for the evaluation of radiation Q based on modal approachrdquo IEEETrans Antennas Propag 60 no pp 4556ndash4567 2012 doi 101109TAP20122207329

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 14 23

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 36: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Mastering Sources

Formulation as Dual Problem ndash Convex Optimization

0 02 04 06 08 10

100

200

ν

qualityfactorsQ

max Qmν Qeν

Q (Iopt)

ka = 03

L L2

Minimization of quality factor Q primal (orange) and dual(blue) problems Dots represent steps of bisection algorithm

A convex combination

Qν =IH ((1minus ν) Xm + νXe) I

IHRI

solves the dual problem

minIQ (I) rarr max

νminIQν

with

((1minus ν) Xm + νXe) I = QνRI

I zero dual gap

9M Capek M Gustafsson and K Schab ldquoMinimization of antenna quality factorrdquo no 2017 [Online] Availablehttpsarxivorgabs161207676

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 15 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 37: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a

10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 38: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Mastering Sources

Natural Basis for Minimum Quality Factor Q

I From an individual radiator to the design of broadband MIMO systems

Spherical shell of radius a 10minus1 100

101

103

105

107

dipoles(6times)

quadrupoles(10times)

octupoles(14times)

hexadecapoles(18times)

ka

Qn

multipoles (asymptotes)

Qn modes

Modes with orthogonal far-fields and minimum quality factors Q

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 16 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 39: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modified

bull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 40: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges)

524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions

268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 41: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

From Current to Antenna Optimization

Feeding Synthesis

I Optimal currents are not compatible with realistic feedingbull Check that Vopt = ZIopt is a dense vector full of non-zero entries

I Structure needs to be modifiedbull currents sub-optimal to Ioptbull heuristic optimization needed

I Shape modification resembles a NP-hard problem

How many DOF do we have

W

N (unknowns) 28 52 120 infin

possibilities (edges) 524 middot 1029 139 middot 1068 115 middot 10199 infin

unique solutions 268 middot 108 450 middot 1015 133 middot 1036 infin

Complexity of structural optimization for a given voltage gap (red line) and N unknowns

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 17 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 42: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 43: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Computational time 12116 s

Result of heuristic structural optimization using MOGANSGAII from AToM-FOPS

Computational time 1155 s

Result of a deterministic in-house algorithm removingldquothe worstrdquo edge in each iteration

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 44: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

From Current to Antenna Optimization

Pixeling with Heuristic OptimizationExample Shape optimization (pixeling) with ad hoc specified feeding placement PEC plate Ltimes L2 ka = 03

Antenna synthesis ndash how far can we go

I At present only the heuristic optimization11

Q (I) Q (Iopt) = 1811

Resulting sub-optimal current approaching the minimalvalue of quality factor Q

Q (I) Q (Iopt) = 1813

Resulting current given by the in-house deterministicalgorithm

11Y Rahmat-Samii and E Michielssen Eds Electromagnetic Optimization by Genetic Algorithm Wiley 1999

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 18 23

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 45: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

From Current to Antenna Optimization

Example Pareto Fronts for Excitation of an Array

Multi-criteria optimization of 4 dipoles of length L = λ2 placed side-by-side and fed with the voltage gaps intothe middle Separation distance is d = L4 (blue) d = L8 (green) and d = L16 (red) respectively

Optimization genes for polarity sk amplitude Ak and phase ϕk of the driven voltage are composed of 188 bits

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 19 23

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 46: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Antenna Toolbox for Matlab (AToM)

Antenna Toolbox for Matlab (AToM)ldquoAntenna source conceptrdquo ndash New approach to antenna design

Third-party product developed by CTU-FEE BUT andMECAS ESI sro supported by TACR under program ALFA

I Part of the project FOPS (Fast Optimiz ProcedureS)

I approx 15 people involved 17 MCZK budget 35 years

I look at antennatoolboxcom

Colleaguersquos side-productnot going into ldquoRIVrdquo

I Based on previous implementation12

I heavily utilizes source concept features

I capable of handling data from third party software

I fast-prototyping in Matlab

I for details (and a recipe for gingerbread) see Appendix

12M Capek P Hamouz P Hazdra et al ldquoImplementation of the theory of characteristic modes in Matlabrdquo IEEEAntennas Propag Mag 55 no pp 176ndash189 2013 doi 101109MAP20136529342

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 20 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 47: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 48: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Afterword

What Next

Summary

I Source concept nowadays recognized as a leading paradigm

bull Techniques and principles introduced by habilitant and his colleagues

I Determination of the optimal currents is well-established

bull New operators derived modal interpretation of the optimality

I AToM package under development

bull Initial code written by habilitant now being transferred into CEM One

Future Work

I All concepts are still half-way to their applicability

bull Excitation placement number of feedersbull Shape modifications

I A good time to come up with great ideas

Can we one day solve the synthesis completely

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 21 23

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 49: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Afterword

Acknowledgment

I to all colleagues actively participating on the research

Left to right Mats Gustafsson (Lund Uni) DorukTayli (Lund Uni) Kurt Schab (NC State Uni)

Lukas Jelınek (CTU) MC (CTU) CasimirEhrenborg (Lund Uni) Guy Vandenbosch (KU

Leuven)

Left to right Filip Kozak (Valeo) Ondrej Kratky

(Siemens) Petr Kadlec (BUT) Vladimır Sedenka

(BUT) MC (CTU) Vıt Losenicky (CTU) PavelHazdra (CTU) Viktor Adler (CTU) Jaroslav

Rymus (MECAS ESI sro) Michal Masek (CTU)

Not in the photos Jan Eichler (CST-MWS) Pavel Hamouz (CMI)

I to all colleagues and friends from Department of Electromagnetic Field

I to family to Eva

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 22 23

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 50: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Questions

Questions

For a complete PDF presentation see capekelmagorg

Miloslav Capekmiloslavcapekfelcvutcz

28 03 2017 v100

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 51: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Appendices (QampA)

Selected Characteristics of Habilitant (valid 28032017 341 PM)Please visit capekelmagorg to get a broader picture

Pedagogy (in requested order and numbering)

1 2 doctoral students (neither defended yet)

bull M Masek (in 2nd year co-supervisor) V Losenicky (in 1st year supervisor)2 5 M Sc students (each thesis received at least one award)3 MTB course (Matlab) visit cwfelcvutczwikicoursesa0b17mtb4 SS1415 1050 WS1516 1028 SS1516 1060 WS1617 ndash

Scientific achievements (in requested order and numbering)

1 3 selected papers from IEEE Trans Antennas Prop

bull 101109TAP20122207329 101109TAP20132287519101109TAP20152414472

2 H-index WOS 6 (4) Scopus 7 (5) Google Scholar 93 Citations WOS 96 (46) Scopus 133 (63) Google Scholar 3094 Lund University (6+05 months) KU Leuven (2 months 4 visits)5 TACR TA04010457 (PI) GACR 15-10280Y (I)6 AToM functionality implemented in commercial EM simulator (CEM One)7 Werner von Siemens Excellence award 2nd place (2014)8 Delegate of The European Association on Antennas and Propagation (2015-2017)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 52: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Appendices (QampA)

Part 1prof Ctyroky

Ad 1) Numbering

I Standard LATEX template is used

I Citations done with respect to the IEEE style (IEEEtransty file) iereferences are in deed numbered from [1] Notice however that somecitations are already used in the Preface (before the main body)

Ad 1) Abbreviations and acronyms (see also last but one slide)

I All abbreviations except IP GPS RFID and GUI are introduced at place oftheir first occurrence The above mentioned ones are considered to becommon

Ad 2) Axes in Figure 11a-c

I Figure 11 have been considered as ldquoillustrativerdquo cartoon not scientific graph

Ad 3) Incomplete citation [55]

I Missing number corrected J L Volakis and K Sertel Integral EquationMethods for Electromagnetics Scitech Publishing Inc 2012

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 53: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Appendices (QampA)

Part 2prof Ctyroky

Ad 4) Usage of λ vs k vs ω vs f in the firstparagraph on p 4

I Nice question to discuss

I λ common in optics f (or ω) common in(antenna) engineering our preference is ka(meant here as a kind of dimensionless frequencysince in vacuum it reads ka = 2πafc0)

I eg in Lund λL highly preferred

Ad 5)

I All CM-related publication in IEEE AP flagshipjournals depicted on right

0 10 20 30 401970

1980

1990

2000

2010

only

IEEE TAPand

IEEE AWPL

212

000101101000021133

69

435

310

7344

117

1610

149

1822

3240

3840

3837

Publications

Year

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 54: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Appendices (QampA)

Part 3prof Ctyroky

500 1000 1500 2000 2500 300010minus2

10minus1

100

101

102

N

t[s

]

QZfit N308

IRAM modes M5 50050 1000100 2000200

GEP solved with QZ and IRAM for a Z matrix of size N timesN warm-up applied 3times repeated and averaged

Ad 6) Publication gap in the 1990s

I The very same questiondiscussed during CM Wokshopin Lund 2016 Consensustheory matured but notapplicable for real-life problems(since the lack of thecomputational resources at thattime)

Ad 7) GEP

I ARPACK13 vs LAPACK

I benchmark in Matlab

13R B Lehoucq D C Sorensen and C Yang ARPACK Usersrsquo Guide SIAM 1998

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 55: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 56: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 57: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Appendices (QampA)

Q1 Dense MoM Impedance Matrixdoc Karban

Finite Elements rarr sparse matrixMethod of Moments rarr dense matrix

Time-consuming parts

I matrix assemblage prop O(mN2

)

I matrix inversion prop O(N3)

Test on spherical helix antenna same

I discretization (1608 unknowns)

I number of frequency samples (500)

I feeding model (4 voltage gaps) used

Reality check

Simulator14 comp time [s]

FEKO15 13041

Makarov16 861

CEM One 895

AToM17 (1) 805

AToM17 (2) 1581

HW Intel Core i7-4970 CPU4GHz 16 GB RAM

14CST-MWS not tested Mesh cannot be imported as a NASTRAN file15Parallel FEKO has been enabled16S N Makarov Antenna and EM Modeling with Matlab Wiley 200217Implicit Matlab parallelization heavily utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 58: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Appendices (QampA)

Q1 Results Spherical Helix Antennadoc Karban

0-2

-1

0

1

2

3

1 2 43 5ka [-]

Zin [kW

]

Z Icirc 1608acute1608 nQuad = 1

Xin

Rin aa

100

AToMFEKOCEM One

Results of AToM-MoM (comparison with FEKO and CEM One)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 59: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Appendices (QampA)

Q2 Optimizationdoc Karban

Leading principle

Do not use Matlab toolboxes for AToM and FOPS development

I Avoid problems with licensing avoid bad practice

Main authors Petr Kadlec (BUT) Martin Marek (BUT)

Unique features

I chains ndash different algorithms can be used simultaneously

bull takes into account the existence of the NFL theorem18

I VND ndash variable number of dimensions

bull use-case signal coverage of a given area is an optimization problem for emittersrsquoposition but also for their count

I full control over all parameters (schemas)

18D H Wolpert and W G Macready ldquoNo free lunch theorems for optimizationrdquo IEEE Trans Evol Comput 1 nopp 67ndash82 1997 doi 1011094235585893

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 60: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Appendices (QampA)

Q2 Optimizationdoc Karban

Multidimensional single-objective (gradient-based) algorithms

I Newton method

I Nelder-Mead

I another classical approaches

Multidimensional algorithms accessible both in single- and multi-objective version

I Differential Evolution (DE)

I Self-organizing Migrating Algorithm19 (SOMA)

I Particle Swarm Optimization (PSO)

I Non-Dominated Sorting Genetic Algorithm20 (NSGAII)

I Variable Number of Dimensions (VND)

19G C Onwubolu and B V Babu New Optimization Techniques in Engineering Springer 200420K Deb Multi-Objective Optimization using Evolutionary Algorithms Wiley 2001

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 61: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 62: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Appendices (QampA)

Q3 AToM Statsdoc Karban

Habilitant wrote first version of AToM and leads development of the package (PI)

022015 032016 0320170

50000

100000

150000

time

num

ber

oflines

lines of code

022015 032016 0320170

1000

2000

3000

time

num

ber

of

m-fi

les

and

m-f

unct

ions

m-filesm-functions

Some statistics of AToM project over time (Matlab+GIT+Jenkinsshell)

directories 155

packages 86

classes 234

m-files 1828

functions 2972

unitTests 1188

lines of code 133322

comments 13381

Stats on 28032017 541 AM

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 63: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Appendices (QampA)

Q3 Lear More About AToM doc Karban

Licensing of AToM and FOPS Toolboxes

I AToM remains property of CTU

I FOPS remains shared property ofCTU and BUT

I antennatoolboxcom

bull white papers news

I questions atinfo[at]antennatoolboxcom

I YouTube channel

I MathWorks pre-product partner

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 64: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Appendices (QampA)

Q3 Why Is Atom Written in Matlabdoc Karban

Pros

I High-definition language

bull excellent for fast-prototypingbull many built-in functions are

embeddedbull implicit amp explicit

parallelization

I New functionality can easily bepublished21

I Maybe others

Cons

I Still not as fast as eg C

bull and to be efficient Matlabneeds very good programmingskill

I Not open-source

I To make standalone applicationis a nightmare

I Maybe others

I What is your opinion

21wwwmathworkscommatlabcentralfileexchange

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 65: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Appendices (QampA)

Q3 Features of Matlab (gtR2015b)doc Karban

I Run-time Type Analysis (gtMatlab 65)

bull data types in m-file are noticed during the first run

I Just-In-Time-Accelerator

bull parts of code that satisfy certain conditions are precompiledbull runs always (gtMatlab 2016a)

I Profiling via profile

bull JIT however deactivated during the profile measurementbull nearly impossible to do a good job without it

I Surprisingly rich Object-Oriented Programming

I Unit-test framework (gt2014b)

I Source Control Integration

bull GIT SVNbull Jenkins can be utilized

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 66: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Appendices (QampA)

Table of Acronymsprof Kasal

IP Internet Protocol CM Characteristic Modes

GPS Global Positioning System PIFA Planar Inverted F Antenna

RFID Radio Frequency Identification MLFMA Multilevel Fast Multipole Algorithm

GUI Graphical User Interface FBW Fractional Bandwidth

PEC Perfect Electric Conductor IBC Impedance Boundary Condition

EFIE Electric Field Integral Equation MIMO Multiple-Input Multiple-Output

MFIE Magnetic Field Integral Eq RWG Rao-Wilton-Glisson

CFIE Combined Field Integral Eq IFS Iterated Function System

MoM Method of Moments

Acronyms from the thesis in order of their appearance (only the bluish ones are used without explanation)

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 67: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Appendices (QampA)

Table of Symbolsprof Kasal

λ wavelength O computational complexity

k wavenumber vacuum k = 2πλ R3 Euclidean space

a radius of circumscribing sphere Prad radiated power

f frequency Plost ohmic losses

ω angular frequency vacuum ω = 2πf Pd dissipated power Pd = Prad + Ploss

Q quality factor Wsto stored energy

G antenna gain Zin input impedance Zin = Rin + jXin

J M electric magnetic currents S Poynting vector (energy flow)

E H intensity of electric magnetic field F far-field F = limrrarrinfinrejkrE

Z impedance matrix Z = R + jX ηr radiation efficiency

λn characteristic numbers

Symbols from the thesis sorted in order of their appearance

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus
Page 68: Source Concept and Its Utilization for the Analysis of ... · PDF fileSource Concept and Its Utilization for the Analysis of Electrically Small Antennas Miloslav Capek Department of

Appendices (QampA)

Bonus Recipe for Gingerbreadndash by Michal Masek AToM colleague

First available result of AToM project (andhopefully not the last one)

I Gingerbread

bull 500 g flour (Bohemian smooth)bull 180 g icing sugarbull 125 g Herabull 125 g honeybull 2 eggsbull 1 teaspoon of cooking sodabull 1 teaspoon of cinnamonbull 2 teaspoons of gingerbread spices

I Icing22

bull 200 g icing sugarbull 1 egg whitebull 1 teaspoon of lemon extract

I Rub icing until smooth

22Logo of the AToM project should respect the existing graphical manual

Capek M Habilitation lecture Source Concept and Its Utilization for Analysis of ESAs 23 23

  • Why Are We Here Today
  • Source Concept
  • Mastering Sources
  • From Current to Antenna Optimization
  • Antenna Toolbox for Matlab (AToM)
  • Afterword
      1. 00
      2. 01
      3. 02
      4. 03
      5. 04
      6. 05
      7. 06
      8. 07
      9. 08
      10. 09
      11. 010
      12. 011
      13. 012
      14. 013
      15. 014
      16. 015
      17. 016
      18. 017
      19. 018
      20. 019
      21. 020
      22. 021
      23. 022
      24. 023
      25. 024
      26. 025
      27. 026
      28. 027
      29. 028
      30. 029
      31. 030
      32. 031
      33. 032
      34. 033
      35. 034
      36. 035
      37. 036
      38. 037
      39. 038
      40. 039
      41. 040
      42. 041
      43. 042
      44. 043
      45. 044
      46. 045
      47. 046
      48. 047
      49. 048
      50. 049
      51. 050
      52. 051
      53. 052
      54. 053
      55. 054
      56. 055
      57. 056
      58. 057
      59. 058
      60. 059
      61. 060
      62. 061
      63. 062
      64. 063
      65. 064
      66. 065
      67. 066
      68. 067
      69. 068
      70. 069
      71. 070
      72. 071
      73. 072
      74. 073
      75. 074
      76. 075
      77. 076
      78. 077
      79. 078
      80. 079
      81. 080
      82. 081
      83. 082
      84. 083
      85. 084
      86. 085
      87. 086
      88. 087
      89. 088
      90. 089
      91. 090
      92. anm0
      93. fdrm0
      94. fdrm1
      95. fdrm2
      96. 10
      97. 11
      98. 12
      99. 13
      100. 14
      101. 15
      102. 16
      103. 17
      104. 18
      105. 19
      106. 110
      107. 111
      108. 112
      109. 113
      110. 114
      111. anm1
      112. 20
      113. 21
      114. 22
      115. 23
      116. 24
      117. 25
      118. 26
      119. 27
      120. 28
      121. 29
      122. 210
      123. 211
      124. 212
      125. 213
      126. 214
      127. anm2
      128. 30
      129. 31
      130. 32
      131. 33
      132. 34
      133. 35
      134. 36
      135. 37
      136. 38
      137. 39
      138. 310
      139. 311
      140. 312
      141. 313
      142. 314
      143. anm3
      144. 40
      145. 41
      146. 42
      147. 43
      148. 44
      149. 45
      150. 46
      151. 47
      152. 48
      153. 49
      154. 410
      155. 411
      156. 412
      157. 413
      158. 414
      159. anm4
      160. fdrm3
      161. 50
      162. 51
      163. 52
      164. 53
      165. 54
      166. 55
      167. 56
      168. 57
      169. 58
      170. 59
      171. 510
      172. 511
      173. 512
      174. 513
      175. 514
      176. 515
      177. 516
      178. 517
      179. 518
      180. 519
      181. 520
      182. 521
      183. 522
      184. 523
      185. 524
      186. 525
      187. 526
      188. 527
      189. 528
      190. 529
      191. 530
      192. 531
      193. 532
      194. 533
      195. 534
      196. 535
      197. 536
      198. 537
      199. 538
      200. 539
      201. 540
      202. 541
      203. 542
      204. 543
      205. 544
      206. 545
      207. 546
      208. 547
      209. 548
      210. 549
      211. 550
      212. 551
      213. 552
      214. 553
      215. 554
      216. 555
      217. 556
      218. 557
      219. 558
      220. 559
      221. 560
      222. 561
      223. 562
      224. 563
      225. 564
      226. 565
      227. 566
      228. 567
      229. 568
      230. 569
      231. 570
      232. 571
      233. anm5
      234. 60
      235. 61
      236. 62
      237. 63
      238. 64
      239. 65
      240. 66
      241. 67
      242. 68
      243. 69
      244. 610
      245. 611
      246. 612
      247. 613
      248. 614
      249. 615
      250. 616
      251. 617
      252. 618
      253. 619
      254. 620
      255. 621
      256. 622
      257. 623
      258. 624
      259. 625
      260. 626
      261. 627
      262. 628
      263. 629
      264. 630
      265. 631
      266. 632
      267. 633
      268. 634
      269. 635
      270. 636
      271. 637
      272. 638
      273. 639
      274. 640
      275. 641
      276. 642
      277. 643
      278. 644
      279. 645
      280. 646
      281. 647
      282. 648
      283. 649
      284. 650
      285. 651
      286. 652
      287. 653
      288. 654
      289. 655
      290. 656
      291. 657
      292. 658
      293. 659
      294. 660
      295. 661
      296. 662
      297. 663
      298. 664
      299. 665
      300. 666
      301. 667
      302. 668
      303. 669
      304. 670
      305. 671
      306. anm6
      307. 70
      308. 71
      309. 72
      310. 73
      311. 74
      312. 75
      313. 76
      314. 77
      315. 78
      316. 79
      317. 710
      318. 711
      319. 712
      320. 713
      321. 714
      322. 715
      323. 716
      324. 717
      325. 718
      326. 719
      327. 720
      328. 721
      329. 722
      330. 723
      331. 724
      332. 725
      333. 726
      334. 727
      335. 728
      336. 729
      337. 730
      338. 731
      339. 732
      340. 733
      341. 734
      342. 735
      343. 736
      344. 737
      345. 738
      346. 739
      347. 740
      348. 741
      349. 742
      350. 743
      351. 744
      352. 745
      353. 746
      354. 747
      355. 748
      356. 749
      357. 750
      358. 751
      359. 752
      360. 753
      361. 754
      362. 755
      363. 756
      364. 757
      365. 758
      366. 759
      367. 760
      368. 761
      369. 762
      370. 763
      371. 764
      372. 765
      373. 766
      374. 767
      375. 768
      376. 769
      377. 770
      378. 771
      379. 772
      380. 773
      381. 774
      382. 775
      383. 776
      384. 777
      385. 778
      386. 779
      387. 780
      388. 781
      389. 782
      390. 783
      391. 784
      392. 785
      393. 786
      394. 787
      395. 788
      396. 789
      397. 790
      398. anm7
      399. 7EndLeft
      400. 7StepLeft
      401. 7PauseLeft
      402. 7PlayLeft
      403. 7PlayPauseLeft
      404. 7PauseRight
      405. 7PlayRight
      406. 7PlayPauseRight
      407. 7StepRight
      408. 7EndRight
      409. 7Minus
      410. 7Reset
      411. 7Plus