wireless channel #8 (hata, cost, itu)ocw.snu.ac.kr/sites/default/files/note/1036.pdfwireless channel...
TRANSCRIPT
Wireless Channel Modeling 1
Hata Model
An empirical formula for propagation lossBased on Okumura’s measurement dataPropagation loss between isotropic antennaOnly for Quasi-smooth terrainStandard formula : urban area propagation loss
The propagation loss in an urban area
A , B – functions of frequency(MHz) and antenna height (m)R – distance (km)
System designs for land mobile radio servicesFrequency : 100 ~ 1500 (MHz) , Distance : 1 ~ 20 (km)Base station antenna height : 30 ~ 200 (m)Vehicular antenna height : 1 ~ 10 (m)
RBALp 10log+=
Wireless Channel Modeling 2
Propagation Loss Between Isotropic Antenna
Received Power Pr
Propagation Loss Lp
)(log10)/()( 102
effur AmdBmPdBmP +=
antennaisotropicanofstrengthfieldReceived:densitypowerReceived:
antennaistropicanofsectioncrossAbsorption:90)120(log10)/()/(
4/
102
2
EP
AmVdBEmdBmP
A
u
eff
u
eff
−−=
=
πμ
πλ
8.145)4/(log10)/()()(
210 +−−=
−=πλμ mVdBEdBWP
PPdBL
t
rtp
antennaistropicanofpowerradiatedEffective:tP
Wireless Channel Modeling 3
Okumura’s Prediction Curves and Propagation Loss
Transform the unit from ERP/dipole to EIRPAbsolute power gain of the dipole antenna : 2.2 dB
Pt (dBW EIRP) = 32.2 dB [when Pt’ = 1 kW (ERP/dipole)]
Propagation Loss Lp ( between the isotropic antenna )
)(2.2)/(')( dBdipoleERPdBWPEIRPdBWP tt +=
)/(log2045.139
)/()4/(log10178)(
10
210
mVdBEf
mVdBEdBL
c
p
μ
μπλ
−+=
−−=
antennaisotropicanofstrengthfieldReceived:)(frequencycarrier:where
EMHzfc
Wireless Channel Modeling 4
Empirical Formula for Propagation Loss
The field strength E
Propagation Loss Lp
RmVdBE 10log)/( βγμ +=
)(&)(bydeterminedconstans:,
MHzfmh cb
βγ
RBAdBLp 10log)( +=2
10178 10log ( / 4 )( )m
Aa h
B
λ πγβ
= −− −
= −
)(heightantennastationvehicularthe
forfactorcorrectionthe:)(
mh
ha
m
m
Wireless Channel Modeling 5
Introduction of the Empirical Formula
A : Value of the Propagation Loss at R = 1 (km)
121.0114.5106.094.5200
123.0116.5108.096.5150
125.0118.0110.098.5100
127.0120.5112.0101.070
129.5122.5114.0103.050
132.0124.5117.0105.530
1500900450150hb (m)
fc (MHz)
c
mb
fhahA
10
10
log16.2655.69)(log82.13
+=−−=
αα
Wireless Channel Modeling 6
Introduction of the Empirical Formula (Cont’d)
B : Slope of the Propagation Loss Curve
29.929.929.429.9200
30.931.130.430.4150
32.232.531.331.5100
33.432.232.533.270
34.133.834.133.450
35.735.735.035.030
1500900450150hb (m)
fc (MHz)
bhB 10log55.69.44 −=
Wireless Channel Modeling 7
Empirical Formula for Propagation Loss
Frequency (fc) : 150 ~ 1500 MHzBase station antenna height (hb): 30 ~ 200 mDistance (R) : 1 ~ 20 kma(hm) : correction factor for the vehicular station antenna height hm(m)
Correction factorsa(hm) : Correction factors in a meium-small or large cityKr : Corrections for SuburbanQr : Corrections for Open areas
Rhhahf
RBAdBL
bm
bc
p
1010
1010
10
log)log55.69.44()(log82.13log16.2655.69
log)(
−+−−+=
+=
Wireless Channel Modeling 8
Correction factors in a medium-small city (1/2)
Correction curves shown by straight lines with hm in linear scale
The reference hm of Lp : 1.5m ⇒ a1.5 = 0 dB at hm = 1.5 m
For a median-small city
)()(5.1 cmc fhfa ηξ −⋅=
7.0)(log1.1)( 10 −⋅= cc ffξ
8.0)(log56.1)( 10 −⋅= cc ffη
Wireless Channel Modeling 9
Correction factors in a medium-small city (2/2)
Error to the linear approximation ∝ fc
The maximum error ; fc = 1500 MHz , hm = 4 ~ 5 m
8.0)(log56.1)7.0)(log1.1()( 1010 −⋅−⋅−⋅= cmcm fhfha
MHzfmh cm 1500~150:,10~1:
Wireless Channel Modeling 10
Correction factors in a large city (1/2)
Large city : the building height average ≥ 15 mCurves can be considered as parabolas
,
)(96.3)54.1log(29.8' 2
103
dBha m
−⋅=
)(63.7)75.11log(2.3' 2
103
dBha m
−⋅=
MHzfc 200≤
MHzfc 400≥
Wireless Channel Modeling 11
Correction factors in a large city (2/2)
Maximum error ; About 1 dB
fc ≤ 200 MHz and hm ≥ 5 m
)(10.1)54.1log(29.8)( 2
10
dBhha mm
−⋅=
)(97.4)75.11log(2.3)( 2
10
dBhha mm
−⋅=
MHzfc 200≤
MHzfc 400≥
Wireless Channel Modeling 12
Estimation of the Approximation Error (1/2)
The error for each frequency very small(1-20 km)Maximum error : 1 dBIndependent of the distance
Only term A depends on freq.
Solid lines : Values from the formulaDashed lines : Values from the prediction curves
c
mb
fhahA
10
10
log16.2655.69)(log82.13
+=−−=
αα
Wireless Channel Modeling 13
Estimation of the Approximation Error (2/2)
The error for each hb
Maximum value : 1 dB
Due to the linear approximation of B
Solid lines : Values from the formulaDashed lines : Values from the prediction curves
bhB 10log55.69.44 −=
Wireless Channel Modeling 14
Corrections for Suburban
The suburban correctionfactor Kr (dB)
The propagation loss insuburban area Lps (dB)
{ }4.5
)28/(log2)( 210
+= cr fdBK
rpps KurbanInLdBL −= )()(
Wireless Channel Modeling 15
Corrections for Open areas
The open area correction factor Qr (dB)
The propagation loss inopen area Lpo (dB)
94.40log33.18
)log(78.4)(
10
210
+−
=
c
cr
ffdBQ
rppo QurbanInLdBL −= )()(
Wireless Channel Modeling 16
Semi-deterministic and empirical models for urban areas ( COST 231-Hata model )
COST 231 – Hata ModelExtension of Hata’s model to the freq. band 1500 ≤ fc (MHz) ≤ 2000
Lb: Basic Transmission Loss
where
restriction :
[ ] mb
mbcb
Ckmdmh
hamhMHzfdBL
+⋅⋅−+
−⋅−⋅+=
)(log)(log55.69.44
)()(log82.13)(log9.333.46)(
⎪⎪
⎩
⎪⎪
⎨
⎧
=
=
=
=
)(20~1
)(10~1
)(200~30
)(2000~1500
kmd
mh
mh
MHzf
m
b
c
⎪⎩
⎪⎨
⎧=
dB
dBCm
3
0 for medium sized city and suburbancenters with medium tree density
for metropolitan centers
Wireless Channel Modeling 17
COST 231-Walfisch-Ikegami model (COST231-WI model) (1/4)
A combination of the Walfisch and Ikegami modelsImproved path-loss estimation
Consider more data : hr (heights of buildings)w (widths of roads), b (building separation)ϕ (road orientation w.r.t. the direct radio path)
Distinguish LOS and non-LOS situations
hb r
b
lw
dBase station Mobile
α
hm
hr
Δhm
Δhb
Wireless Channel Modeling 18
COST231-WI model (2/4)
LOS case ( a simple propagation loss formula )for d ≥ 20 m
To determine first constant, using free space loss for d = 20 m
Non-LOS caseComposed of three terms
L0 ; free space lossLmsd ; multiple diffraction lossLrts ; rooftop-to-street diffraction and scatter loss
)(log20)(log266.42)( MHzfkmddBL cb ⋅+⋅+=
⎩⎨⎧
≤+>+++
=00
0
0
msdrts
msdrtsmsdrtsb LLforL
LLforLLLL
)(log20)(log204.32)(0 MHzfkmddBL c⋅+⋅+=
Wireless Channel Modeling 19
COST231-WI model (3/4)
Lrts is mainly based on Ikegami’s model
where
Lmsd is based on Walfisch and Bertoni model
Lbsh ; Path loss due to Δ hb > 0
Orim
crts
Lmh
MHzfmwdBL
+Δ⋅+
⋅+⋅−−=
)(log20
)(log10)(log102.8)(
[ ][ ]⎪
⎪⎩
⎪⎪⎨
⎧
≤≤−⋅−
<≤−⋅+
<≤⋅+−
=
905555)deg(114.00.4
553535)deg(075.05.2
350)deg(354.010
ϕϕ
ϕϕ
ϕϕ
for
for
for
LOri
)(log9)(log)(log)( mbMHzfkkmdkkLdBL cfdabshmsd ⋅−⋅+⋅++=
( )⎪⎩
⎪⎨⎧
≤
≥Δ+⋅−=
rb
rbbbsh
hh
hhmhL
,0
,)(1log18 10
Wireless Channel Modeling 20
COST231-WI model (4/4)
ka; the increase of the path loss for Δ hb < 0
kd, kf ; the multi-screen diffraction loss vs. distance and frequency
Default values
hr = 3 × ( # of floors) + roof-height , roof-height =
b = 20 ~ 50 m, w = b / 2, ϕ = 90°
Restrictionsfc : 800 ~ 2000 MHz, hb: 4 ~ 50 m, hm: 1~3 m, d: 0.02 ~ 5 km
⎪⎩
⎪⎨
⎧
≤Δ⋅−
>
=rb
r
b
rb
d hhmhmh
hhk
,)(
)(1518
,18 [ ]
[ ]⎪⎪⎩
⎪⎪⎨
⎧
−⋅
−⋅
+−=
1925/)(5.1
1925/)(7.0
4
MHzf
MHzf
k
c
c
f
, medium sized city and suburban centers
, metropolitan centers
⎪⎪⎩
⎪⎪⎨
⎧
≤<⋅Δ⋅−
≤≥Δ⋅−
>
=
rbb
rbb
rb
a
hhandkmdkmdh
hhandkmdh
hh
k
5.0,)(6.154
5.0,8.054
,54
⎩⎨⎧
flat;)(0
pitched;)(3
m
m
Wireless Channel Modeling 21
Multiple screen diffraction (1/5)
Relatively uniform height buildings modeled as absorbing half-screensA process of multiple diffraction past rows of buildingsAssumptions
Propagation perpendicular to the rows of buildingsMagnetic field polarized parallel to the ground (vertically polarized)Consider the problem of plane-wave diffraction past a semi-infinitesequence of rows labeled n = 0, 1, 2, ⋅ ⋅ ⋅
b x
α
hr
hb
hm
d
Wireless Channel Modeling 22
Multiple screen diffraction (2/5)Elevated antennas
For elevated antennas
Numerical resultsHn(y = 0) for elevated antennas settles to a nearly constant value for n large enough
Excess path loss due to multiple screen diffraction, Le
Le = -10 log (Q2)Q depends on BS antenna heightH & row spacing d through the dimensionless parameter n=0 1 2 3 N-1 N
b
x
y
Nbtanαα
∫∞ −
+ +=0
4/
1 ')coscos()'(2
)( dyr
eyHeyHjkr
n
j
n αδλ
π
rbyybr /cos,)'(where 22 =−+= δ
Wireless Channel Modeling 23
Multiple screen diffraction (3/5)Elevated antennas
The dimensionless parameter gp
Over the range of 0.01 < gp < 0.4
Ex) for 900 MHz, typical row spacing of b = 40 m, and hb – hr= 10 m⇒ Range of gp correspond to 0.3 km < d < 11 km
In order to apply the theory for smaller values of d
Over the range of 0.01 < gp < 1 ⇒ 0.11 km < d < 11 km (900 MHz)
λα bgp = d
hhd
hh rbrb −≈
−= −1tanwhereα
9.035.2)( pp ggQ =
32 962.0327.3502.3)( pppp ggggQ +−=
Wireless Channel Modeling 24
Multiple screen diffraction (4/5)Low antennas
Le for low antennasThe factor QN giving the reduction of the field at the top of last screen due to screens
QN depends on the source height y0 above or below the rooftops
and the row separation b
[ ]∑∞
=
+=0
,2!
11q
qNq
cN Ijgq
NQ π rbc hhyb
yg −== 00 ,1whereλ
1, 1
, , 2 1/ 21
( 1) 12( 1) ( )2 ( 1)
NN q
N q N qn
IN qI IN N nNπ
−−
−=
−= +
+ −+∑
)conditioninitial()1(
14
1,)1(
1
02/32/31,2/30, ∑
= −+=
+=
N
nNN nNn
IN
Iπ
Wireless Channel Modeling 25
Multiple screen diffraction (5/5)Examples
Field after multiple diffraction over absorbing screens
For antennas above the rooftops (y0 > 0)A rate of decrease for field is less than 1/N, but approaches 1/N
For antennas below the rooftops (y0 < 0)The field initially decreases more rapidly than 1/N, but quicklyapproaches the 1/N variation
fc = 900 MHz and b = 50 m fc = 1800 MHz and b = 50 mN N
NQ NQ
Wireless Channel Modeling 26
ITU-R P. 1411 model
Estimating path loss for short-range(less than 1km) outdoor radio systems
Propagation affected primarily by buildings and trees
Classified into 3 categories according to propagation situation ,rooftops-NLOS(1), street canyons-NLOS(2), LOS(3)
Rooftops-NLOS model is similar to COST231-WI model
BS1
BS2
MS1
MS2
MS3
MS4
(1)
(2)(3) (3)
Wireless Channel Modeling 27
Rooftops NLOS model (P. 1411)
The formula is same as the COST231-WI modelExtension of COST231-WI model to the freq. band fc (MHz) ≤ 5000for Δhb > hr (ITU-R P.1411-3, March, 2005)
ka = 71.4, kf = – 8 for Δhb > hr and fc > 2000 MHz
AssumptionThe roof-top heights differ only by an amount less than the 1st Fresnel-zone radius over a path of length l, hr = the average roof-top heightThe roof-top heights vary by much more than the 1st Fresnel-zone radius: a preferred method of knife-edge diffraction calculation due tothe the highest buildings along the path is recommended to replace themulti-screen modelWhere l : distance over which the buildings extend
Wireless Channel Modeling 28
Lmsd has two formulas according to Δhb and incidence angleA criterion for grazing incidence : settled field distance ds
When l > ds, Lmsd has same formula as COST231-WI modelWhere l : distance over which the buildings extend
When l < ds
rbbb
s hhhhdd −=Δ
Δ
⋅= where2
2λ
)(log10 210 Nmsd QL ⋅−=
⎪⎪⎪⎪⎪
⎩
⎪⎪⎪⎪⎪
⎨
⎧
<⎟⎟⎠
⎞⎜⎜⎝
⎛+
−
≈
>⎟⎟⎠
⎞⎜⎜⎝
⎛ Δ⋅
=
rb
rb
rbb
N
hhford
b
hhfordb
hhforbdh
Q
,2
112
,
,35.29.0
θπθρλ
π
λ
( )22
arctan
bh
bh
b
b
+Δ=
Δ=
ρ
θ
Rooftops NLOS model (P. 1411)
Wireless Channel Modeling 29
Street canyons NLOS model (P. 1411)
Situations where both antennas are below roof-top level
Lr : reflection path loss
Ld : diffraction path loss
BS
MSx1
x2
w1
w2
α
( )10/10/10 1010log10)( dr LL
SC dBL −− +⋅−=
⎟⎠⎞
⎜⎝⎛+++=
λπα 4log20)()(log20)( 10
21122110 ww
fxxxxdBLr
παα
α <<≈= )()34(6.086.3)( 5.3 radwheref
[ ]
⎟⎠⎞
⎜⎝⎛+⎟
⎠⎞
⎜⎝⎛ −−
++=
λπ
πα 4log20180901.0
2)(log10)(
10
211210 ad DxxxxdBL
⎥⎥⎦
⎤
⎢⎢⎣
⎡−⎟⎟
⎠
⎞⎜⎜⎝
⎛+⎟⎟
⎠
⎞⎜⎜⎝
⎛⎟⎟⎠
⎞⎜⎜⎝
⎛=
2arctanarctan
240
1
1
2
2 ππ w
xwxDa
Wireless Channel Modeling 30
LOS situations within street canyons (P. 1411)
Basic transmission loss can be characterized by two slopes and a single breakpointAn approximate lower bound
An approximate upper bound
⎪⎩
⎪⎨⎧
>
≤+=
bpbp
bpbp
bpRdforRd
RdforRdLL
lLOS )/(log40
)/(log20
10
10
,
⎪⎩
⎪⎨⎧
>
≤++=
bpbp
bpbp
bpRdforRd
RdforRdLL
uLOS )/(log40
)/(log2520
10
10
,
pointbreaktheatlossontransmissibasicthe:8
log20where2
10 ⎟⎟⎠
⎞⎜⎜⎝
⎛=
mbbp hh
Lπλ
distancebreakpointthe:4
whereλ
mbbp
hhR
⋅≈
Wireless Channel Modeling 31
ITU-R P. 1546 model (1/4)
Point-to-area predictions for terrestrial servicesFrequency range : 30 MHz to 3000 MHzThe distance range : 1 km to 1000 km
The propagation curves at nominal frequencies of 100, 600 and 2000 MHz as a function of various parameters used
Curves are based on measurement dataRepresent the field-strength values for 1 kW e.r.p. exceeded for 50%, 10% and 1 % of time
Interpolation and extrapolation for nominal values such as frequency, distance, percentage time, base antenna height and mixed land sea path are used
Wireless Channel Modeling 32
ITU-R P. 1546 model (2/4)
Interpolation / ExtrapolationFor frequency, distance and BS antenna height
Einf, Esup: Field strength value for lower/higher nominal value
hb, d, f: required value for prediction
hinf, dinf, finf / hsup, dsup, fsup: lower/higher nominal value
For percentage time
Qt = Qi(t/100), t: percentage time
Qi(x): inverse complementary cumulative normal distribution function
),,/,,log(),,/,,log(
)(infinfinfsupsupsup
infinfinfinfsupinf fdhfdh
fdhfdhEEEE b−+=
)/()()/()( supinfsupinfsupinfinfsup QQQQEQQQQEE tt −−+−−=
Wireless Channel Modeling 33
ITU-R P. 1546 model (3/4)
Mixed pathsStep1. The total length of path that lies over land dl
Step 2. The quantity Δ
Step 3. The mixed path value at the MS antenna distance, dtotal
Step 4. The difference between mixed-path and land path field strength
Step 5. An interpolation factor
Step 6. The field strength for the mixed path
[ ]⎩⎨⎧
−
<−=Δ
otherwisedEdEkmdforkmEkmEd
lsealland
lsealandl
,)()(1,)1()1(
Δ+= )()( totalseatotalmix dEdE
)()( totallandtotalmix dEdEE −=Δ
)]([exp)1( 0003527.042.2 BShld −⋅−−+= βααχ
χ⋅Δ+= EdEE totalland )(
0001.0,3.0 == βα
Wireless Channel Modeling 34
ITU-R P. 1546 model (4/4)
CorrectionsMS antenna height
Effective terrain clearance angle θeff from BS antenna
Terrain clearance angle θ from MS antenna
Location variabilityEquivalent basic transmission loss
Lb = 139 – E + 20log fc
Wireless Channel Modeling 35
Impulse response in a Multipath Environment
Received Voltage:an=amplitude of nth ray
τn=Ln/c=Travel time of the nth ray
For non-overlapping impulses
( ) ( )∑ −=n
nn tatV τδ
( ) ( ) ( ) ( ) ( )
powersrayofsum
222
=
−=−−== ∑∑∑n
nnn m
mnmn tattaatVtP τδτδτδ
Wireless Channel Modeling 36
Finite Width Pulse Response in Multipath
Received voltage:
For partially overlapping pulses
( ) ( )∑ −−=n
jjwtjkLnn
nn eeetpatV φτ
( ) ( ) ( ) ( ) ( )mn LLjk
n mmnmn etptpaatVtP −−∑∑ −−== ττ*2
Wireless Channel Modeling 37
Tapped Delay Line ModelThe discrete time frequency-selective fading channel model
Tapped delay line (TDL) with spacing ts and time varyingcoefficients g(t)
ts: Poisson arrival distribution
g(t): Rician or Rayleigh distribution
ts ts tss(t)
g0(t) g1(t) g2(t) gL-1(t) gL(t)
y(t)
Wireless Channel Modeling 38
Saleh & Valenzuela ModelChannel model
{θkl}: statistically independent uniform r.v. over [0, 2π){β 2
kl}: monotonically decreasing functions of {Tl} and {tl}The first rays of the clusters & subsequent rays
Poisson arrival distribution ( rate Λ, γ )
tT0 T1 Tl
)(2 tβ
Γ− /Teγ/te−
tT0 T1 Tl
00β
10β 01βl0β
)()(0 0
klljθ
l kkl τTtδeβth kl −−= ∑ ∑
∞
=
∞
=
0 )],(exp[)|()](exp[Λ)( )1()1(kl11 >−−=−−= −−−− l, k, pTTΛ|TTp lkkllkllll ττλλττ
Wireless Channel Modeling 39
ETSI Channel Model
Classify test environment into 3 casesSeparate Channel w.r.t RMS delay spread
55400040370Vehicular(High Antenna)
557504045Outdoor to Indoor andPedestrian
451005035Indoor Office
P(B) (%)RMS delay ofChannel B (nsec)P(A) (%)RMS delay of
Channel A (nsec)Test Environment
Wireless Channel Modeling 40
FLAT-25.2700-32.03106
FLAT-18.0500-26.02905
FLAT-10.8300-18.01704
FLAT-7.2200-10.01103
FLAT-3.6100-3.0502
FLAT00001
Avg. Power (dB)Rel. Delay (nsec)Avg. Power
(dB)Rel. Delay (nsec)Doppler
Spectrum
Channel BChannel ATap
Examples ( ETSI Model ) Indoor office case
Channel A Channel B
0 Relative Delay (nsec)500 10000 Relative Delay (nsec)500 1000
Wireless Channel Modeling 41
JTC Channel Model
Classify test environment into 9 casesSeparate Channel w.r.t RMS delay spread
Residential
Urban/Suburban Low-Rise
Urban High-Rise
Vehicular
Residential
Urban/Suburban Low-Rise
Urban High-Rise
Pedestrian
Outdoor
Commercial
Office
Residential
Indoor
Wireless Channel Modeling 42
Examples ( JTC Model )Indoor Office Case
Avg. PowerRel. Delay
2400
1100
700
500
200
0
Channel C
-16.3
-1.0
-4.8
-2.4
-1.4
0
FLAT-25.27006
FLAT-18.05005
FLAT-10.83004
FLAT-7.22003
FLAT-3.6100-8.51002
FLAT00001
Avg. PowerRel. DelayAvg. PowerRel. DelayDoppler
Spectrum
Channel BChannel ATap
0 Relative Delay (nsec)500 0 Relative Delay (nsec)500 1000 15000 Relative Delay (nsec)500
Channel A Channel B Channel C
Wireless Channel Modeling 43
IEEE-802.11 WLANs TGn channel model
Five delay profile models proposed for different indoor environmentsInclude spatial information for channel model
6
4
3
2
2
1 tap
# of clusters
503 / – ∞LOS/NLOSD Residential homes and small offices
150 / – ∞LOS/NLOSB
300 / – ∞LOS/NLOSC
1506 / – ∞LOS/NLOSF
1006 / – ∞LOS/NLOSE
Optional, 1 tap at 0 ns delay model
00 / – ∞LOS/NLOSA
EnvironmentsRMS delay
(ns)(NLOS)
K (dB)LOS/NLOS
ConditionModel
Wireless Channel Modeling 44
Examples (WLANs TGn channel model B)
25.425.425.425.425.425.425.4AS (°)AS(Tx)
106.5106.5106.5106.5106.5106.5106.5AoD (°)AoD
25.225.225.225.225.225.225.2AS (°)AS(Rx)
118.4118.4118.4118.4118.4118.4118.4AoA (°)AoA
- 21.8- 18.7- 15.6- 12.5- 9.4- 6.3- 3.2Power (dB)Cluster 2
14.414.414.414.414.4AS (°)AS(Tx)
225.1225.1225.1225.1225.1AoD (°)AoD
14.414.414.414.414.4AS (°)AS(Rx)
4343434343AoA (°)AoA
-21.7-16.2-10.8- 5.40Power (dB)Cluster 1
80706050403020100Excess delay (ns)
987654321Tap index
Wireless Channel Modeling 45
• Moving Focus : from Propagation Channel to Radio Channel- Spatial description of channel attributes (DoA, DoT, AS, DS, per-path PDP, etc)- Inclusion of the antenna pattern- Time-dependent trajectory of the MS
CODER MODEMIF/RF
STAGESCODERMODEM
IF/RFSTAGES
RADIO CHANNEL
PROPAGATION CHANNEL
DIGITAL CHANNEL
MODULATIONCHANNEL
TRANSMITTER RECEIVER
@Mobile Radio Comm by Raymond
Spatial Channel Model (SCM) (1/2)
Wireless Channel Modeling 46
Example of spatial channel information• Real Measurement Data : Angle vs Delay vs Power -
Spatial Channel Model (SCM) (2/2)
Wireless Channel Modeling 47
• Link Level Spatial Channel Parameters (Base Station and Terminal Specific)
- Mean Angle of Arrival- Rms Angle Spread- Power Azimuth Spectrum- Behavior per Resolvable Path- Ricean and Rayleigh Fading
• System Level Spatial Channel Parameters (System Wide Parameters)
- BS, MS Positions- AOA, AS, PAS for each terminal- Random MS Orientation- Mixture of Channel Models- Explicit Spatial Interference Modeling- Per-path spatial parameters
Link level vs System level SCM
Wireless Channel Modeling 48
Link Level SCM assumptions (1/3): parameters• Only one snapshot of the channel behavior• Not used to compare performance of different algorithms.• Only for calibration : comparison of performance results from different implementations
of a given algorithm.
Wireless Channel Modeling 49
Link Level SCM assumptions (2/3): parameters
Wireless Channel Modeling 50
3 Sector Antenna Pattern
-25
-20
-15
-10
-5
0
-120 -100 -80 -60 -40 -20 0 20 40 60 80 100 120
Azimuth in Degrees
Gai
n in
dB
.
( )2
3
min 12 , where 180 180mdB
A A⎡ ⎤⎛ ⎞θ
θ = − − ≤ θ ≤⎢ ⎥⎜ ⎟θ⎢ ⎥⎝ ⎠⎣ ⎦
Antenna Boresight in direction of arrow
3-Sector Scenario
BS
• Am is max attenuation 20 dB for 3 sector, 23dB for 6 sector• θ3dB is 70° for 3 sector, 35° for 6 sector
Link Level SCM assumptions (3/3)BS antnena parameters
Wireless Channel Modeling 51
• Am is max attenuation 20 dB for 3 sector, 23dB for 6 sector• θ3dB is 70° for 3 sector, 35° for 6 sector
Link Level SCM Reference Values for calibration
Wireless Channel Modeling 52
Assumptions• Multiple cells environments : BSs & MSs• Performance metrics (throughput, delay etc) are collected over D drops• Quasi-static channel for each drop
- the channel undergoes fast fading - the locations of the MSs are fixed for each drop.
Final Goal• For an S element BS array and a U element MS array,
obtain the channel coefficients for one of N multipath components
)(tnH S -by- U matrix of complex amplitudes
at nth path
( )[ ]( )
( ) tθjkkdjG
jkdG
MP
th vAoAnmM
mmnAoAmnuAoAmnMS
AoDmnsAoDmnBSnnus )cos(exp
)sin(exp)(
)sin(exp)()( ,,1
,,,,,
,,,,,, θ−⋅
⎟⎟⎟
⎠
⎞
⎜⎜⎜
⎝
⎛
Φ+θθ
×θθ= ∑ =
v
General description for System level SCM (1/3)
Wireless Channel Modeling 53
Procedure for generating the channel matrix
• Specify an environment : urban/suburban macro, or urban micro• Obtain the parameters to be used in simulations• Generate the channel coefficients
Suburban Macro Options: None
Urban Macro Options: Urban canyon Far scattering cluster Line of sight
Urban Micro Options: Urban canyon Far scattering cluster Line of sight
Pathloss model: (Hata) Pathloss model: (Walfish-Ikegami)
Generate coefficients
)(tnH
General description for System level SCM (2/3)
Wireless Channel Modeling 54
Ray-based Scattering Model• N time-delayed multipath (N = 6 or 3)
• M subpaths per multipath (M = 20)
BSθ
AoDn ,δ
, ,n m AoDΔ
AoDmn ,,θ
BSΩ
N
NCluster n
AoAmn ,,θ
, ,n m AoAΔ
,n AoAδ
MSΩ
MSθ
θv
BS array broadside
MS array broadside
BS array
MS directionof travel
MS array
Subpath m
v
General description for System level SCM (3/3)
Wireless Channel Modeling 55
• Macrocell ; BS to BS 3 km with antennas above rooftop • Microcell ; BS to BS 1 km with antennas at rooftop height
( )10 ^ , ~ (0,1)AS AS ASx xσ = ε + μ η
Scenario Parameters (1/2)
Wireless Channel Modeling 56
Path Loss Model• Macrocell : Modified Hata Model• Microcell : Walfisch-Ikegami Model
Scenario Parameters (2/2)
Wireless Channel Modeling 57
• Angle of Departure (AoD) for N multipath
Wish to generate RVs : δn,AoD corresponding to nth multipath• δn’ ~ η(0,σ2AoD) , σAoD = rAS * σAS
• Proportional factor rAS = angle occurrence (σAoD) / power weighted angular spread(σAS)
@3GPP TSGR1#25 (02) 0613
• Obtain δn,AoD as increasing order : δ1,AoD < … < δn,AoD
AoD for N multipath