c© 2011

Zhonghao Hu

All Rights Reserved

To my parents

Page iii

Page iv

Contents

Contents v

Abstract xiii

Statement of Originality xv

Acknowledgments xvii

Conventions xix

Publications xxi

Abbreviations xxiii

List of Figures xxvii

List of Tables xxxv

Chapter 1. Introduction and Motivation 1

1.1 Research Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.2 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.3 Original Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

1.4 Thesis Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

Chapter 2. RFID Background 13

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.2 The History of RFID . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.3 RFID Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

2.3.1 Mode of Excitation . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

Page v

Contents

2.3.2 Operating Frequency . . . . . . . . . . . . . . . . . . . . . . . . . . 17

2.4 Regulations and Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

2.4.1 Regulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

2.4.2 Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

2.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

Chapter 3. Operating Range Evaluation of RFID Systems 25

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

3.2 Fundamental Parameters of Antennas and the Friis Equation . . . . . . . 26

3.2.1 Power Transmission in a Tag . . . . . . . . . . . . . . . . . . . . . 26

3.2.2 Effective Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

3.2.3 Effective Length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

3.2.4 Gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

3.2.5 EIRP and ERP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

3.2.6 Polarisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

3.2.7 The Friis Transmission Equation . . . . . . . . . . . . . . . . . . . 34

3.3 Tag Antenna Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

3.4 Threshold Power of a Transponder . . . . . . . . . . . . . . . . . . . . . . 36

3.4.1 Modulator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

3.4.2 Rectifier Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

3.4.3 Memory Chosen . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

3.5 The Reader Sensitivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

3.6 The Literature Review on the Existing Work in Evaluating Operating

Range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

3.7 Interpretation and limitations of the Friis Transmission Equation in an

RFID Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

3.7.1 Forward Link . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

3.7.2 Backward Link . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

Page vi

Contents

3.7.3 Limitations in Implementing the Friis Transmission Equation . . 47

3.8 The Use of S-parameters in Analysing the Operating Range of RFID Sys-

tems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

3.8.1 Formula Derivation . . . . . . . . . . . . . . . . . . . . . . . . . . 49

3.8.2 Formula Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

3.9 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

Chapter 4. Analysis and Design of Meander Line Dipole Antennas 61

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

4.2 Introduction and Validation of the Formula for Calculating Resonant

Frequency of an MDA in Free Space . . . . . . . . . . . . . . . . . . . . . 63

4.2.1 Formula Derivation . . . . . . . . . . . . . . . . . . . . . . . . . . 63

4.2.2 Validation of Equation (4.11) . . . . . . . . . . . . . . . . . . . . . 66

4.3 Modifications on Equation (4.11) for RFID Tag Antenna Design . . . . . 68

4.3.1 Limitations of Equation (4.11) in RFID Tag Antenna Design . . . 68

4.3.2 Method for Calculating Relative Effective Permittivity of an MDA

on a Dielectric Substrate . . . . . . . . . . . . . . . . . . . . . . . . 69

4.3.3 Further Validation of the Method for Calculating the εre f f of an

MDA on a Dielectric Substrate . . . . . . . . . . . . . . . . . . . . 76

4.4 Experimental Validation of Equation (4.12) . . . . . . . . . . . . . . . . . 78

4.5 Radiation Pattern and Efficiency . . . . . . . . . . . . . . . . . . . . . . . 79

4.5.1 Physical Dimension of MDA . . . . . . . . . . . . . . . . . . . . . 80

4.5.2 Dielectric Substrate . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

4.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

Chapter 5. A Security Tag Design 85

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

5.2 T-Seal Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

5.3 Chip and Antenna Selection . . . . . . . . . . . . . . . . . . . . . . . . . . 89

Page vii

Contents

5.3.1 Chip Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

5.3.2 Antenna Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

5.4 The Security Tag Antenna Design . . . . . . . . . . . . . . . . . . . . . . . 91

5.4.1 Semi Finished Tag Design . . . . . . . . . . . . . . . . . . . . . . . 92

5.4.2 Completely Finished Tag Design . . . . . . . . . . . . . . . . . . . 95

5.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

Chapter 6. Solutions for the Antenna on Metal Problem 103

6.1 Introduction and Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104

6.2 The Antenna on Metal Problem . . . . . . . . . . . . . . . . . . . . . . . . 104

6.2.1 Metallic Boundary Conditions . . . . . . . . . . . . . . . . . . . . 104

6.2.2 Antenna Parameters in Proximity to Metal . . . . . . . . . . . . . 107

6.2.3 The Performance of Commercial Tags Above a Metal Plate . . . . 108

6.3 Previous Solutions to the Problem . . . . . . . . . . . . . . . . . . . . . . 112

6.3.1 One Quarter Wavelength Isolator Solution . . . . . . . . . . . . . 112

6.3.2 Antenna Selection Solutions . . . . . . . . . . . . . . . . . . . . . . 113

6.3.3 Artificial Magnetic Conductor Solutions . . . . . . . . . . . . . . . 114

6.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118

Chapter 7. The Slitted Decoupler Design for Metallic Item Detection 121

7.1 Introduction and Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122

7.2 Structure of the Slitted Decoupler . . . . . . . . . . . . . . . . . . . . . . . 124

7.3 Design Principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125

7.4 Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126

7.4.1 Construction of the Simulated Devices . . . . . . . . . . . . . . . . 126

7.4.2 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 126

7.5 Patch Antenna Resonant Property Analysis . . . . . . . . . . . . . . . . . 130

7.5.1 Theoretical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 131

Page viii

Contents

7.5.2 Simulation results . . . . . . . . . . . . . . . . . . . . . . . . . . . 136

7.6 Slitted Decoupler Parameter Settings . . . . . . . . . . . . . . . . . . . . . 144

7.6.1 Simulation Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144

7.6.2 Length and Width of Each Top Patch Selection . . . . . . . . . . . 146

7.6.3 Dielectric Material Layer Thickness Selection . . . . . . . . . . . . 147

7.6.4 Slit Width Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . 148

7.6.5 Dielectric Material Selection . . . . . . . . . . . . . . . . . . . . . . 149

7.6.6 The Ground Plane Size Selection . . . . . . . . . . . . . . . . . . . 150

7.6.7 Design Principles for the Slitted Decoupler . . . . . . . . . . . . . 152

7.7 A Dipole on the Slitted Decoupler . . . . . . . . . . . . . . . . . . . . . . 153

7.7.1 Induced Voltage in the Middle Port of the Dipole on the Decoupler154

7.7.2 Input Impedance of the Half Wavelength Dipole on the Decoupler 158

7.7.3 Power Collected by the Half Wavelength Dipole on the Decoupler 159

7.7.4 Antenna Design Principles for the Slitted Decoupler . . . . . . . . 161

7.8 Measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162

7.8.1 Measurement Facilities . . . . . . . . . . . . . . . . . . . . . . . . . 163

7.8.2 Measurement Results and Comparison . . . . . . . . . . . . . . . 165

7.9 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168

Chapter 8. Detection of Massive Numbers of DVDs 171

8.1 Introduction and Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . 172

8.1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172

8.1.2 An Operational Constraint . . . . . . . . . . . . . . . . . . . . . . 173

8.1.3 Some General Perspectives . . . . . . . . . . . . . . . . . . . . . . 173

8.1.4 Literature Treatments . . . . . . . . . . . . . . . . . . . . . . . . . 174

8.1.5 Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175

8.2 Parameters of a Packaged DVD Product . . . . . . . . . . . . . . . . . . . 177

Page ix

Contents

8.3 Theoretical Analysis and Simulation Verification of the Effect on a Uni-

form Plane Wave from a Thin Metal Film . . . . . . . . . . . . . . . . . . 179

8.3.1 Surface Resistance of a Thin Metal Film . . . . . . . . . . . . . . . 179

8.3.2 Simulation on a DVD Disc . . . . . . . . . . . . . . . . . . . . . . . 183

8.4 Investigation of Tag Labelling Method . . . . . . . . . . . . . . . . . . . . 185

8.4.1 Tag Lying on the Case Cover . . . . . . . . . . . . . . . . . . . . . 186

8.4.2 Tag Lying on the Case Faces: Opening A and Spine . . . . . . . . 188

8.4.3 Tag Folded on the Case Faces: Opening A and Spine . . . . . . . 189

8.5 DVD Detection in a Stack . . . . . . . . . . . . . . . . . . . . . . . . . . . 191

8.5.1 Testing Strategy and DVD Stack Description . . . . . . . . . . . . 192

8.5.2 Single Tagged DVD Film in a DVD Stack . . . . . . . . . . . . . . 195

8.5.3 Multiple Tag Detection in a DVD Stack . . . . . . . . . . . . . . . 198

8.6 Further Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202

8.6.1 Q Parameter in EPC C1G2 Protocol for Anti-Collision . . . . . . . 202

8.6.2 Method of Packaging and Stacking DVDs in Industry . . . . . . . 203

8.6.3 Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205

8.7 The Optimisation of the Distance Between the Reader Antenna and the

DVD stack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215

8.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217

8.8.1 Stacking Policies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217

8.8.2 Results for Side and Base Stacking . . . . . . . . . . . . . . . . . . 218

8.8.3 Further Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219

Chapter 9. Conclusions and Future Work 221

9.1 Review of and Conclusions from the Work in This Thesis . . . . . . . . . 222

9.2 Recommendations on Future Work . . . . . . . . . . . . . . . . . . . . . . 224

9.3 Summary of Original Contributions to Knowledge . . . . . . . . . . . . . 228

9.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230

Page x

Contents

Appendix A. Tests of the Tags in Chapter 5 231

A.1 Test Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232

A.2 Test Result on the Semi-finished Security Tag . . . . . . . . . . . . . . . . 233

A.3 Test Result on the Final Design of the Security Tag . . . . . . . . . . . . . 233

Appendix B. Open Circuit Voltage of A Half Wavelength Dipole 235

Appendix C. Original Testing Data Corresponding to the Work in Section 8.4 239

Appendix D. Evaluation of Reflections in the Aperture Surrounded by Absorbing

Foam Used in Chapter 8 241

D.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242

D.2 Reflection Coefficient of Waves Incident on a Lossless Dielectric Interface 242

D.3 The Structure of the Absorbing Foam and Its Reflection Coefficient . . . 246

D.4 Reflection in the Aperture Surrounded by the Absorbing Foam . . . . . 248

D.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250

Bibliography 253

Page xi

Page xii

Abstract

Radio frequency identification (RFID) is an auto-identification technology realised by

radio waves. The ultimate goal of RFID is the item-level tagging of all kinds of prod-

ucts in supply chains. This goal challenges industry and academia in many aspects.

Passive UHF RFID systems, when compared with other RFID systems, are believed to

possess advantages in achieving that goal. However, UHF RFID systems possess two

serious disadvantages: (i) the relatively large antenna size, and (ii) the sensitiveness

to the metallic items on which a tag is mounted. Those two deficiencies make a large

number of small size objects and metallic objects hard to tag. In addition, different

applications also bring special requirements or limitations in adopting UHF RFID sys-

tems, such as in the case of a container seal, the requirement for tags to have a physical

security function, and in other cases such as pallet shipping, the requirement for de-

tecting massive numbers of items densely stacked together. Finally, of course, cost is

one of the key limitations if one intends to apply his or her design down to item-level

tagging commercially. Hence each of the inherent deficiencies of the system itself and

the limitations caused by the application, or a combination of all or some of the de-

ficiencies and limitations make a large number of items hard to tag and impedes the

item-level tagging target.

The research in this thesis aims, by antenna design and electromagnetic wave analysis,

to provide feasible and affordable solutions for some of those hard-to-tag objects in

UHF RFID systems, and the thesis can be divided into five parts.

In detail, the first part of the thesis gives the motivations, contributions and structure

of this thesis. In addition it also provides a brief introduction to RFID systems and

about how they are operated, developed, classified, regulated and standardised.

The second part of this thesis presents basic terminologies and design criteria in tag

antenna design, transponder IC design and reader design. Factors which limit the op-

erating range of UHF RFID systems are discussed. Following this discussion, a novel

Page xiii

Abstract

method making use of a scattering matrix for evaluating the operating range of a UHF

RFID system deployed in an arbitrary environment is proposed.

In the third part, concerning the meander line dipole antenna (MDA), one of the ap-

proaches to minimising tag antenna size is analysed in terms of its resonant frequency,

size reduction contributors, radiation pattern and efficiency. An analytic formula for

calculating the resonant frequency of an MDA on a dielectric substrate as an RFID tag

antenna is established. Based on the analysis, a novel tag antenna with a physical secu-

rity function (an electronic seal) for protecting shipping containers was designed and

experimentally verified.

The fourth part of this thesis puts emphasis on metallic item detection. The reason of

why common dipole based tag antennas cannot work well in close proximity to metal

is given. Previous solutions and their own demerits in solving this problem are sum-

marised. Then, a low profile, simple structure, compact size solution is introduced via

the artificial magnetic conductor concept. Furthermore, a general DVD disc contains a

very thin metal layer inside for the purpose of reflecting laser. That layer may not bring

many troubles in identifying a single DVD by a UHF RFID system, but if thousands

of DVDs were stacked, the role the metal component plays in degrading the detection

of each DVD in the stack should be investigated. An approach in detecting a large

number of DVDs (up to 2000) densely stacked is thus presented.

Conclusions of the work in this thesis are drawn as the last part of the thesis. Besides

conclusions the last part also includes some recommendations for future work and the

description of the original contributions of this thesis.

The potential benefits of item-level tagging in supply chains are enormous. The exis-

tence of a large number of hard-to-tag objects is one of the main challenges in achieving

item-level tagging. The studies in this thesis extend the scope of the detectable objects

and this extension makes item-level tagging more realisable.

Page xiv

Statement of Originality

This work contains no material that has been accepted for the award of any other de-

gree or diploma in any university or other tertiary institution and, to the best of my

knowledge and belief, contains no material previously published or written by an-

other person, except where due reference has been made in the text.

I give consent to this copy of the thesis, when deposited in the University Library,

being available for loan, photocopying and dissemination through the library digital

thesis collection.

The author of this thesis acknowledges that copyright of published work contained

within this thesis (as listed in the publications page) resides with the copyright holder(s)

of that work.

Signed Date

Page xv

Page xvi

Acknowledgments

First and foremost, I must recognise my principal supervisor Prof. Peter H. Cole for

his constant and patient guidance during the period of my Ph.D study. His meticulous

approach to learning and tolerance towards others will influence me for the rest of my

life. Talking with him not only in professional areas but also in music and culture has

always been enjoyable and beneficial. In addition, many thanks to him for generously

providing my living allowance.

I would also like to wholeheartedly thank my co-supervisors Dr Christophe Fumeaux

and Dr Christopher Coleman for providing valuable suggestions to my research.

Sincere thanks to my colleagues and also to successful graduates in the Auto-ID Lab,

Adelaide for their unselfish help in extending my knowledge. They are Behnam Jamali,

David Hall, Damith Ranasinghe, Ng Mun Leng and Kin Seong Leong. Thanks also to

Mr Alfio R. Grasso for arranging projects with industrial partners.

I am indebted for the work done by the staff in the School of Electrical and Elec-

tronic Engineering, particularly, Mr Pavel Simcik and Mr Brandon Pullen who fabri-

cated most of my designs, and Associate Professor Michael Liebelt, Associate Professor

Cheng Chew Lim and Mr Stephen Guest who managed my scholarship and travelling

issues. Of course thanks to the four kind ladies in the school office.

Many thanks to my friends, specifically Thomas McLean, Matthew Trinkle, Yang Ruit-

ing, Guo Bin, Wang Yuexian in Adelaide and also Liu Tan, Wu Xiao, Ye Yang in China

for their constant support and encouragement during my Ph.D studies.

I am grateful to my parents who dedicate their love to me. Since I first entered primary

school, all through my educational journey of twenty years, their love has always ac-

companied me. Last but not least, thanks for the unconditional love and encourage-

ment from my girlfriend Xin Xia. Without her, I could not imagine how I could have

accomplished this work.

Zhonghao Hu (September 2010)

Page xvii

Page xviii

Conventions

Typesetting

This thesis is typeset using the LATEX2e software.

The fonts used in this thesis are Times New Roman and Sans Serif.

Referencing

Referencing and citation style in this thesis are based on the Institute of Electrical and

Electronics Engineers (IEEE) Transaction style [1].

For electronic references, the last accessed date is shown at the end of a reference.

Units

The units used in this thesis are based on the International System of Units (SI units) [2].

Spelling

The Australian English spelling is adopted in this thesis.

Page xix

Page xx

Publications

Book Chapter

[1] P. H. Cole, L. Turner, Z. Hu, and D. Ranasinghe, “The Future of RFID,” in Unique Radio Innovation

for the 21st Century, D. Ranasinghe, M. Sheng, and S. Zeadally, Eds. Springer, 2010.

Book Chapter Accepted

[1] P. H. Cole and Z. Hu, “Operating Range Evaluation of UHF RFID Systems,” in Advances in RFID

Tags. Vienna, Austria: InTech, 2010.

[2] Z. Hu, P. H. Cole, and C. Fumeaux, “Analysis and Design of Meander Line Dipole Antennas,” in

Chipless Radio Frequency Identification: Systems for Ubiquitous Tagging, N. Karmakar, Ed. Hershey,

USA: IGI Global, 2011.

Journal Accepted

[1] Z. Hu, P. H. Cole, and A. Grasso, “Compact solution for metallic item detection in RFID systems by

means of artificial magnetic conductor,” International Journal of Radio Frequency Identification Technol-

ogy and Applications (IJRFITA), 2010.

[2] Z. Hu and P. H. Cole, “Detection of Massive Numbers of DVDs by a UHF RFID system,” Progress In

Electromagnetics Research B (PIER-B), 2010.

Conference

[1] Z. Hu, P. H. Cole, and L. Zhang, “A method for calculating the resonant frequency of meander-line

dipole antenna,” in 4th IEEE Conference on Industrial Electronics and Applications, ICIEA 2009, Xi’an,

China, May 2009, pp. 1783–1786.

[2] Z. Hu and P. H. Cole, “Detection of DVDs in a stack by an RFID system,” in Asia-Pacific Symposium

on Electromagnetic Compatibility, APEMC 2010, Beijing, China, April 2010.

[3] Z. Hu and P. H. Cole, “The Slitted Decouple Design for Metallic Item Detection in UHF RFID Sys-

tems,” in Asia-Pacific Symposium on Electromagnetic Compatibility, APEMC 2010, Beijing, China, April

2010.

Page xxi

Publications

[4] Z. Hu and P. H. Cole, “Bottle Packaged Wine Product Detection By UHF RFID Systems,” in In-

ternational Conference on Electromagnetics in Advanced Applications, ICEAA 2010, Sydney, Australia,

Septermber 2010.

Non-refereed

[1] P. H. Cole and Z. Hu, “Solving the Water and Metal Problem,” RFID Journal, April 2009. [Online].

Available: http://www.rfidjournal.com/article/view/4755 [29 July 2010].

[2] P. H. Cole and Z. Hu, “Every DVD Tells a Story,” RFID Journal, July 2010. [Online]. Available:

http://www.rfidjournal.com/article/view/7717 [29 July 2010].

Page xxii

Abbreviations

AC Alternating Current

ACMA Australia Communications and Media Authority

AMC Artificial Magnetic Conductor

ASK Amplitude Shift Keying

BOPP Biaxially Oriented Polypropylene

BTA Bow Tie Antenna

CMOS Complementary Metal Oxide Semiconductor

CPS Coplanar Strip

CST Computer Simulation Technology (a commercial simulation software)

DC Direct Current

DVD Digital Video Disc

DVD R Recordable Digital Video Disc

DVD-ROM Digital Video Disc-Read Only Memory

EAS Electronic Article Surveillance

EAN European Article Numbering

EBG Electromagnetic Band Gap

EEPROM Electrically Erasable Programmable Read Only Memory

EIRP Equivalent Isotropic Radiated Power

EPC Electronic Product Code

Page xxiii

Abbreviations

ERP Effective Radiated Power

ETSI European Telecommunications Standards Institute

FCC Federal Communications Commission

FDA Folded Dipole Antenna

FEM Finite Element Method

FeRAM Ferroelectric Random Access Memory

FHSS Frequency Hopping Spread Spectrum

FSS Frequency Selective Surfaces

GA Genetic Algorithms

GS1 Global Standards 1

HF High Frequency

HFSS High Frequency Structural Simulator (a commercial simulation software)

IC Integrated Circuit

IEC International Electrotechnical Commission

IFA Inverted F Antenna

IPICO Intellectual Property and Innovation Company

ISM Industrial, Scientific and Medical (frequency range)

ISO International Organization for Standardization

LBT Listen Before Talk

LF Low Frequency

MDA Meander Line Dipole Antenna

MOM Method Of Moments

Page xxiv

Abbreviations

PBG Photonic Band Gap

PIE Pulse Interval Encoding

PP Polypropylene

PSK Phase Shift Keying

PZT Lead Zirconate Titanate

RAM Random Access Memory

RCS Radar Cross Section

RF Radio Frequency

RFID Radio Frequency Identification

SAW Surface Acoustic Waves

SBT Strontium Bismuth Tantalate

SHF Super High Frequency

SMA Sub-Miniature version A

UCC Uniform Code Council

UHF Ultra High Frequency

UID Ubiquitous Identification

Page xxv

Page xxvi

List of Figures

1.1 Components of a basic RFID system . . . . . . . . . . . . . . . . . . . . . 2

1.2 The tree diagram of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . 12

2.1 Power limitation in EN 302 208 . . . . . . . . . . . . . . . . . . . . . . . . 19

3.1 Thevenin equivalent of a receiving antenna . . . . . . . . . . . . . . . . . 27

3.2 Thevenin equivalent of a transponder . . . . . . . . . . . . . . . . . . . . 27

3.3 Coordinate used in the definition of effective length . . . . . . . . . . . . 30

3.4 Block chart of a transponder . . . . . . . . . . . . . . . . . . . . . . . . . . 37

3.5 Two port junction representing coupled antennas in an RFID system . . 50

3.6 A self-made tag used in experiment . . . . . . . . . . . . . . . . . . . . . 55

3.7 The chip impedance illustration . . . . . . . . . . . . . . . . . . . . . . . . 55

3.8 A shielding tunnel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

3.9 Comparison between the reading range calculated by (3.61) after deriv-

ing the S parameters from the simulation and the tested reading range . 59

4.1 A sample of meander line dipole antenna with approximate current dis-

tribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

4.2 Meander line dipole antenna loaded with two meanders . . . . . . . . . 64

4.3 Three models of MDA with various numbers of meander lines . . . . . . 66

4.4 The resonant frequency of MDA as a function of its physical parameters 67

4.5 Two coplanar strips on a dielectric substrate . . . . . . . . . . . . . . . . . 70

4.6 The transverse electric field distribution in the cross section of an CPS

on board . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

Page xxvii

List of Figures

4.7 An MDA loaded with four identical meander lines . . . . . . . . . . . . . 72

4.8 Cross section view of electric field magnitude distribution of the MDA

shown in Figure 4.7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

4.9 The relative effective permittivity of the MDA in Figure 4.7 . . . . . . . . 74

4.10 The variation of the MDA’s electric field magnitude distribution at the

resonant frequency along with the variation of the εr . . . . . . . . . . . 76

4.11 An MDA loaded with three different meander lines . . . . . . . . . . . . 77

4.12 The relative effective permittivity of the MDA in Figure 4.11 . . . . . . . 78

4.13 The half MDA on a ground plane being tested . . . . . . . . . . . . . . . 79

4.14 Smith chart derived by the network analyser 8714C showing input impedance

of the half MDA on a ground plane . . . . . . . . . . . . . . . . . . . . . . 80

4.15 A tag based on the MDA in Figure 4.11 . . . . . . . . . . . . . . . . . . . 80

4.16 Radiation efficiency comparison between two types of MDA . . . . . . . 83

5.1 T-seal structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

5.2 A regular sample of MDA with two meander lines . . . . . . . . . . . . . 91

5.3 The semi finished tag shape . . . . . . . . . . . . . . . . . . . . . . . . . . 93

5.4 Simulated gain pattern of the semi finished tag antenna by HFSS . . . . 94

5.5 A fabricated sample of semi finished tag . . . . . . . . . . . . . . . . . . . 95

5.6 Semi finished tag with a loop in the down-narrow part of the board . . . 96

5.7 Simulated gain pattern of the semi finished tag antenna with a complete

loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96

5.8 Simulated gain pattern of the semi finished tag antenna with a incom-

plete loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

5.9 The final design of the security tag . . . . . . . . . . . . . . . . . . . . . . 100

5.10 Simulated gain pattern of Tag1 . . . . . . . . . . . . . . . . . . . . . . . . 100

5.11 Simulated gain pattern of Tag2 . . . . . . . . . . . . . . . . . . . . . . . . 101

Page xxviii

List of Figures

6.1 Boundary conditions at a perfect conductor surface . . . . . . . . . . . . 105

6.2 Electric field when a charge is put above the perfect conductor . . . . . . 106

6.3 A straight wire carrying current and its image underneath the ground

plane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107


6.5 Reading ranges of labelled commercial tags when they are placed above

the aluminium plate at various distances . . . . . . . . . . . . . . . . . . 111

6.6 Side view of an antenna placed at one quarter wavelength distance above

a metal plate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112

6.7 Sievenpiper high impedance electromagnetic surface . . . . . . . . . . . 115

6.8 Origin of the capacitance and inductance in each cell . . . . . . . . . . . 116

6.9 Three conductive layer high impedance electromagnetic surface . . . . . 116

6.10 Hilbert curve in various orders . . . . . . . . . . . . . . . . . . . . . . . . 117

6.11 Hilbert curve AMC based on order 4 Hilbert curves . . . . . . . . . . . . 119

7.1 The structure of the slitted decoupler . . . . . . . . . . . . . . . . . . . . . 124

7.2 Slitted decoupler placement illustration . . . . . . . . . . . . . . . . . . . 124

7.3 The simulated slitted decoupler . . . . . . . . . . . . . . . . . . . . . . . . 127

7.4 Magnitude of the r.m.s. phasors representing the simulated electric fields

of the slitted decoupler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128

7.5 The magnitude of y-directed electric field variation along the y and x

axes at various heights. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129

7.6 The structure of a simple rectangular patch antenna without excitation . 130

7.7 Electric field distribution of a rectangular patch antenna . . . . . . . . . . 131

7.8 Patch width and length values making the antenna resonant at 923MHz 133

7.9 Charge and current distribution in a rectangular patch antenna . . . . . 134

Page xxix

List of Figures

7.10 A rectangular patch antenna fed by a coaxial cable . . . . . . . . . . . . . 137

7.11 A typical input impedance of patch antenna as a function of frequency . 138

7.12 The equivalent circuit of a patch antenna which is fed by a coax cable . . 139

7.13 The comparison between the simulation results and the theoretical re-

sults in terms of patch size at resonance . . . . . . . . . . . . . . . . . . . 140

7.14 The comparison between the simulation results and the theoretical re-

sults in terms of the resonant input impedance . . . . . . . . . . . . . . . 141

7.15 The r.m.s phasor of the electric field distribution underneath top patch

obtained by HFSS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142

7.16 The y-directed electric fields as a function of the patch width at various

patch length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143

7.17 The structure of the slitted decoupler illuminated by a uniform plane wave144

7.18 The y-directed electric fields in the slit as a function of the patch width

at various patch length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146

7.19 The y-directed electric fields in the slit as a function of the dielectric layer

thickness at the particular patch size 90.5mm×32.5mm . . . . . . . . . . 147

7.20 The y-directed electric fields in the slit at various slit width . . . . . . . . 148

7.21 The y-directed electric fields in the slit as a function of the patch width

at various patch lengths when the loss tangent is increased to 0.02 . . . . 149

7.22 Slitted decoupler with a ground plane larger than the top layer . . . . . 150

7.23 The y-directed electric fields in the slit as a function of margin at 923MHz 151

7.24 A dipole on the slitted decoupler . . . . . . . . . . . . . . . . . . . . . . . 154

7.25 The induced voltage comparison among the dipole on the slitted decou-

pler, the dipole on the metal and the dipole in free space . . . . . . . . . 155

7.26 The induced voltage of a short dipole on the slitted decoupler . . . . . . 157

7.27 The induced voltages of the half wavelength dipole on the slitted decou-

pler as a function of the slit width . . . . . . . . . . . . . . . . . . . . . . . 158

7.28 The input impedance of the dipole in various distances above the slitted

decoupler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159

Page xxx

List of Figures

7.29 The four fabricated slitted decouplers . . . . . . . . . . . . . . . . . . . . 165

7.30 The placement of the tag on both the decoupler and the plate. . . . . . . 167

8.1 The structure of a regular DVD case and the SPI code on it . . . . . . . . 178

8.2 The structure of a regular DVD disc . . . . . . . . . . . . . . . . . . . . . 178

8.3 Transmission line model of a uniform plane wave perpendicularly inci-

dent on an infinite aluminium metal film . . . . . . . . . . . . . . . . . . 180

8.4 Simulation model of the square aluminium film . . . . . . . . . . . . . . 181

8.5 Total electric field distribution shown in the xz plane of the simulation

on the square aluminium film . . . . . . . . . . . . . . . . . . . . . . . . . 182

8.6 Simulation model of the aluminium film in the disc . . . . . . . . . . . . 184

8.7 Total electric field distribution shown in the xz plane of the simulation

on the aluminium film in the disc . . . . . . . . . . . . . . . . . . . . . . . 184


8.9 Tag lying on the case cover . . . . . . . . . . . . . . . . . . . . . . . . . . . 186

8.10 Tag lying on the case faces: opening A and spine . . . . . . . . . . . . . . 188

8.11 Tag folded on a DVD case . . . . . . . . . . . . . . . . . . . . . . . . . . . 189

8.12 Tag staggered on a DVD case . . . . . . . . . . . . . . . . . . . . . . . . . 190

8.13 Three selected testing schemes . . . . . . . . . . . . . . . . . . . . . . . . 192

8.14 Testing strategy illustration . . . . . . . . . . . . . . . . . . . . . . . . . . 193

8.15 Three forms of testing a DVD stack in terms of the three testing schemes

shown in Figure 8.13 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194

8.16 Aperture structure illustration . . . . . . . . . . . . . . . . . . . . . . . . . 195

8.17 Two types of DVD stack in the aperture . . . . . . . . . . . . . . . . . . . 196

8.18 The level and floor division of the stack shown in Figure 8.17(a) . . . . . 199

8.19 Flow chart of the method examining the testing schemes “1”, “2” and

“3” respectively. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200

Page xxxi

List of Figures

8.20 DVD carton and its dimension . . . . . . . . . . . . . . . . . . . . . . . . 203

8.21 A sample of a real pallet. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204

8.22 The DVD stack structure for testing scheme “3” . . . . . . . . . . . . . . 206

8.23 The real DVD stack for testing scheme “3” . . . . . . . . . . . . . . . . . . 207

8.24 The reader antenna’s positions in relation to the stack in terms of the

testing scheme “3” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207

8.25 The DVD stack structure for testing scheme “2” . . . . . . . . . . . . . . 210

8.26 The real DVD stack for testing scheme “2” . . . . . . . . . . . . . . . . . . 211

8.27 The reader antenna’s positions in relation to the stack in terms of the

testing scheme “2” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212

8.28 The DVD stack structure for testing scheme 2. The reader scans the back

side of the stack. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213

8.29 Illustrating reflection symmetry of tag positions about a vertical mid-plane215

8.30 The variation of the reader antenna input impedance in the form of

Smith Chart along with the variation of the distance between the reader

antenna and the DVD stack de measured by the network analyzer . . . . 216

A.1 A shielding tunnel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232

A.2 Two tested tags . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233

B.1 A half wavelength dipole in the rectangular coordinate system . . . . . . 237

B.2 |Vinr.m.s| as a function of the ratio z/λ at 923MHz . . . . . . . . . . . . . . 238

D.1 Plane wave incident on a dielectric interface . . . . . . . . . . . . . . . . . 243

D.2 The reflection coefficient at a dielectric interface as a function of incident

angle, for εr = 1.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245

Page xxxii

List of Figures

D.3 Cross section of the absorbing foam . . . . . . . . . . . . . . . . . . . . . 247

D.4 Aperture structure illustration . . . . . . . . . . . . . . . . . . . . . . . . . 249

D.5 The deployment of the reader antenna in front of the aperture . . . . . . 251

Page xxxiii

Page xxxiv

List of Tables

2.1 Comparison among RFID systems . . . . . . . . . . . . . . . . . . . . . . 17

2.2 Regulation status of UHF RFID among countries . . . . . . . . . . . . . . 20

3.1 Comparison among RAM, EEPROM and FeRAM . . . . . . . . . . . . . 41

3.2 Reading ranges of the self-made tag in proximity to the aluminium plate

by experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

3.3 Reading ranges of the self-made tag in proximity to the aluminium plate

calculated by (3.61) after deriving S parameters from the simulation . . . 59

6.1 Reading ranges of commercial tags in free space . . . . . . . . . . . . . . 110

7.1 Geometrical parameters of the fabricated slitted decoupler . . . . . . . . 165

7.2 Reading ranges of the tag on the decouplers varied in size . . . . . . . . 166

7.3 Reading ranges of the tag on both the decoupler and the aluminium plate 167

7.4 Reading ranges of the tag above the decouplers in a certain distance

(Dz=6.5mm) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168

8.1 Reading range test results of the tag shown in Figure 8.9 . . . . . . . . . 187



8.4 Outside reading range of the tag at the end of the DVD stack . . . . . . . 197

8.5 Pallet top surface dimensions standardised by ISO . . . . . . . . . . . . . 205

8.6 Misreading tag distribution in the stack shown in Figure 8.24(a). For

these results the reader antenna occupied four positions. . . . . . . . . . 208

8.7 Misreading tag distribution in the stack shown in Figure 8.24(b). For

these results the reader antenna occupied twelve positions. . . . . . . . . 210

8.8 Misreading tag distribution in the stack shown in Figure 8.27(a). For

these results the reader antenna occupied four positions. . . . . . . . . . 212

Page xxxv

List of Tables

8.9 Misreading tag distribution in the stack shown in Figure 8.27(b). For

these results the reader antenna occupied ten positions. . . . . . . . . . . 212

8.10 Misreading tag distribution when those tags are read from the back of

the stack. For these results the reader antenna occupied ten positions. . . 214

C.1 Original testing data corresponding to Table 8.1, unit: mm . . . . . . . . 240



Page xxxvi

Chapter 1

Introduction andMotivation

THIS chapter gives a brief introduction to the research area (RFID

systems) of this thesis. The motivation for doing the particular

work described in this thesis and the contributions of the thesis

are also presented. Finally, the thesis structure is discussed and the content

of each chapter is summarised.

Page 1

1.1 Research Area

1.1 Research Area

The research work in this thesis focuses on Radio Frequency Identification (RFID).

RFID is a type of automatic identification technology making use of radio waves. An

RFID system generally consists of three main components: (i) A tag. An RFID tag

is composed of an antenna and a chip. It is attached to an item for detection. (ii) A

reader. A reader contains a reader antenna and some signal processing circuits. (iii) A

host computer.

Figure 1.1 shows these three parts in a basic RFID system. The operating scheme of

that system is described as follows. Firstly, the reader sends an interrogating signal or

wave, represented by the red arrow in Figure 1.1 to an RFID tag. Then, the tag will

respond to the interrogating signal by scattering some identifying signal, denoted by

the green arrow in Figure 1.1, back to the reader. Once the signal scattered from the

tag is received by the reader, the signal will be transmitted to the host computer for

processing.

The Auto-ID Center has introduced an Electronic Product Code (EPC) concept [3]. In

this concept, each tag only contains a unique 96-bit long code; all the other informa-

tion is stored in the database. Hence, the object attached to an EPC tag has a unique

identity. The adoption of this concept avoids the requirement for large memory for

storing object information in tag chips and results in low tag costs. The proposal of

EPC concept makes the item level tagging possible.

Host Computer RFID Reader Radio Waves RFID tag attachedto an item

Reader Antenna

Tag

Figure 1.1. Components of a basic RFID system.

The introduction above demonstrates that RFID systems are complex systems which

involve various areas of study, such as antenna design, microwave communication

analysis, signal processing and integrated circuit design. This thesis puts emphasis

on the antenna design and the analysis of microwave communication in passive UHF

Page 2

Chapter 1 Introduction and Motivation

RFID systems. UHF is the abbreviation for ultra high frequency band. In RFID ap-

plications this frequency band is defined as extending from 860MHz to 960MHz. The

term “passive” means that there is no extra power supply in the tag. The backscattered

power of the tag comes from the radiating power of the reader antenna.

More details about RFID systems will be introduced in Chapter 2 and Chapter 3.

1.2 Motivation

The ultimate goal of RFID is the item-level tagging for all kinds of products in supply

chains. This goal challenges the industry and academia in many aspects.

As mentioned before, the proposal of the EPC concept makes the item level tagging

possible. However, the ECP concept is a superstructural concept in achieving item

level tagging; in terms of the application level, there are still many challenges.

Among various types of RFID systems, of which details are introduced in Chapter 2,

passive UHF RFID systems are believed to have advantages in meeting these chal-

lenges. However, UHF RFID systems also have some inherent deficiencies: (i) Because

of their working spectrum (around 1 GHz), the wavelength is about 300mm, and the

size of the tag antenna resonant within this spectrum has to be proportional to half

of the wavelength in order to obtain acceptable radiation performance, which makes

the tag antenna’s size relatively large. (ii) Also because of their working spectrum, the

radiation elements (antennas) and wave propagation in systems are very sensitive to

the metallic and liquid items which makes the detection of those items or deployment

of a system surrounded by those items difficult. In addition, different applications also

bring special requirements or limitations in adopting UHF RFID systems, such as in

the case of a container seal, the requirement for tags to have a physical security func-

tion, and in other cases such as pallet shipping, the requirement for detecting massive

numbers of items densely stacked together. Finally, of course, cost is one of the key

limitations if one intends to apply his or her design down to item-level tagging com-

mercially. Hence each of the inherent deficiencies of the system and limitations caused

Page 3

1.2 Motivation

by the application, or combinations of all or some of the deficiencies and limitations

make a large number of items hard to tag and impedes the item-level tagging target.

The research in this thesis aims to provide feasible and affordable solutions for some of

those hard-to-tag objects in UHF RFID systems by antenna design and electromagnetic

wave analysis. Those hard-to-tag items the detection of which this thesis intends to ac-

complish may be divided into the four categories shown in the following four bulleted

paragraphs.

• Physically small items

There are hundreds of thousands kinds of physically small items needed to be

detected. The tag antennas for identifying these small items have to be designed

to be smaller than, or at least as the same size as, the items themselves. Other-

wise, the tag will suffer from being bulky or easily damaged. The meander line

dipole antenna (MDA) is commonly used as the tag antenna pattern to address

this problem, because of its compact size and flexible shape compared with the

half wavelength dipole antenna. However, the conventional method of design-

ing an RFID tag antenna based on the MDA pattern is a numerical method and

to arrive at the optimal MDA design, such numerical computations have to be

iterated, with the results that the calculations are inefficient. Therefore, a simple

method or formula is desired to speed the MDA design cycle.

• Items with security requirement

Large numbers of valuable items, for example shipping containers, need protec-

tion from theft. Nowadays, the protection is usually provided by conventional

seals. However, an automated, real time, remote and high security level protec-

tion is expected by making use of UHF RFID systems. The expectation cannot

be satisfied by most of the commercial RFID tags. Hence, new tag designs with

security functions are desired.

• Metallic items

The metallic boundary condition tells us that there is no tangential electric field

on the surface of a metallic item. Hence, the dipole based tag antennas in which

Page 4


excitation is significantly dependent on that field cannot identify metallic items.

Many solutions to this problem have been reported, e.g. one quarter wavelength

isolator solution [4] [5] [6], antenna selection solutions [7] [8] [9] [10] [11] and

artificial magnetic conductor solutions [12] [13]. However, they all suffer from

their own weaknesses , such as being bulky, high cost or complex structure, when

these solutions are applied in RFID systems. Hence, other solutions for the metal-

lic items identification must be identified. Low cost, low profile, small size and

simple structure solutions are preferable.

• Massive numbers of DVDs

Packaged DVDs are very common commodities in our life. Currently, they are

identified by barcode systems. However, in supply chains, a large number of

DVDs (the large number is in the region of 2000) can be densely stacked together

for shipping or distributing. A barcode system cannot detect all of them without

unpacking the stack and scanning the DVDs one by one.

UHF RFID systems are thought to be potentially useful in solving this problem

because of the UHF radio waves’s long range propagation ability. However,

when thousands of DVDs are stacked on a pallet, there is no certainty that each

of them can be identified successfully, since the condensed placement of the tags

enhances the mutual coupling. In addition, the fact that the DVD disc contains a

thin metal layer inside influences the propagation and makes the detection even

harder. Hence, it is worth investigating the feasibility of UHF RFID systems in

solving this problem.

Since each DVD disc contains a very thin metallic layer to reflect the laser beam,

the detection of DVDs could be included in the scope discussed in the last itemi-

sation which is the detection of metallic items. But we choose to discuss this

problem separately, because the difficulties in achieving that detection are caused

by not just the metallic component in the DVD but by the combination of that fact

and the large number.

To allow a thorough and full deployment of RFID, the work in this thesis aims to pro-

vide some feasible solutions to satisfy the requirements of each of the categories of

Page 5

1.3 Original Contributions

hard-to-tag items which have been introduced above. While tackling these challenges,

the principle that providing practical solutions in terms of low cost, small size and

simple deployment is always taken into account.


The original contributions of this thesis begin in Chapter 3. The factors tag antenna

design, chip design, reader design and deployed environment analysis, which could

affect the operating range of a UHF RFID system, are summarised. Factors critically

deciding the reading range are identified. The limitations of the Friis equation which

is commonly adopted in analysing the operating range are discussed. In order to over-

come these limitations, a novel method in evaluating the operating range of a UHF

RFID system by making use of a scattering matrix is proposed. By using this method,

the operating range of the UHF RFID system deployed in an arbitrary environment

can be predicted.

In addition, following the discussion in Section 1.2, the main objective of this research

is to provide feasible solutions to the problem of detecting those hard-to-tag items in

various categories. Meanwhile, the solutions maintain a balance between cost, size,

performance and deployment. Hence, the original contributions are further extended

and expressed separately in the following four bulleted paragraphs as each paragraph

describes a solution corresponding to one category of hard-to-tag items introduced in

Section 1.2.

• The solution to detect physically small items

As mentioned in Section 1.2, simple formula is needed in designing MDA which

is used to detect physically small items. Endo et al. [14] proposed a useful an-

alytic formula to calculate the resonant frequency from an MDA’s geometrical

parameters when the MDA is in free space. However, the formula cannot anal-

yse the MDA’s resonant frequency when it is built on a dielectric substrate which

is sometimes the case of RFID applications. In addition, even for analysing the

MDA in free space, the result derived from the formula proposed by Endo et

Page 6


al. [14] is still not accurate for designing UHF RFID tag antennas, because of the

UHF RFID tag antennas’ special need for a conjugate impedance matching con-

dition to the chip.

Hence in Chapter 4, an original method for calculating the effective permittiv-

ity brought by a dielectric substrate to an MDA is proposed. Furthermore, after

considering the effective permittivity and the UHF RFID tag antennas’ special

impedance matching need, the formula proposed by Endo et al. [14] is modified

to be able to design the MDA on a dielectric substrate in terms of RFID appli-

cations. The combination of the modified formula and a simulation software

can speed the design cycle dramatically compared with just using the simulation

software.

• The solution to detect items with security requirement

Large number of valuable items, for example shipping containers need protec-

tion from theft. Nowadays, the protection is mainly provided by conventional

seals. However, this kind of seal has to be checked manually which results in low

efficiency in supply chains and also extra human labor expense. Moreover the

security level is low since the conventional seals can be easily duplicated.

An electronic seal realised by means of a passive UHF RFID tag to provide an

automatic protection and identification for containers at a high security level is

described in Chapter 5. The security function is carried out by the combination

of mechanical design and antenna design. A novel antenna based on an MDA

pattern is so designed, fabricated and tested in this work.

• The solution to detect metallic items

As mentioned in Section 1.2, the electromagnetic characteristic of metal brings

some obstacles in detecting metallic items by dipole based tag antennas. Many

solutions to this problem have been reported, but suffer from their own weak-

nesses, such as being bulky, high cost or complex in structure, when these solu-

tions are applied to RFID systems.

Page 7


A new device named as “Electromagnetic Radiation Decoupler” was invented by

Brown et al. [15] to address the antenna on metal problem. This device attracts

our attention since it is found to have benefits in thickness and simplicity. How-

ever, the inventors did not give explanations of their device. More significantly,

a disadvantage that the device is too large, particularly in its width can be iden-

tified. Brown et al. [15] drew the conclusion that there is a tradeoff between the

performance and the width of the decoupler.

The work in Chapter 7 provides an explanation of the operational principles of

this decoupler. Some parameters and factors such as the size of the attached

metallic item and the interaction between the tag and the decoupler, which are

left out by Brown et al. [15], are discussed. In addition, according to the expla-

nation of the device’s operational principles, the decoupler’s width is reduced

dramatically, meanwhile the performance is even enhanced. Both the benefits

are obtained only at the expense of the decoupler bandwidth which is a minor

factor in affecting the decoupler’s performance since the UHF RFID system does

not place strict demands on that factor.

• The solution to detect massive numbers of DVDs

As discussed in Section 1.2, there are difficulties in detecting a large number of

DVDs. The large number is considered to be in the region of 2000 and all of the

DVDs are stacked within about 1 stere volume.

To the best of our knowledge, there are no publications discussing such dense

detection either for DVDs or other items.

A complete solution to this problem of reading massive numbers of DVDs is

established after investigating the physical, electrical parameters of a packaged

DVD, and conducting many experiments. The exploration includes: (i) the la-

belling method of a tag on a DVD, which defines the position of the tag mounted

on the packaged DVD; (ii) the pattern of incident field which defines the polari-

sation of the reader antenna; (iii) the testing strategy which defines the reader an-

tenna’s placement in front of the DVD stack; and (iv) the stacking policy which

defines how the DVDs are stacked above a wooden pallet. In addition, after

Page 8


considering the real packaging and stacking methods used in industry, the ex-

periments on a significant portion of 2000 DVDs (320 DVDs) were conducted

to verify the success of the solution in its electro-magnetically complex environ-

ment.

The results indicate that with an appropriate labelling method, pattern of inci-

dent field, testing strategy and stacking policy, perfect detection of 2000 DVDs in

a stack can be realised.

Besides solving the problem of reading massive numbers of DVDs, more crit-

ically, the solution sets an example and method for successors who intend to

detect massive numbers of other kinds of items in a dense placement.

1.4 Thesis Structure

The thesis can be divided into five parts. The first part provides the motivations, contri-

butions and structure of this thesis. In addition it also gives a brief introduction to RFID

systems about how it is operated, developed, classified, regulated and standardised.

This part includes Chapters 1 and 2. The second part is Chapter 3 in which the operat-

ing range of a UHF RFID system is evaluated. The third part, including Chapter 4 and

Chapter 5, introduces the analysis and applications of meander line dipole antennas

which is one of approaches to minimise the size of the tag antenna and provides secu-

rity function. The forth part is composed of Chapter 6, Chapter 7 and Chapter 8 which

mainly discuss the reason why the common tags are hard to detect when they are close

to metallic items and solutions to this problem. The last part is Chapter 9 describing

conclusions of the work done in this thesis and making recommendations of possible

work to extend this research in the future. The content in each chapter is summarised

as follows. A tree diagram of the thesis is also given in Figure 1.2.

Chapter 1 provides a brief introduction to RFID systems. In addition, the motivation

for doing the work, the contributions to knowledge provided by the thesis, and the

structure of the thesis are also discussed.

Page 9


Chapter 2 presents the background of RFID. In detail, the background includes RFID

history, classification, regulations and standards.

Chapter 3 introduces the operating range evaluation of UHF RFID systems. Within

this chapter, some fundamental ideas underlying the design of UHF RFID systems are

discussed and evaluated. A novel method for analysing the operating range of a UHF

RFID system in an arbitrary environment by means of a scattering matrix is proposed.

The research topic of this thesis is also narrowed down into meeting some particular

challenges in RFID applications.

Chapter 4 describes an analytic method for obtaining the resonant properties of mean-

der line dipole antennas (MDAs) either in free space or on a dielectric substrate in the

context of an RFID tag antenna design. MDAs provide one approach to the detection

of physically small items.

Chapter 5 provides a security tag design. The tag is designed to protect any container

(large or small) that has either a) two parts that join together to create a sealing chamber

or b) a finger that slots into a chamber, for example as in a shipping containers, from

theft. The security function of this tag is achieved by the combination of the mechanical

design and the tag antenna design.

Chapter 6 gives the reasons underlying difficulties in detecting metallic items by RFID

systems. The problem is well known as the antenna on metal problem. The existing

solutions to this problem found in an extensive literature survey are summarised and

the disadvantages of these solutions are discussed.

Chapter 7 introduces one solution to detecting metallic items in UHF RFID systems.

This solution is called the slitted decoupler solution. Compared with other solutions

summarised in Chapter 6, the slitted decoupler offers some significant advantages,

such as a low profile, simple structure and compact size.

Chapter 8 establishes a complete solution to the problem of reading a large number of

DVDs in a stack. The large number of DVDs considered here is in the region of 2000,

and those DVDs are densely stacked within the minimum practicable volume. The

solution has also been verified by experiments.

Page 10


Chapter 9 is the last chapter in this thesis. It reviews and concludes the thesis. In

addition, some recommendations for future work are given. Finally, the original con-

tributions to knowledge are re-summarised.

Page 11


Intro

du

ction

&B

ackgro

un

d

So

lutio

ns to

Hard

-to-T

agItem

s

Con

clusio

ns

&F

utu

reW

ork

Ch

ap

ter1

Intro

du

ction

&M

otiv

ation

Ch

ap

ter2

RF

IDB

ackg

rou

nd

Ch

ap

ter5

AS

ecurity

Tag

Desig

n

Ch

ap

ter6

So

lutio

ns fo

r the

An

tenn

a on

Metal

Pro

blem

Ch

ap

ter7

Th

e Slitted

Deco

up

lerD

esign

Ch

ap

ter9

Co

nclu

sion

san

dF

utu

reW

ork

Th

esis

Ap

pen

dix

A

Ch

ap

ter4

An

alysis an

d D

esign

of M

eand

er Lin

eD

ipo

leA

nten

nas

Ch

ap

ter8

Detectio

n o

fM

assive N

um

bers

of D

VD

s

Ap

pen

dix

BA

pp

end

ix C

& D

Operatin

g R

ang

eE

valu

ation

Ch

ap

ter3

Op

erating

Ran

ge

Ev

aluatio

no

f RF

ID sy

stems

So

lutio

ns to

Metallic ItemD

etection

So

lutio

ns to

Hard

-to-T

agItem

s

Mean

der L

ine

Dip

ole

An

tenn

as

Fig

ure

1.2

.T

he

treedia

gra

mofth

eth

esis.

Page 12

Chapter 2

RFID Background

THIS chapter contains more details of RFID systems than were

introduced in Chapter 1. The history, classification, regulations

and standards of RFID systems in this chapter provide a clear

view of RFID background, which is significant to understand the following

contents in this thesis.

Page 13

2.1 Introduction

2.1 Introduction

Before addressing some particular aspects in RFID systems, it is significant to under-

stand some background of RFID. Hence, the history, classification, regulations and

standards of RFID systems are introduced in this chapter.

The history of RFID demonstrates a clear sequence of its development and related

technology. The classification provides options and explanations in choosing a suitable

RFID system for a particular application. Moreover, the regulations and standards

establish boundaries on how the technology may be applied.

2.2 The History of RFID

RFID systems are based on the man’s understanding of the Electromagnetic world. The

significant subsequent developments of RFID are listed in the following itemisations.

• In 1906, Ernst F. W Alexanderson achieved the first continuous wave radio gen-

eration and transmission of radio signals. The accomplishment critically acceler-

ated the process of the birth of radar, which is the original application of RFID

[16].

• In 1935, probably the earliest active RFID system was invented by Sir Robert

Alexander Waston-Watt. This system could tell friendly aircraft from unfriendly

aircraft by the signal transmitted from the former [17].

• In 1948, one landmark paper of RFID’s theoretical foundation by Harry Stock-

man, “Communication by means of Reflected Power”, was published [18]. It is

discussed in this paper that the backscattered wave can be utilised to detect a

remote object.

• In the 1960s, researchers accumulated more theoretical knowledge, for exam-

ple, in 1964, R. F. Harrington published his paper, “Theory of loaded scatterers”.

Moreover, some inventors focused on RFID related inventions, such as “Remote

activated radio frequency powered devices” by Robert Richard [16]. Because of

Page 14

Chapter 2 RFID Background

the accumulated knowledge, the wheel of the RFID’s commercial use was turn-

ing. In the late 1960s, electronic article surveillance (EAS) system was deployed

to cope with theft of merchandise. This early stage of RFID products was af-

fordable due to the adoption of 1 bit tag. A 1 bit tag only states two situations:

presence or absence. When the tag, attached to an object, exceeds the surveillance

area without any legal permission, an alarm is triggered. The details of EAS are

introduced in [19].

• In the 1970s, further development was achieved. In 1972, the patent “Electronic

Surveillance System” was presented by Cole [20] [21], in which surface acous-

tic waves (SAW) are made use of to excite passive tags and coded signals can

be returned to a reader. In 1975, Koelle et al. in Los Alamos Scientific Labora-

tory, Northwestern University, America, presented “Short-Range Radio Teleme-

try for Electronic Identification Using Modulated Backscatter”. This develop-

ment achieved a practical short range-operation by passive tags [16]. Not only

the academia but also some large companies were actively working on RFID,

such as Raytheon’s Raytag and Richard Klensch of RCA [16]. Meanwhile, the

functionality and miniaturisation were improved since low power-consumption

CMOS logic circuits started to be involved in tag design.

• In the 1980s, the technology of other aspects brought improvement to RFID, for

example, the applications of personal computer enabled efficient and economical

data collection and analysis for RFID systems. Moreover, the development of

integrated circuits (IC) resulted in further reductions of tags’ dimensions [16].

In the late part of this decade, the original toll-collection systems by RFID were

installed in Norway (1987) and the United States (1989) [16].

• In the 1990s, the biggest achievement was that RFID toll collection systems were

widely applied. This was resulted from three main improvements. Firstly, the toll

could be charged even when cars are driven at high speeds. In addition, the com-

bination of toll collection and traffic management systems appeared. Thirdly, the

multi-protocol capability in toll collection systems was improved which means

cars attached with different protocol tags can be charged in the same system.

Page 15

2.3 RFID Classification

Meanwhile, the development in other areas also enhanced RFID systems’ perfor-

mance. In this decade, the schottky diodes were integrated in a CMOS circuit

which enabled smaller size and higher power efficiency [16].

• In 1999, the Auto-ID Center was founded at the Massachusetts Institute of Tech-

nology. A new concept, the Electronic Product Code (EPC), was proposed by

this organisation. The data stored in a tag is just a serial number instead of com-

plex information. The data related to the serial number can be accessible over a

networked database. This concept decreases the costs of the chip manufacture

dramatically because of less memory being needed in chips. After four years de-

velopment, the Auto-ID center had attracted more than 100 large companies to

support this project. Several labs had also been opened in Australia, The United

Kingdom, Switzerland, Japan, Korea and China to provide technology support.

Another organisation EPCglobal was founded to develop the numbering scheme

and standard of RFID systems [22]. Additionally, International Organization for

Standardization (ISO) [23] also joined into the establishment of RFID standards.

These standards are introduced in Subsection 2.4.2.

2.3 RFID Classification

2.3.1 Mode of Excitation

By powered mode, RFID devices can be divided into two categories: active and pas-

sive. An active tag is excited by an external power source or an integrated battery. Pas-

sive transponders are powered by RF fields which are emitted by interrogator anten-

nas. Compared with passive tags, active tags can achieve longer operational distance

due to the stable and sufficient power supply. However, the integration of battery and

periodic maintenance also add extra costs, increase tag size and lower the lifespan of

an active transponder [24]. As a result, active tags are impractical to be applied in item

level for retail business. They are currently deployed in specific applications and have

not been standardised [25].

Page 16


Table 2.1. Comparison among RFID systems.Frequency Band LF HF UHF Microwave

Operating frequency 125-135kHz 13.56MHz 860-960MHz 2.45, 5.8GHz

Communication Range 0.2-1m 0.1-0.7m 3-10m 3m

Data Rate Low High Medium Medium

Applied area Animal track Public transportation Airline baggage, Toll collection

Costs Low———————————————————————————————High

Sensitivity to water Low———————————————————————————————High

Tag size Large——————————————————————————————–Small

2.3.2 Operating Frequency

The operating frequency of RFID systems ranges widely from 125kHz to 5.8GHz, which

bands are all in ISM (Industrial, Science and Medical) band. Generally, RFID systems

can be divided into four types by the frequency of operation: Low Frequency (LF 125-

135kHz), High Frequency (HF 13.56MHZ), Ultra High Frequency (UHF 860-960MHz)

and so called microwave (2.45GHz, 5.8GHz). Further more, both LF and HF RFID tags

are powered by inductive (near field) coupling and the other two employ radiated (far

field) coupling. The criterion for distinguishing near and far field for a small dipole

is r = λ2π , where r is the distance between the field point and the source, and λ is the

wavelength in free space. More information about the coupling modes can be found

in [26].

Due to the varied coupling methods, RFID systems have their own advantages and dis-

advantages. For example, the systems which employ inductive coupling (LF & HF) are

not sensitive to wet environment, so these transponders are capable to label moisture

objects which radiated coupling tags cannot accomplish. Nevertheless, reading range

of LF and HF RFID systems are lower compared to UHF and Microwave systems. A

comparison among LF, HF, UHF and Microwave RFID systems is made in Table 2.1.

2.4 Regulations and Standards

The research work presented mainly focuses on UHF RFID systems, so only regula-

tions and standards related to UHF RFID are involved here.

Page 17


2.4.1 Regulations

More and more microwave devices, working at different frequency bands, are invented

and manufactured to assist people’s work and improve people’s life. However, the fre-

quency spectrum resource is limited. In order to avoid the mutual interference among

these devices and make the usage of the frequency spectrum efficient, regulations are

needed to be specified. Although the details of regulations vary among countries and

regions, in general, three main aspects are involved which are frequency spectrum us-

age, maximum radiated power allowance and frequency channel selection techniques.

There are two common regulations currently. They are specified by European coun-

tries (ETSI-EN 302 208) [27] and the United States of America (47 CFR Part 15.247) [28]

respectively. The two regulations are discussed in the following two itemisations.

• ETSI-EN 302 208

This regulation is drafted by European Telecommunications Standards Institute

(ETSI). A maximum 2W ERP power limitation is settled for RFID devices (the

term ERP is introduced in Subsection 3.2.5). The frequency band ranges from

865MHz to 868MHz. On the edge of this range, the emitted power is restricted

strictly, which goes down to 100mW-500mW. The power limitation along the fre-

quency range is shown in Figure 2.1. When it comes to the frequency selection

techniques, the mandatory “listen before talk (LBT)” mode in Version 1.1.1 [29].

should be abandoned on 31 December 2009. A new frequency hopping spread

spectrum (FHSS) techniques will be used in Version 1.2.1 [27]. In the new ver-

sion, the frequency spectrum from 865MHz to 868MHz is divided into 15 chan-

nels (1-15). The band of each channel is 200kHz and the transmission of RFID

interrogators is restricted in channel numbers 4, 7, 10, and 13 [27].

• Title 47 CFR Part 15.247

This protocol is specified by the USA Federal Communications Commission (FCC).

This regulation adopts frequency hopping spread spectrum (FHSS) in the range

from 902MHz to 928MHz. At least 50 hopping frequencies are required on condi-

tion that “the 20dB bandwidth of the hopping channel is less than 250kHz” [28].

The signal cannot occupy the same hopping channel more than 0.4 seconds in

Page 18


865.6MHz 867.6MHz 868MHz

100mW

2W

500mW

865MHz

Figure 2.1. Power limitation in EN 302 208.

20 seconds on average. Otherwise, if the 20dB bandwidth of hopping channel

is equal to or exceed 250kHz, at least 25 hopping channels are needed and the

average time of occupancy shall be less than 0.4 second in a 10 seconds period.

Moreover, the 20dB bandwidth of the hopping channel is limited under 500kHz.

The maximum allowed peak conducted output power is 4W EIRP (the term EIRP

is introduced in Subsection 3.2.5), if the directional gain of an antenna is less than

6dBi [28]. More details can be found in 15.247(c), if the gain of an antenna exceeds

6dBi.

The Australia Communications and Media Authority (ACMA) has also made some

regulations for UHF RFID devices [30]. A 6MHz band is assigned for UHF RFID equip-

ment, which is from 920MHz to 926MHz. The maximum emitted power is 4W EIRP

with a licence, which is in the charge of GS1 Australia, (GS1 Australia is a not-for-profit

organisation). In addition interrogators must comply with ISO/IEC 18000-6c which is

introduced in Subsection 2.4.2.

According to [31], several countries’ regulation status about UHF RFID systems are

listed in Table 2.2.

Page 19


Table 2.2. Regulation status of UHF RFID among countries.

Country Assigned spectrum Maximum transmitted power Technique

Australia 920-926MHz 4W EIRP (with licence) FHSS

United States 902-928MHz 4W EIRP FHSS

United Kingdom 865.6-867.6 MHz 2W ERP (with licence) FHSS

China 840.5-844.5MHz 2W ERP FHSS

Japan 952-956.4MHz 4W EIRP (with licence) LBT

2.4.2 Standards

The purpose of specifying standards is to improve the compatibility and to encourage

inter-operability of RFID devices manufactured by different companies. In this subsec-

tion, we are going to introduce the significant organisations in specifying standards for

RFID systems, the categories of these standards and two important protocols in UHF

band.

(a) Organisations for standardisation

Nowadays, many organisations join into the fierce competition of RFID standardisa-

tion. Three organisations with global influence are listed as follows:

• EPCglobal

It is a joint venture between European Article Numbering (EAN) International

and the Uniform Code Council (UCC) and funded by many transnational corpo-

rations. As mentioned in Section 2.2, the Electronic Product Code (EPC) concept

is advanced by this organisation and plenty of standards relating to UHF band

have been specified. An HF Generation 2 tag protocol standard is being devel-

oped.

• International Organisation for Standardisation (ISO)

ISO is the largest publisher of International Standards all over the world. It has

established branches in 159 countries. Unlike EPCglobal just focuses on the UHF

Page 20


band and starts to concern HF band, ISO specifies RFID standards covering from

LF band to microwave band.

• Ubiquitous ID (UID)

The Ubiquitous ID Center was founded by the T-Engine Forum. It is mainly

supported by Japanese companies.

(b) Categories of RFID standards

Specification of RFID standards is an enormous project which is composed of many

components. However, generally, it consists of the following four aspects.

• Technical protocol

This kind of protocol mainly specifies the air interface regulations including the

forward and return link parameters, maximum radiated power, operating fre-

quency spectrum, occupied bandwidth, modulation, anti-collision issue and bit

rate, etc. The typical technical protocols in the UHF band are ISO 18000-6 and

EPC Class1 Gen2 for which details are introduced in the Subsubsection (c).

• Data content protocol

This kind of protocol specifies data encoding and compaction rules. In detail, it

copes with the processing of date and provides guidelines of encoding data to

be objects in a particular application. For example, ISO/IEC 15961 and 15962 are

examples of this kind of standards.

• Conformance protocol

It defines RFID device performance test methods including product packing and

tagging etc. The common conformance protocols are ISO/IEC 18046 and 18047-6.

• Application standards

It specifies the RFID technology deployed in a particular application such as ISO

18185-2: Freight containers–Electronic seals, and ISO 14223-1: Radio Frequency

Identification of animals.

Page 21


(c) Important RFID technical protocols in UHF band

Only the technical protocols in the UHF band are involved here, since this work mainly

puts emphasis on UHF tags’ hardware design. As mentioned before, ISO and EPC-

global are the two biggest organisations in the RFID standardisation, so the technical

protocols developed by them are introduced here.

• EPCglobal Class1 Generation2

The Class1 Gen2 standard is the latest specification published by EPCglobal in

order to define the air interface of passive RFID systems in the spectrum from

860MHz to 960MHz. Pulse Interval Encoding (PIE) and Amplitude-shift keying

(ASK) are utilised in the forward link. Additionally, the backscattered link de-

ploys FM0 or Miller-modulated subcarrier. In terms of the anti-collision issue,

an interrogator commands tags in the readable area to load a Q-bit random (or

pseudo-random) number into their slot counters. The tags in their arbitrate state

decrement their slot counter every time when they receive a command from the

interrogator. The tags reply when the value in their slot counter is zero [32].

Hence, the tags reply in different time slot.

• ISO 18000-6

This standard can be classified into 3 types: Type A, Type B and Type C. Type

C is the same as the EPC Class1 Gen2 standard. Both Type A and Type B adopt

ASK to be the modulation mode and the spectrum from 860 to 960MHz to be the

operational band. The unique ID length is 64bits. “Both types use a common re-

turn link and are reader talks first. Type A uses Pulse Interval Encoding (PIE) in

the forward link, and an adaptive ALOHA collision arbitration algorithm. Type

B uses Manchester in the forward link and an adaptive binary tree collision arbi-

tration algorithm” [33].

Page 22


2.5 Conclusion

This chapter has introduced how RFID systems are developed, classified, regulated

and standardised and has put emphasis on an introduction to UHF RFID systems,

since UHF RFID systems are the major concern of the work in this thesis. The contents

of this chapter provide a clear view of RFID background and serve as complementary

materials for understanding the rest of the thesis.

Page 23

Chapter 3

Operating RangeEvaluation of RFID

Systems

THE operating range of an RFID system is evaluated by analysing

each system component, and some factors which may enable im-

provement in the operating range are identified. Previous meth-

ods for analysing the operating range are summarised and their limitations

are listed. For overcoming these limitations, a new method for performing

such analysis by means of a scattering matrix is proposed.

Page 25

3.1 Introduction

3.1 Introduction

Operating range is one of the most significant criteria in evaluating the performance of

RFID systems, especially UHF RFID systems. That is because a longer operating range

can create more potential application opportunities and ensure a more reliable perfor-

mance. The operating range is determined by the whole RFID system design rather

than just a part of it. Hence, it is worth doing some analysis to find out the methods

for evaluating the operating range and which factors, in an overall system design, may

play a key role in improving the operating range. In order to achieve this target, this

chapter provides in Section 3.2 an RFID technology background by explaining some

relevant terminologies in the antenna performance. The considerations of designing

tag antennas in reality are described in Section 3.3. Then Section 3.4 and Section 3.5

analyse two limitations 1) threshold power in exciting a transponder, and 2) sensitivity

of a reader, in achieving a successful communication between the transponder and the

reader. Section 3.6 summarises the existing work in the literature on analysing the op-

erating range of UHF RFID systems. The existing work is based on either theoretical

analysis according to the Friis equation or totally experimental analysis in real RFID

systems. The experimental analysis is a direct solution but may be expensive in cost

and time. The limitations of using the Friis equation are given in Section 3.7. In or-

der to overcome these limitations, Section 3.8 provides a novel method for evaluating

the operating range of RFID systems via a scattering matrix. Lastly the key factors in

designing a long operating range RFID systems are identified in Section 3.9.

3.2 Fundamental Parameters of Antennas and the Friis

Equation

3.2.1 Power Transmission in a Tag

Figure 3.2 shows a Thevenin equivalent circuit of an antenna in its receiving mode.

Zant = Rant + jXant is the input impedance of the antenna in which Rant is composed

of loss resistance Rl and radiation resistance Rr. The receiving antenna is connected to

Page 26

Chapter 3 Operating Range Evaluation of RFID Systems

A’

Rl

B’

I

A

BRr

Xant

Vin

Rc

Xc

Z0

Figure 3.1. Thevenin equivalent of a receiving antenna.

Rl

I

A

BRr

Xtant

Vin

Rchip

Xchip

Figure 3.2. Thevenin equivalent of a transponder.

its load Zc = Rc + jXc by a transmission line of which the characteristic impedance is

Z0. Vin is the induced voltage caused by the incident wave. In the diagram the induced

voltage is represented by a peak value phasor. The source causes a current represented

by a peak value phasor I to circulate in the direction shown through all elements of the

circuit.

If the receiving antenna shown in Figure 3.2 is a tag antenna, then the load presented

to the tag antenna is a chip. In fact, the transmission line between the tag antenna

and the chip is very short, hence the antenna output impedance at port AB is nearly

the same to the transferred impedance at port A′B′. Figure 3.1 is thus simplified to

Figure 3.2, in which the symbols representing the impedance elements are rewritten to

keep correspondence with the situation here. Zchip = Rchip + Xchip is the impedance of

the chip presented to the tag antenna. Ztant = Rtant + Xtant is the tag antenna’s output

impedance. Rtant is composed of loss resistance Rl and radiation resistance Rr.

The phasor representing the circulating current is given by

I =Vin

Rtant + Rchip + j(Xtant + Xchip)(3.1)

Page 27

3.2 Fundamental Parameters of Antennas and the Friis Equation

The power captured by the tag (chip and tag antenna) is expressed as follows

Ptag =|I|2(Rchip + Rtant)

2=

|Vin|2(Rchip + Rtant)2(Rtant + Rchip)2 + 2(Xtant + Xchip)2 (3.2)

The power delivered to the chip is given by

Pchipr =

|I|2Rchip

2=

|Vin|2Rchip

2[(Rtant + Rchip)2 + (Xtant + Xchip)2](3.3)

The load impedance for maximum power transfer is the complex conjugate of the an-

tenna impedance. Thus in this case

Rchip = Rtant (3.4)

Xchip = −Xtant (3.5)

The total power captured is then |Vin|24Rtant

according to (3.2). Half of the power is delivered

to the load which is the maximum available power PA = |Vin|28Rtant

. The other half is

consumed by the antenna in the form of scattered power and ohmic losses. If the

antenna is lossless which means that Rl = 0 so Rtant = Rr, the backscattered power is|Vin|28Rr

.

When the impedance is unmatched whether or not the antenna is a lossless antenna,

the chip can only get part of the maximum available power, the ratio of the power Pchipr

delivered to the unmatched load to the maximum available power PA is then

Pchipr

PA=

4RtantRchip

(Rtant + Rchip)2 + (Xtant + Xchip)2 (3.6)

We can use the identity |Zchip + Ztant|2 − |Zchip − Z∗tant|2 = 4RtantRchip to write the

result above as

Pchipr

PA=|Zchip + Ztant|2 − |Zchip − Z∗tant|2

|Zchip + Ztant|2 = 1− |Zchip − Z∗tant

Zchip + Ztant|2 = 1− |θ|2 (3.7)

where θ =Zchip−Z∗tantZchip+Ztant

is defined as the reflection coefficient in many publications [34]

[35] [36], but we notice that the expression of θ here is not analogous to a reflection coef-

ficient as defined in most text books because of the conjugate symbol in the numerator.

Hence, we would rather just call it the theta parameter. It has the property that its

Page 28


magnitude squared is the fraction of the available source power that is not delivered

to the chip.

Using the circuit of Figure 3.2 and our definition of the theta parameter, we may derive

the expression for the current I

I =Vin

2Rtant(1− θ) (3.8)

The sum of the powers dissipated within and backscattered from the tag antenna be-

comes

Ptagsum =

|I|2Rtant

2=|Vin|28Rtant

|1− θ|2 = PA|1− θ|2 (3.9)

The backscattered power into the air becomes

Ptagbs =

|I|2Rr

2=|Vin|28Rtant

Rr

Rtant|1− θ|2 = PAer|1− θ|2 (3.10)

where

PA = the maximum available power of load,

er = RrRtant

is known as the radiation efficiency.

3.2.2 Effective Area

The power capturing characteristics of a receiving antenna can also be described in

terms of effective area, which is defined as the ratio of the available power at the termi-

nals of the receiving antenna to the power flux density of a plane wave incident on the

antenna on condition that the polarisation of the receiving antenna and the impinging

wave is matched. In mathematical form, it is shown as (3.11) which can be found on

page 89 of [37].

Ae =Pr

Wi(3.11)

where

Ae = effective area (m2),

Pr = available source power (W),

Wi = power density of incident wave (W/m2).

Page 29


3.2.3 Effective Length

The induced voltage Vin of the receiving antenna shown in Figure 3.2 can also be ex-

pressed in terms of the antenna effective length.

In order to provide clarity in the definition of the concept of effective length, we intro-

duce as shown in Figure 3.3 the definitions of input current and induced voltage for a

general antenna.

z

Pq

A

B

r

aq

Iin

Vin

+

-

le

Figure 3.3. Coordinate used in the definition of effective length.

The general antenna we consider is excited, when it is driven, by injecting a peak value

phasor input current Iin at two terminals A,B, shown in Figure 3.3. These terminals

are also used as the output terminals for the induced voltage Vin sensed as shown in

Figure 3.3 when the antenna receives a signal from an incident field.

Without loss of generality, we place in Figure 3.3 terminal B directly above terminal

A, and establish a spherical polar co-ordinate system with its origin at the midpoint at

the interval AB, and the reference z axis for the polar angle θ in the direction from A to

B (other satisfactory co-ordinate systems could be defined, but the one being defined

here has the advantage of being defined in a clear way).

It is noted that the antenna need not be a wire antenna. All that is needed is that it have

terminals A,B allowing the definition of an input current and of a terminal voltage, and

a co-ordinate system for describing the far field.

Page 30


We note that the far electric field will be in the direction aθ and we will be defining an

effective length vector le to be in the same direction as aθ and to have a magnitude to

be defined shortly.

The value of the effective length vector le is determined by far-zone field Ea radiated

by this antenna which can be found on page 88 of [37].

Ea = −jηkIin

4πrlee−jkr (3.12)

where k is the free space propagation constant and η is the characteristic impedance in

free space.

For a uniform incident electric field represented by a peak value phasor Ei, using the

reciprocity theorem, the voltage Vin induced at the terminal of a receiving antenna

which is shown in Figure 3.2 depends on the same effective length of the antenna as

shown in equation (3.13) [37] below.

Vin = Ei · le (3.13)

3.2.4 Gain

Gain is one of the parameters that describe an antenna’s radiating ability. The absolute

gain of an antenna (in a given direction) is defined as the ratio of the power density of

an antenna radiated to a certain far field point to the power density at the same point

which would be radiated by a lossless isotropic emitter. It is expressed as

g =4πr2Wrad

4πr2W irad

=Wrad

W irad

(3.14)

where g is the gain of the subject antenna, r is the distance from the antenna to a point

in far-field zone, which should be lager than 2D2/λ, (D is the largest dimension of the

subject antenna). Wrad is the radiation density generated at that point by the subject

antenna [37], W irad is the power density of the lossless isotropic emitter.

The physical meaning of gain is related to the two factors: (1) directivity Dd and (2)

radiation efficiency er [37]. Gain can also be expressed in the other form (3.15) by

means of these two factors.

g = Dd × er (3.15)

Page 31


Besides the expression of gain in terms of absolute value as introduced above, another

two forms are also widely used. These are dBi and dBd. GdBi is the form of gain which

is written in decibels (dB). In mathematical forms, it is shown

GdBi = 10 log10Wrad

W irad

= 10 log10 g (3.16)

Clearly, the reference is still a lossless isotropic emitter. The concept of dBd is similar

to dBi. The only difference is that the reference object is changed to a lossless half

wavelength dipole antenna instead of a lossless isotropic antenna. Therefore, gain in

dBd can be expressed as (3.17).

GdBd = 10 log10Wrad

Wdrad

= 10 log10(Wrad

W irad× W i

rad

Wdrad

)

= 10(log10 g− log10 gd) (3.17)

where Wdrad is the radiation density of the half wavelength dipole antenna and gd is the

gain of the dipole which is 1.64. In terms of dBi, it is 2.15dB. Hence, (3.17) becomes

GdBd = GdBi − 2.15dB (3.18)

3.2.5 EIRP and ERP

In order to avoid the effects brought by RFID power transmitter to other radio wave

devices, many countries impose regulations on the power usage. More details can

be found in Subsection 2.4.1. The radiated power limitation is usually expressed in

terms of “EIRP” and “ERP”. EIRP and ERP are the acronyms of Equivalent Isotropic

Radiated Power and Effective Radiated Power respectively. The regulators do not care

about how much power actually is radiated from the reader antenna, although the

limitation is described in terms of power. What they really care about is the maximum

power density.

The radiation power density of a reader antenna at a distance r can be expressed as

(3.19).

Wrad =Prant

t greader

4πr2 (3.19)

Page 32


where Prantt is the input power to the reader antenna, and greader is the gain of this

antenna.

The radiation density caused at the same distance r by a lossless isotropic emitter with

input power PEIRP is given in (3.20).

Wrad =PEIRP

4πr2 (3.20)

The PEIRP that achieves for a lossless isotropic emitter at a given distance the same

radiation density as the antenna of gain greader and input power Prantt is given by

Prantt =

PEIRP

greader (3.21)

ERP is a similar concept to EIRP, however, the reference emitter is changed to a lossless

half wavelength dipole instead of a lossless isotropic emitter. The absolute gain of a

lossless half wavelength dipole is 1.64. Therefore,

Prantt =

1.64PERP

greader (3.22)

The EIRP and ERP has the following relationship, derived by (3.21) and (3.22).

PEIRP = 1.64PERP (3.23)

3.2.6 Polarisation

The electric field vector at a point may trace a curve as a function of time. The type of

the curve can be used to classify polarisation patterns. Generally, polarisation can be

classified into three types which are linear, circular or elliptical polarisation.

When the receiving and transmitting antennas are polarised in the same pattern, the

receiving antenna can capture the maximum power emitted from the transmitting one.

However, in general, the polarisations of these communicating antennas working in

the same system are different, which causes polarisation mismatch.

Polarisation efficiency is involved to evaluate this mismatch. This factor is defined as

the ratio of the actual power received by an antenna to the possible maximum received

Page 33


power which can be accomplished by optimising the matching condition between the

polarisation of incident wave and that of receiving antenna. In mathematics, it is ex-

pressed as (3.24) on page 77 in [37],

ep =|le ·Ei|2|le|2|Ei|2 (3.24)

where

le = vector effective length of the receiving antenna which has been introduced in Sub-

section 3.2.3,

Ei = incident electric field.

UHF RFID systems usually adopt linearly polarised antennas as tag antennas because

of their low cost and easy fabrication. However, most RFID systems are used to de-

tect mobile items, for example, in the RFID application of supply chains, the cargo on

which is mounted a tag will be transported along a supply chain. If the reader antenna

is linearly polarised, it is possible that the tag antenna and the reader antenna can be

aligned orthogonally to each other. When that happens, the reader will not be able to

read or program RFID tags. Hence, RFID reader antennas often adopt circular polari-

sation to ensure in most of the cases the system can perform correctly. As a result, the

polarisation efficiency between a reader antenna in circular polarisation and a tag an-

tenna in linear polarisation is 0.5 i.e. -3dB. If a reader antenna is elliptically polarised,

the polarisation mismatch between a linearly and an elliptically polarised antenna can

be obtained in [38].

3.2.7 The Friis Transmission Equation

After introducing the fundamental parameters for describing an antenna, the Friis

transmission equation commonly used in designing and analysing communication

systems is given in (3.25). This equation relates the power delivered to the load of

a receiving antenna Pr to the available power Pt from a transmitter which is placed at a

distance r > 2D2/λ in free space, where D is the largest dimension of either antenna.

Pr = Pt(1− |Γt|2)(1− |Γr|2)gtgr(λ

4πr)2ep (3.25)

Page 34


In (3.25), Γt, Γr are the reflection coefficients of the transmitting antenna and the receiv-

ing antenna respectively, gt and gr are the gain of the transmitting and the receiving

antenna respectively, as defined in Subsection 3.2.4, and ep denotes the polarisation

efficiency which is explained in Subsection 3.2.6.

If the two antenna’s impedances are perfectly matched to their source or load and their

polarisation is matched as well, an ideal form of (3.25) is expressed as follows.

Pr = Ptgtgr(λ

4πr)2 (3.26)

(3.25) is an idealised form of the Friis transmission equation. When this equation is

applied in analysing RFID systems, a few changes should be made according to the

special needs of RFID systems, which are identified in Section 3.7.

In addition, the factor ( λ4πr )

2 which is defined as the path gain describes the depen-

dence of the power received by the transponder on the wavelength and the distance r.

Normally, this factor is much less than 1, and we speak of there being a loss. However,

this path loss occurs in free space. Most of RFID systems are installed in a building

or even a room. Therefore, the path loss in a more complicated environment should

be considered before applying it to an RFID system. The evaluation of the in-building

path loss has been introduced in [26] and described in Section 3.7.

3.3 Tag Antenna Design

In Section 3.2, a few fundamental parameters such as gain, impedance match, polari-

sation etc, in designing antennas are discussed. Besides those parameters, some other

parameters e.g. the antenna size, cost and deployed environment should be considered

as well if the antenna is expected to be used in reality. Usually, the tag antenna design

is more limited by those parameters required by the reality than the reader antenna

design, hence only the tag antenna design is discussed in this section. The parameters

required by the reality are discussed respectively in the three following itemisations.

Some work in other chapters in this thesis is mentioned.

• Size

Page 35

3.4 Threshold Power of a Transponder

Generally for tag antennas the smaller the better. However, the small size will

also affect other factors, such as gain, impedance match and bandwidth. Most of

the commercial tags are less than 140mm×40mm. Analysis of the meander line

dipole antenna which is one of the antenna types to realise the minimisation of

the tag antenna is provided in Chapter 4.

• Applied environment or attached objects

Definitely, an RFID system will not be deployed in free space. The applied envi-

ronment especially when a tag is attached to a metallic object will have a critical

impact on the performance of the RFID system. The reason why the metal inter-

feres in antenna performance is introduced in Section 6.2. As a result, a solution

to this problem is needed before completing an antenna design. To address the

antenna on metal problem, some existing solutions are introduced in Chapter 6,

and a new slitted decoupler is also analysed in Chapter 7.

• Cost

Generally speaking, a 96-bit EPC inlay (chip and antenna mounted on a sub-

strate) costs from 7 to 15 U.S. cents [39]. Low cost tags are always required by

the industry for a wide range of applications. One of the possible solutions to

reduce the cost significantly is the use of printed electronics, especially printed

silicon electronics, which is out of the scope of the work in this thesis. More de-

tails of the printed electronics and its costs can be found in [40], co-authored by

the author of the thesis.

Unfortunately and not surprisingly, the factors discussed in this section and the an-

tenna parameters discussed in Section 3.2 are interacting and usually are not positively

related. Some tradeoffs, depending on the system requirements, should be made dur-

ing the antenna design.


Chips require a minimum power or voltage to be operated which are called threshold

power or threshold voltage. Generally, the threshold power is about 100µW [19] but

Page 36


can be even less down to 16.7µW [34]. If the distance between a tag and a reader is

too far for the tag to collect more power than the threshold, that tag is unable to be

detected. The amount of power or voltage, which can be collected by transponders at

a certain distance, depends on the tag antenna design which is briefly discussed in Sec-

tion 3.3. Apparently, this threshold is critical to evaluate the reading range of an RFID

system and it is definitely decided by the chip IC design. As shown in Figure 3.4, a typi-

cal transponder IC consists of several principal components which are decoder, voltage

multiplier, modulator, control logic and memory unit. Each component’s power con-

sumption or power transfer efficiency can influence the threshold power. These factors

are discussed in the subsections below.

Decoder

Modulator

VoltageMultiplier

MemoryUnits

Logic

VDD

data

data

data

Antenna

Front end

clk

Figure 3.4. Block chart of a transponder.

3.4.1 Modulator

The power transfer efficiency influenced by impedance matching situation has been

analysed in Subsection 3.2.1. If the ideal impedance match is obtained which means

the chip input impedance is the complex conjugate of the antenna impedance, half of

the captured power is delivered to the chip, the other half is consumed by the antenna

linked to the chip. However, in this case, the signals carrying backscattered power are

all in the same phase and magnitude, and cannot carry any information. Therefore,

a modulator is employed in the chip circuit to adjust the front-end impedance into

two different states, Zchip1 and Zchip2. Hence, phase or magnitude of the backscattered

wave can be changed to form a useful signal back to the base station antenna. The

Page 37


input RF power to the chip becomes (3.27).

Pchipr,1,2 = PA(1− |θ1,2|2) (3.27)

where

θ1,2 =Zchip1,2−Z∗tantZchip1,2+Ztant

,

PA=maximum available power.

The power reflected from the chip for backscattering also then varies between two

states according to (3.10). In terms of the modulation modes, ASK (Amplitude Shift

Keying) or PSK (Phase Shift Keying) could be employed. For ASK, the amplitude

difference of the backscattered wave between the two states brought by θ1, θ2 should

be large enough to allow the reader to tell them apart. Similarly, for PSK, the phase

difference of the backscattered wave between the two states brought by θ1, θ2 should

be large enough to allow the reader to tell them apart. The difference of the two states

determines the error probability.

As a result, the θ parameter is a decisive factor in designing an RFID system. It deter-

mines through (3.27) how much RF power is distributed to the chip rectifier to be con-

verted into dc power and through (3.10) how much RF power is assigned to backscatter

to the reader for it to decode under a particular modulation mode either ASK or PSK.

The optimisation of the two states of θ depending on the modulation modes to achieve

the best usage of the RF power received by the transponder is discussed by Karthaus

et al. [34]. The selection of the two states of θ under either ASK mode or PSK mode

for obtaining reading range oriented RFID system or bit-rate oriented RFID system is

reported by Vita et al. [41]. The task of optimising the factor of θ is out of the scope of

the work in this thesis. Hence, it is not discussed further.

3.4.2 Rectifier Efficiency

Once the RF power is received, it will be transmitted to the inside circuit, including

voltage multiplier, decoder, control logic and memory units. However, the RF power

cannot be used by these components directly and the induced voltage in the terminal

Page 38


of the tag antenna is too small to excite the circuit. As a result, a voltage multiplier

is needed to rectify the ac current to dc, and to enlarge the induced ac voltage. This

process definitely brings power loss due to the diode and capacitor composing of mul-

tiplier. The ratio of dc power produced by the voltage multiplier to the input RF power

is called rectifier efficiency. Clearly, threshold power will be increased by a low rectifier

efficiency. It was reported that rectifier efficiency ranged from 5-25% in 2003 [19]. For

example, Karthaus and Fischer [34] achieved a 18% rectifier efficiency. However, with

the recent years development of semiconductor technology and circuit design, rectifier

efficiency has been improved significantly. Nakamoto et al. [42] even made the factor

to be 36.6%.

3.4.3 Memory Chosen

The threshold power, can be divided into two types: 1) the threshold power for reading

and 2) the threshold power for programming. Those two types of threshold power are

also related to the memory which is used to store data in the transponder. The data

carriers, currently applied, can be categorised into the three types of RAM, EEPROM

as well as FeRAM. A comparison among these memories is made below:

• RAM

This kind of memory can store data only temporarily. When the voltage supply

disappears, the stored data is lost. This form of memory can be used in a passive

tag as a temporary information storage when the tag is being read or written.

Additionally, it can also be applied in an active tag.

• EEPROM

Compared to RAM, EEPROM is a long-term storage memory which can provide

reliable data for around ten years [19]. The reading operation with this memory

needs a relative low supply voltage which is usually below 5V [19], [34]. Che et

al. even succeed in lowering the threshold voltage to be 0.75V [43]. Moreover,

a considerably large voltage (around 17V) is needed to activate the tunnel effect,

so that data can be written. Although a charging pump is integrated into the

Page 39


circuit to provide this large voltage and EEPROM is used widely as an RFID tag

memory, it still has two serious weakness. Firstly, the power consumption of

programming is much lager than that of reading due to the large voltage needed

in writing. As a result, the tag integrated with EEPROM cannot be read and

written at the same range. Usually, the writing range is only about 20% of the

reading range. Secondly, the programming is a time-expensive operation due to

the tunneling principle [42]. In general, it needs 5-10ms for each single-bit or

multiple-bit operation.

• FeRAM

FeRAM is invented to solve the weaknesses which are faced by EEPROM. The

ferroelctric effect is taken advantage of to store data and achieve a balanced

power consumption in both reading and programming. In particular, Nakamoto

et al. [42] addressed this unbalanced reading and writing barriers by employing

FeRAM memory. The writing time is also improved to 0.1µs [19] [42]. A 4m op-

erating distance approximately balanced in reading and writing was derived for

a 4W EIRP transmitted power. The actual input power of both working modes is

nearly the same which are 13µW in reading and 15.7µW in writing. The writing

speed of FeRAM is 100 times faster than that of EEPROM. However, FeRAM has

not been widely used in place of EEPROM because FeRAM cells are difficult to

combine with CMOS processes, since a high temperature treatment is needed to

crystallise the memory materials (PZT or SBT) into ferroelectric phases before the

cell is connected to the CMOS [19] [44].

Table 3.1 provides a comparison among the three memories [19] [45].

In conclusion, as long as the modulation mode, the rectifier efficiency, the dc power

needed by the chip circuit and the type of memory units are known, the threshold

power of transponder can be derived. In particular, Karthaus and Fischer [34] made a

tag which could be read at a distance of 4.5m under only 500mW ERP radiated power.

In this case with on-wafer measurements, the rectifier efficiency was established to

be 18%, the dc power consumed by the chip circuit was 2.25µW (1.5µA, 1.5V). As a

result, the minimum input RF power for operation is 12.5µW (2.25µW18% ). The threshold

Page 40


Table 3.1. Comparison among RAM, EEPROM and FeRAM.

Comparison parameters RAM EEPROM FeRAM

Size of memory cell ∼ ∼130(µm)2 ∼80(µm)2

Lifetime in write cycles ∞ 105 1010 ∼ 1012

Read cycle (ns) 12 ∼ 70 200 110

Write cycle 12∼70ns 3∼10ms 0.1µs

Data write Overwrite Erase + Write Overwrite

Write voltage (V) 3.3 15 ∼ 20 2 ∼ 3.3

Energy for Writing ∼ 100µJ 0.0001µJ

RF power for reading is the sum of the minimum backscattered power (4.2µW) derived

in [34] and the minimum input RF power (12.5µW). However, the threshold power

for programming is much larger than that for reading because an EEPROM memory

is chosen which choice leads to an unbalanced operating range between reading and

programming. The optimisation of all factors discussed in this section is beyond our

work, so they will not be discussed further in this thesis.

3.5 The Reader Sensitivity

The mathematical expression of a general receiver’s sensitivity is found in [46], and is

reproduced as follows.

Sen = (S/N)minkTB(NF) (3.28)

where

Sen = sensitivity,

(S/N)min = the minimum signal to noise ratio required to demodulate the replying

signal,

k = Boltzman’s constant,

B = bandwidth of the receiver,

NF = noise factor of the receiving equipment,

Page 41

3.6 The Literature Review on the Existing Work in Evaluating Operating Range

T = absolute reference temperature used in the definition of the noise factor.

In the case of an RFID reader the sensitivity can be influenced by several additional fac-

tors including receiver implementation details, receiver gain, communication protocol

specifics and interference generated both within the reader and externally by other

users of the spectrum. A figure for sensitivity is usually available from the reader

manual, and is commonly -70dBm. However, for passive tags the sensitivity is usually

good enough for detecting the backscattered signal [47], and the range is limited by tag

excitation, not receiver sensitivity.

3.6 The Literature Review on the Existing Work in Eval-

uating Operating Range

Significant work has been done in evaluating operating range of RFID systems recent

years.

Griffin et al. [48] reported two radio link budgets based on the Friis equation. The first

budget links the power received by a chip to the power radiated from a reader antenna.

The second budget establishes the relationship between the power received by the

reader from the backscattered power of the tag and the power radiated from the reader

antenna. The contribution of Griffin et al. [48] is to add a new factor named as gain

penalty in the modified Friis transmission equation. The gain penalty shows to what

extent the materials close to the tag can reduce the antenna’s gain. However, Griffin et

al. [48] assumes the tag antenna’s impedance is always matched to the chip. This is not

an accurate assumption because 1) the requirement of the modulation needs at least

one state of impedance mismatching, 2) the existence of electro-magnetically sensitive

materials in close proximity to the tag will critically vary the output impedance of the

tag antenna [49] [50].

Nikitin and Rao [51] introduced a new method in describing and measuring the backscat-

tered power from the tag antenna by means of radar cross section (RCS) based on

the Friis transmission equation in free space. Compared with the study by Griffin et

al. [48], the impedance mismatch occurring in the tag and caused by the modulation is

Page 42


considered. The RCS of a meander line dipole antenna in three different situations is

investigated by assuming the antenna is placed in free space. The three situations are 1)

the antenna is loaded with a chip, 2) the antenna is shorted and 3) the antenna is open

circuit. The measurement of the RCS was thus conducted in an anechoic chamber after

background substraction. However, when the tag is deployed in a more complicated

environment than in free space, this method is not applicable.

Jiang et al. [52] proposed another concept response rate in evaluating the operating

range of an RFID system by experiments. Most of the exciting readers support a “poll”

mode, wherein the reader continually scans for the presence of RFID tags. For exam-

ple, a reader sends N polls within a second, and counts the number of the responses

(Nr) from the particularly tag being observed. Therefore, the response rate from that

tag is defined as α = Nr/N. The larger the response rate is, the more probability the

tag will be read. By placing the tag in different positions each time in a complex envi-

ronment, and counting the response rate of the tag, the readable probability of the tag

in various positions can be derived. The optimum position could be found and this

optimisation definitely involves the influence of the environment. In addition, people

can even place many tags in the complex environment at one time and get the response

rate of each tag by experiments. The method not only considers the effects from the

environment but also the effects from the mutual coupling among the tags.

Hodges et al. [53] optimised the position where the tag should be attached on each bot-

tle of wine within a case containing six identical bottles based on a modified response

rate test. The test is modified by setting a threshold response rate and attenuating the

transmitting power from the reader programmablly to meet that threshold response

rate. Then the RF margin for the tag in each location on the wine bottle is tested and

the optimum location is determined.

According to the discussion above, the existing work is based on either theoretical

analysis according to the Friis equation or totally experimental analysis in a real RFID

system. The experimental analysis is a direct solution but may be expensive in cost

or in time. In addition, the limitation of using the Friis equation is also obvious in

Page 43

3.7 Interpretation and limitations of the Friis Transmission Equation in an RFIDPerspective

that it cannot deal with a complex environment. More details of the Friis equation’s

limitation in evaluating the operating range of an RFID system are given in Section 3.7.

Furthermore, the simulation tools such as Ansoft HFSS or CST can accomplish a full

wave analysis on the transmission between two antennas or among multiple anten-

nas. A complex environment can be built in the simulation model and considered in

the simulation process. The accomplishment of the simulation is definitely dependent

on the computing ability of the equipment used. The influences of the environment

on the antenna gains and input impedance can be obtained directly, hence people may

argue that the Friis equation could still be used combining with the simulation results

about the antenna impedance and gain which is similar to what Griffin et al. [48] did

by involving a gain penalty, but the path loss caused in the propagation cannot be ob-

tained directly which is required by the Friis transmission equation. Hence, we totally

abandon the Friis equation but turn to evaluating the reading range of an RFID system

in any environment by a scattering matrix which takes all the relevant matters into ac-

count. More importantly, a scattering matrix can be obtained by both simulation and

experiments. This novel method in evaluating the operating range of an RFID system

is introduced in Section 3.8.

3.7 Interpretation and limitations of the Friis Transmis-

sion Equation in an RFID Perspective

In Subsection 3.2.7, a common form of the Friis transmission equation is given in (3.25).

In addition, (3.25) is simplified to (3.26) in an ideal condition. In this section, the phys-

ical meaning of each factor in the Friss transmission equation and its usage is inter-

preted in an RFID perspective. With respect to the radio wave communication be-

tween a reader and a passive tag, it is known that the reader firstly interrogates the

tag, which is named as forward-link. Then, the tag receives the power from the inter-

rogating wave and makes use of this power to backscatter a signal to the reader, which

process is named as backward-link. The Friis transmission equation may be used once

in each link. We therefore discuss the use of the Friis transmission equation in the two

Page 44


links respectively and identify its limitations in analysing operating range of an RFID

system.

3.7.1 Forward Link

In the forward-link, the reader antenna is in the transmitting mode. Conversely, the

tag antenna is in the receiving mode. The Friis transmission equation used in this link

is written as follows according to (3.25).

Pchipr = Preader

t (1− |Γrant|2)(1− |θ|2)greadergtag 1pl

ep (3.29)

Preadert represents the available source power from the reader generator, which has been

designed to produce optimum power into a load of real impedance Z0 and has been

connected to the reader antenna by a cable of characteristic impedance Z0. Pchipr is the

power received by the chip. Γrant is the reflection coefficient between the reader an-

tenna and the reader which is expressed in (3.30a). Zrant is the input impedance of the

reader antenna, Z0 is the characteristic impedance of the transmission line connected

to the reader antenna, which is usually 50Ω. θ is the parameter the magnitude squared

of which describes the fraction of the available source power not delivered to the tag

circuit as defined in Subsection 3.2.1 and rewritten in (3.30b) in which Zchip is the chip

impedance, Ztant is the output impedance of the tag antenna and Z∗tant is conjugate to

Ztant. greader and gtag are the gains of the reader antenna and the tag antenna respec-

tively. The path gain factor ( λ4πR )2 in (3.25) is changed to be 1

pl , since the RFID system

considered here is not assumed to be operated in free space but a more practical and

complex environment.

Γrant =Zrant − Z0

Zrant + Z0(3.30a)

θ =Zchip − Z∗tant

Zchip + Ztant(3.30b)

The expression of the power input into the reader antenna is given in (3.31) according

to (3.21).

Prantt = Preader

t (1− |Γrant|2) =PEIRP

greader (3.31)

Page 45


where PEIRP is the equivalent isotropic radiated power which meaning is given in Sub-

section 3.2.5. The involvement of PEIRP is because the maximum power allowed to be

radiated is usually described in terms of PEIRP. According to (3.31), (3.29) becomes:

Pchipr = PEIRP(1− |θ|2)gtag 1

plep (3.32)

The maximum value of Pchipr is obtained when PEIRP is set to be maximum which is

regulated differently in different countries and regions. To make the tag readable, Pchipr

has to be larger than the threshold power for operating the chip, which was discussed

in Section 3.4.

In (3.7), another form of Pchipr is given in terms of maximum available power PA and

the theta parameter θ, which is rewritten as follows.

Pchipr = PA(1− |θ|2) (3.33)

3.7.2 Backward Link

In the backward-link, the tag antenna is in the transmitting mode. Conversely, the

reader antenna is in the receiving mode. The Friis transmission equation used in this

link is written as follows.

Preaderr = Ptag

sum(1− |Γrant|2)greadergtag 1pl

ep (3.34)

where Preaderr is the power received by the reader and Ptag

sum is the sum of the powers

dissipated within and backscattered from the tag antenna. The expression of Ptagsum has

been given in (3.9) which is rewritten in (3.35). The path loss factor remains the same

as that in (3.32), since the propagating path in the forward link is the same as in the

backward link.

Ptagsum = PA|1− θ|2 (3.35)

Solving for Pchipr according to (3.35) and (3.33) gives:

Pchipr =

1− |θ|2|1− θ|2 Ptag

sum (3.36)

Page 46


Substituting (3.36) into (3.29), another expression of Ptagsum is derived.

Ptagsum = Preader

t (1− |Γrant|2)|1− θ|2greadergtag 1pl

ep (3.37)

Substituting (3.37) into (3.34), then

Preaderr = Preader

t [(1− |Γrant|2)|1− θ|greadergtag 1pl

ep]2 (3.38)

(3.38) establishes the relationship between the power transmitted from the reader Preadert

and the power received by the reader Preaderr after the transmitted wave is backscattered

from the tag antenna. Preaderr has to be larger than the sensitivity of the reader which

was introduced in Section 3.5.

According to (3.31), Preadert is replaced by PEIRP/[(1− |Γrant|2)greader], (3.38) becomes:

Preaderr = PEIRP(1− |Γrant|2)greader[|1− θ|gtag 1

plep]2 (3.39)

3.7.3 Limitations in Implementing the Friis Transmission Equation

In Subsections 3.7.1 and 3.7.2, the power transfer between the transponder and the

reader in the forward and backward link is established in (3.29) and (3.38) by means of

the Friis transmission equation.

However, there are a few limitations in implementing the Friis transmission equations

for evaluating the operating range of an RFID system, if the system is deployed in a

very complex environment, e.g. 1) when a tag is mounted on a metallic item or a liquid

item, or 2) when the testing environment contains a lot of metal reflectors. The reasons

of the limitations are given as follows.

1. Far field condition

To implement the Friis transmission equation, the two antennas in communica-

tion should be sufficiently far away from each other. The distance between them

should be larger than 2D2/λ, where D is the largest dimension of either antenna,

and λ is the free space wavelength at the resonant frequency. However, when

an RFID system is placed in the very complex environment as mentioned before,

Page 47


the reader antenna has to be very close to the tag in order to read it. Hence, the

distance between them is not sufficient to meet the far field criterion.

2. Gain and impedance variation

In the Friis transmission equation, the gain and input/output impedance of the

tag/reader antenna are involved. However, again the RFID system is placed in

a very complex environment. The gain pattern and impedance will vary from

the intentionally designed values. The effects brought by metals in proximity to

a tag antenna to the antenna’s output impedance and gain are discussed in [48]

and [49]. It would be possible to investigate those effects by means of simulation

or experiments, but that would require effort.

3. Unknown path loss factor

As shown in (3.31) and (3.38), path loss factor 1pl is still unknown. If the RFID

system is deployed in free space, 1pl is equal to ( λ

4πr )2, where r is the distance

between the two communicating antennas. Most RFID systems are not deployed

in free space but in an in-building environment consisting of many obstacles in

the signal propagating path, and the system may be composed of multiple read-

ers and tags. Because of the obstacles in building-environment where an RFID

system is deployed, there are more losses brought by path obstruction, reflection,

multi-path propagation, absorption and other attenuation effects. In addition,

there are also more losses brought by the interaction between the multiple read-

ers and tags.

The analysis of path loss of a dense reader environment can be found in [26].

The path loss in dB of a two-antenna RFID system (one tag antenna, one reader

antenna) in building is introduced [54]:

PL(dB) = PL(d0) + 10× n× log10(dd0

) (3.40)

where d0 is an arbitrary reference distance; n is a value that depends on the sur-

roundings and building type; d is the distance between the reader antenna and

the tag antenna. The reference distance d0 should be selected to be much smaller

than the size of the building, so that the reflection in this small distance is not

Page 48


significant and the path loss in this small distance d0 can be considered approxi-

mately equal to the path loss in free space.

Path loss represented by (3.40) is a rough evaluation of the general case of an

RFID system in building. It does not have the universality of all situations and

especially is not suitable for defining the path loss factor in complex environ-

ments, e.g. metallic items in near proximity to a tag.

Based on the limitations in implementing the Friis transmission equation in evaluating

the operating range of an RFID system, a novel method by means of the scattering

matrix is therefore proposed in Section 3.8.

3.8 The Use of S-parameters in Analysing the Operating

Range of RFID Systems

3.8.1 Formula Derivation

We consider the two antennas (a reader antenna and a tag antenna) transmission sys-

tem to be a two port system, as shown in Figure 3.5, in which the reader and chip are

connected to the reader antenna and the tag antenna by transmission lines of which

the characteristic impedance is Z0. In Figure 3.5, the reader antenna is represented by

the two red lines for which the input impedance, taking into account the coupling be-

tween the antennas, is Zrant, and the tag antenna is represented by the two blue lines

for which the output impedance, taking into account the coupling between the anten-

nas, is Ztant. The resistance of the reader Rreader is deliberately designed to be equal to

Z0 (50Ω). In addition, the transmission line between the tag and the chip is very short.

In the following discussion, we will make use of scattering parameters to establish the

relationship between the power received by the chip and the power transmitted from

the reader antenna. All the values involving voltage and current are represented by

peak value phasors.

On the right side of Figure 3.5, the voltage V0 and current I0 at the load port are ex-

pressed in (3.41).

Page 49

3.8 The Use of S-parameters in Analysing the Operating Range of RFID Systems

Vin

Rchip

Xchip

Rreader

Port 2

V1

+

V1

-

V2

+

V2

-

Z0 Z0

Port 1

Zrant Ztant

I1

V1

I2

V2

V0

+

V0

-

Port 0

Figure 3.5. Two port junction representing coupled antennas in an RFID system.

V0 = V+0 + V−0 (3.41a)

I0 = I+0 + I−0 (3.41b)

The current I+0 and I−0 can also be expressed by the voltage in and out of the load port

as shown in (3.42).

I+0 =V+

0Z0

(3.42a)

I−0 = −V−0Z0

(3.42b)

The ratio of V−0 /V+0 is equal to the reflection coefficient looking into the chip impedance

from the terminal of the transmission line, which is written as follows.

V−0V+

0= sL =

Zchip − Z0

Zchip + Z0(3.43)

The power received by the chip Pchipr is obtained by (3.44).

Pchipr =

|I0|22

Rchip =12| V0

Zchip|2Rchip =

|V+0 + V−0 |2Rchip

2|Zchip|2=|V+

0 |2|1 + sL|2Rchip

2|Zchip|2(3.44)

As mentioned before, the transmission line between the chip and the tag antenna is

very short (its length is nearly zero), hence, V+0 = V−2 and V−0 = V+

2 . Then (3.43) be-

comes (3.45). In addition, replacing V+0 in (3.44) by V−2 , (3.46) is derived.

Page 50


V+2

V−2= sL =

Zchip − Z0

Zchip + Z0(3.45)

Pchipr =

|V−2 |2|1 + sL|2Rchip

2|Zchip|2(3.46)

Similarly, on the left side of Figure 3.5, the voltage V1 and current I1 on the port one is

expressed in (3.47).

V1 = V+1 + V−1 (3.47a)

I1 = I+1 + I−1 (3.47b)

The current I+1 and I−1 can also be expressed by the voltage in and out of the port one

as shown in (3.48).

I+1 =V+

1Z0

(3.48a)

I−1 = −V−1Z0

(3.48b)

The ratio of V−1 /V+1 is equal to the reflection coefficient Γrant which is expressed in

Section 3.7 and rewritten as follows.

V−1V+

1= Γrant =

Zrant − Z0

Zrant + Z0(3.49)

The power transmitted from the reader antenna Prantt is obtained by (3.50).

Prantt =

12

Re(V1 · I∗1) =12

Re[1

Z0(V+

1 + V−1 )(V+1 −V−1 )∗]

=12

Re[1

Z0|V+

1 |2(1 + Γrant)(1− Γrant)∗] =|V+

1 |22Z0

(1− |Γrant|2) (3.50)

A scattering matrix can be built according to the simplified two port system shown in

Figure 3.5 as below. V−1

V−2

=

s11 s12

s21 s22

V+

1

V+2

(3.51)

Page 51


According to the above matrix, the V−1 and V−2 can be written into (3.52).

V−1 = s11V+1 + s12V+

2 (3.52a)

V−2 = s21V+1 + s22V+

2 (3.52b)

Substituting the first of (3.45) and (3.49) into (3.52), solving for V−1 /V+1 and V−2 /V+

1

givesV−1V+

1= Γrant = s11 − s12s21sL

s22sL − 1(3.53)

V−2V+

1=

s21

1− s22sL(3.54)

Hence,

V−2 = V+1

s21

1− s22sL(3.55)

(3.53) illustrates how the impedance mismatch in the transponder and the testing en-

vironment considered in the S parameters affect the reflection occurring between the

reader and the reader antenna.

Inserting (3.55) into (3.46):

Pchipr =

|V+1 |2|s21|2|1 + sL|2Rchip

2|1− s22sL|2|Zchip|2(3.56)

(3.56) demonstrates that the power received by the chip is partially related to |V+1 |2.

The value of |V+1 |2 can be defined by the combination of (3.31) and (3.50) as follows:

Prantt =

|V+1 |2

2Z0(1− |Γrant|2) =

PEIRP

greader (3.57)

Then (3.56) becomes:

Pchipr =

PEIRP

greader |s21|2RchipZ0

|Zchip|2|1 + sL|2

(1− |Γrant|2)|1− s22sL|2 (3.58)

In (3.57), |V+1 |2

2Z0represents the available source power from the reader generator. The

product of this power and (1− |Γrant|2) denotes the power radiated from the reader

antenna. This radiated antenna power can be expressed in terms of PEIRP by multi-

plying by the gain of the reader antenna greader. If PEIRP is set to be the maximum

Page 52


power specified by regulations, then (3.57) tells us that no matter what the reflection

between the reader antenna and the transmission line is, the reader antenna can always

be made to radiate the same amount of power PEIRPgreader by adjusting the available source

power |V+1 |2

2Z0.

Finally, the power received by the chip is represented by means of scattering param-

eters which can be obtained by the simulation tools or experiments. The complex en-

vironment in which the RFID system is deployed can be built in the simulation model

and considered in the simulation process. In terms of experiments, the environment is

certainly considered. PEIRP is specified by the regulations in different countries and re-

gions separately. In Australia, this factor is equal to 4Wor 36dBm as introduced before.

greader is dependent on the reader antenna deployed. sL can be calculated by (3.43). The

reflection occurring between the reader and the reader antenna represented by Γrant is

caused by the testing environment represented by S parameters and sL as shown in

(3.53). As a result, Pchipr can be obtained. When Pchip

r is less than the threshold power

of the chip which is in the order of -10dBm, the reading fails and the maximum reading

range can be read in the simulation model or measured directly in experiments. Here,

the backward link is not considered since it is concluded in [47] that the limitation of

the reading range of a passive RFID systems mainly comes from the forward link not

the backward link because usually the reader’s sensitivity is, as mentioned before, low

enough to detect the signal from the successfully excited tag.

3.8.2 Formula Validation

In the last subsection, (3.58) has been derived to calculate the power received by the

chip. In this subsection, it is verified by simulation and experiments. However, as

mentioned before, to implement (3.58), the available source power of the reader gener-

ator should be adjusted according to Γrant to keep the radiation power from the reader

antenna as PEIRP/greader. The implemented condition brings some obstacles in the ex-

perimental validation, since the available source power of most real reader generators

cannot be adjusted arbitrarily. The available source power can only be set stage by

stage and the gap between the adjacent stages is large (in our case the gap is 0.1W).

Page 53


The approach we adopted to solve that problem is to keep the available source power

unchanged as PEIRP/greader which is very easy to achieve by the real reader and leads

to:|V+

1 |22Z0

=PEIRP

greader (3.59)

After substituting (3.59) into (3.56):

Pchipr =

PEIRP

greader |s21|2RchipZ0

|Zchip|2|1 + sL|2|1− s22sL|2 (3.60)

Equation (3.60) is more convenient to be used in the form of dB, which is shown in

(3.61).

Pchipr (dBm) = PEIRP(dBm)− Greader(dBi) + |S21|(dB)

+10 log10RchipZ0

|Zchip|2+ 20 log10 |

1 + sL

1− s22sL| (3.61)

In the following discussion, (3.58) is verified indirectly by verifying (3.61) by simula-

tion and experiments. The experiments were conducted by testing the reading range

of a self-made tag. The equipment used in the experiments is introduced first.

• Self-made tag

The self-made tag shown in Figure 3.6 is used. The chip is manufactured by Alien

Technology which model is Higgs-2. The chip conforms to the EPCglobal Class

1 Gen 2 specifications. It is implemented in a CMOS process and uses EEPROM

memory. The equivalent input impedance of the chip in parallel is shown in Fig-

ure 3.7(a) in which the parallel resistance Rp is 1500Ω and the parallel capacitance

Cp is 1.2pF. Usually, the input impedance of a tag antenna is presented in series.

Hence, in order to simplify the analysis, the chip impedance is transformed into

a series representation, so Figure 3.7(a) becomes Figure 3.7(b). At 923MHz which

is the centre frequency of UHF RFID band in Australia, the input impedance in

series is about 13.6-j142Ω. Hence, Zchip in (3.61) can be obtained. Typically the

threshold power of this chip is -14dBm, but the threshold power is dependent on

the manufacturing quality control, the worst could be -11dBm. More details of

the chip can be found in [55].

Page 54


The tag antenna is a meander line dipole antenna fabricated on FR4 board which

thickness is 1.6mm and the dielectric constant is 4.4. The footprint of this antenna

is 43.8mm×28.8mm. The output impedance of this antenna is designed to be

approximately conjugate matched to the chip impedance. More details of this

antenna can be found in Chapter 4.

The chip is installed on the antenna by electrically conductive adhesive transfer

tape manufactured by 3M (Model 9703) [56]. Currents can pass perpendicularly

through the sticky tape. The tape and adhesive material on it will bring losses and

chip impedance changes. However, previous experiences with this tape reported

by other colleagues in our laboratory indicated that these losses and impedance

changes are negligible [8].

Figure 3.6. A self-made tag used in experiment.

A

B

Cp Rp

(a) In parallel

A

B

C s

R s

(b) In series

Figure 3.7. The chip impedance illustration.

• Reader

The RFID reader used in the experiment is manufactured by FEIG Electronics

which model is ID ISC.LRU2000. The reader antenna is the linearly polarised

patch antenna with 8dBi gain and manufactured by Cushcraft Corporation, model:

Page 55


S9028P. The reason why the linearly polarised reader antenna is used is to sim-

plify the model building in the simulation discussed later.

• Shielding tunnel

The reading range experiments were conducted by placing both the self-made

tag and the reader antenna inside of a shielding tunnel. The size of the tunnel is

1826mm×915mm×690mm, which is shown in Figure 3.8. The shielding tunnel

is surrounded by electromagnetic wave absorbing foam. The absorbing foam

is manufactured by the Emerson & Cuming company for the frequency range

from 600MHz to 4GHz. These absorbing foams can achieve maximum -22dB

reflectivity around 1GHz. The inside space of the tunnel can thus be considered

to be effectively free space.

690mm

915mm

Figure 3.8. A shielding tunnel. The size of this tunnel inside is 1826mm×915mm×690mm.

As mentioned before, the reading range of the self-made tag were measured by plac-

ing the tag and the reader antenna in the shielding tunnel. Since the tunnel inside can

be regarded as free space, it is not the complex environment as described in Subsec-

tion 3.7.3. In order to make the environment complex, a square aluminium plate which

length is 260mm is placed behind the tag. Various reading ranges of this tag were tested

by varying the distance between the tag and the plate. The reading ranges are shown

in Table 3.2. In Table 3.2, dt is the distance between the tag and the aluminium plate.

Page 56


In the experiments dt is formed by inserting one or two kinds of materials in slice be-

tween the tag and the plate. The materials are bubble wrap of which the thickness is

3mm and Teflon sheet of which the thickness is 0.97mm. It is believed that the effec-

tive permittivity of the bubble wrap is close to be 1. The relative permittivity of Teflon

is usually about 2 with very low losses [57] [58]. In order to minimise the effects of

the Teflon, the Teflon sheet is cut into a much smaller footprint (6mm×8mm) than the

tag. Given the low profile structure and small size, it is believed that the insertion of

the Teflon sheet will not affect the results much either. The reading range tests were

conducted by the equipment introduced before and under Australian UHF RFID reg-

ulations which frequency band is from 920MHz to 926MHz and the available source

power of the reader generator is set to be 4W EIRP (36dBm). The reading range ac-

tually is the distance criterion after which the power received by the tag drops below

its threshold power. In addition, the reading range of the tag in free space under the

Australian regulations and tested by the equipment introduced before is about 5.2m.

Table 3.2. Reading ranges of the self-made tag in proximity to the aluminium plate by

experiments.

dt (mm) 3 4 5 6 7 8

Reading range (mm) 230 350 470 890 1000 1140

As illustrated by Table 3.2, the further the tag is away from the aluminium plate, the

longer the reading range that is obtained. This phenomenon is easily understood since

the metal beside will degrade the performance of the tag antenna. More details of this

degradation is discussed in Chapter 6.

Then, the tag antenna, the aluminum plate behind the tag antenna and the reader an-

tenna were built in the simulation tool Ansoft HFSS. The two antennas’ terminals are

connected to two lumped ports separately. In HFSS, such ports possess implied trans-

mission line characteristic impedances. These lines could be connected to the ports and

those lines allow scattering parameters to be defined.

Page 57


In terms of the characteristic impedance, it can be set in HFSS as an arbitrary complex

impedance. But, in reality the characteristic impedance of the transmission line be-

tween the reader antenna and the reader is 50Ω. As for the characteristic impedance of

the transmission line between the tag antenna and the chip, it can be assumed to be any

value, since its length is nearly zero, its characteristic impedance does not really matter.

But, in order to get the symmetrical scattering matrix, it is set to be 50Ω as well in the

simulation. In terms of source, since in reality the reader antenna is active and the tag

antenna is passive, the lumped ports connected to the two antennas are therefore set to

be an active port and a passive port respectively. In addition, the reader antenna in the

simulation is not exactly the same to the one used in the experiment, since the reader

antenna used in the experiment is a commercial antenna which is enclosed, so that the

inside structure cannot be seen. But it is known that this commercial antenna design

is based on a patch antenna. Hence, in the simulation we designed a patch antenna as

the reader antenna with geometrical and electrical parameters similar to the one in the

experiments.

After building, setting and simulating the model, the S parameters are derived directly

at the two lumped ports. Furthermore, we have already known that the Higgs-2 chip’s

impedance Zchip at 923MHz is about 13.6-j142Ω and the characteristic impedance Z0

of the transmission line is 50Ω. Hence, inserting the derived S parameters, Zchip and

Z0 into (3.61), the power received by the chip at any relative distances among the alu-

minium plate, the tag and the reader antenna can be derived.

As mentioned before, as long as the communication between the reader and the tag is

successful, the power received by the chip should be larger than the threshold power

of the chip which is typically -14dBm. In other words, the longest reading range ap-

pears when the received power falls to -14dBm. Hence, in the simulation, the distance

dt between the aluminium plate and the tag, and the distance between the tag and

the reader antenna will not stop varying until the power calculated by (3.61) reaches

−14dBm to get the longest reading range. The results are shown in Table 3.3.

In order to compare the data in Table 3.2 and Table 3.3, these results are plotted in

Figure 3.9. In Figure 3.9, the x axis represents the distance between the tag and the

Page 58


Table 3.3. Reading ranges of the self-made tag in proximity to the aluminium plate calcu-

lated by (3.61) after deriving S parameters from the simulation.

dt (mm) 3 4 5 6 7 8

Reading range (mm) 200 390 570 880 1040 1160

3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8200

300

400

500

600

700

800

900

1000

1100

1200

dt

(mm)

Rea

din

g r

ang

e (m

m)

Experimental results

Calculated results

Figure 3.9. Comparison between the reading range calculated by (3.61) after deriving the

S parameters from the simulation and the tested reading range.

aluminium plate. The y axis represents the reading range. The red curve comes from

the experimental results which are given in Table 3.2 and the blue one comes from the

calculated results by (3.61) after deriving the S parameters from the simulation which

are given in Table 3.3. The coincidence between the two curves validates (3.61). It may

be noticed that the differences between the two curves is relatively large when dt is less

than 6mm. This is because when dt is small the reading range is very sensitive to the

changes in dt. In the simulation, dt is exactly as the number you give to the simulation,

but in the experiments, as we mentioned before, the distance dt is formed by inserting

the bubble wrap and Teflon sheet between the tag and the aluminum plate. The Teflon

sheet is hard and its thickness at 0.97mm is very close to the 1mm assumed in the

Page 59

3.9 Conclusion

simulation. The thickness of the bubble wrap is about 3mm but it is soft and shape-

flexible, hence the thickness may not be very accurately established. This may be the

reason causing the error.

3.9 Conclusion

According to the discussion above, every aspect, e.g. the transponder IC design, the

tag antenna design, the reader antenna design, and the deployed environment, in an

RFID system affects the operating range of that system. Among all of them, there are a

few factors which we believe play a significant role. (i) The selection of the parameter

θ, the magnitude squared of which establishes the fraction of the available tag antenna

power that is not delivered to the tag chip is one of the keys to lengthening the oper-

ating range, since it governs how much power would be delivered to power the chip

and how much will be backscattered to sense the reader. (ii) The rectifier design is crit-

ical since the enhancement of the rectifier efficiency can lower the threshold power of

the chip. (iii) The environment in which the system is deployed could be an obstacle

in obtaining long operating range, especially when the environment involves many

electro-magnetically sensitive materials surrounding the tags or even very close to the

tag. Those materials include metal and water etc.

To show how to avoid the negative effects brought by the application environment is

part of the work in this thesis. The negative effects can be avoided by the tag antenna

design or decoupler design as part of the tag antenna. One of the decoupler designs

is introduced in Chapter 7. In addition, the negative environmental effects can also

be eliminated by analysing the working environment of the systems in terms of the

electromagnetic wave propagation and making use of the existing tag. In Chapter 8,

the method of reading a large number of DVDs up to 2000 densely stacked is given

by analysing the electromagnetic wave propagation within the DVD stack and making

use of one type of commercial tag.

Page 60

Chapter 4

Analysis and Design ofMeander Line Dipole

Antennas

Asimple analytic formula, found in the literature, for calcu-

lating the resonant frequency of a meander line dipole an-

tenna (MDA) in free space from its physical parameters is

described. The formula is modified to calculate the resonant frequency of

an MDA on a dielectric substrate and used as an RFID tag antenna by taking

two factors into account: (i) the effects of dielectric material underneath the

MDA, (ii) the special needs of an impedance matching condition in RFID

tag antenna design. The parameter of relative effective permittivity for an

MDA on a dielectric board, and the method for deriving this parameter, are

introduced. Experiments to verify the modified formula are reported. Test

results such as input impedance and reading range of an RFID tag antenna

design based on an MDA on a dielectric board are provided. Following

that, the radiation pattern and efficiency of an MDA either in free space or

on a board are investigated.

Page 61

4.1 Introduction

4.1 Introduction

Nowadays the meander line dipole antenna (MDA) is used widely in UHF RFID tag

antenna design [59] [60] because of its size reduction property and relative high radia-

tion efficiency. The MDA is actually a dipole loaded with meander lines. One example

of an MDA loaded with six meander lines is shown in Figure 4.1. Significant research

work has been done on MDAs. Nakano et al. [61], inspired by the appearance of the

meander monopole antenna [62], firstly proposed the meander line dipole antenna

(MDA). Nakano et al. [61] not only proposed the MDA, but also analysed its radia-

tion pattern, input impedance and size reduction ratio relative to the half wavelength

dipole, when they are both resonant at the same frequency. The radiation efficiency of

the MDA has been studied by Marrocco [63]. According to his research, an approxi-

mate current distribution on an MDA loaded with six meander lines is also shown in

Figure 4.1. Since the currents on the adjacent vertical segment of each meander cell

are opposite, these currents actually do not contribute to the radiation but bring losses.

The radiation resistance is mainly determined by the horizontal segments of the MDA.

Moreover, because the large currents occur near the centre of the MDA as shown in Fig-

ure 4.1, one should not place meander lines, especially long meander lines (with large

vertical dimension), near the centre. Genetic algorithms (GA) are also used to obtain a

gain-optimised MDA within a fixed maximum available area in [63]. However, most

analyses of MDAs are based on numerical methods, and to arrive at the optimal MDA

design, such numerical computations have to be iterated, with the result that the cal-

culations are extensive. As a result, a simple analytic formula was proposed by Endo

et al. [14] to calculate the resonant frequency from an MDA’s geometrical parameters.

However, all the literature introduced above including the formula proposed by Endo

et al. [14] assumes that MDA is working in free space, whereas for the purpose of an-

tenna protection, size reduction [64] and high radiation efficiency [65], large numbers

of MDAs are fabricated on dielectric substrates. Although numerical electromagnetic

methods such as MOM (Method Of Moments) and FEM (Finite Element Method) can

provide us reliable and accurate characteristics of MDAs on dielectric substrates, a sim-

ple method is still needed to avoid the burdens of the numerical methods in analysing

the MDA.

Page 62

Chapter 4 Analysis and Design of Meander Line Dipole Antennas

Figure 4.1. A sample of meander line dipole antenna with approximate current distribution.

The arrows on the antenna represent the current flow direction and the number of the

arrows denotes the magnitude of the current distribution.

This chapter aims, by summarising the existing literature, to give a complete analysis

of resonant frequency, radiation pattern and radiation efficiency of an MDA and by

contributing original thoughts to modify the formula proposed by Endo et al. in [14]

to calculate the resonant frequency of an MDA not only in free space but also above a

dielectric board for RFID tag antenna design.

The outline of this chapter is as follows. Section 4.2 introduces and validates the for-

mula proposed by Endo et al. [14]. Section 4.3 gives the limitations of the formula

proposed by Endo et al. [14] for calculating the resonant frequency of an MDA on a

dielectric board and used for an RFID tag antenna. In order to overcome these limita-

tions, several modifications are made to the formula. One significant modification is to

add a new factor named as relative effective permittivity εre f f in the formula so that the

effects of the dielectric substrate can be taken into account. The method for deriving

the factor εre f f is given and verified by simulation software Ansoft HFSS. An RFID tag

antenna based on the MDA pattern is investigated and tested in Section 4.4 to verify

the modified formula. Following that, the radiation pattern and radiation efficiency of

MDA are discussed in Section 4.5. Finally, in Section 4.6, conclusions are drawn.

4.2 Introduction and Validation of the Formula for Calcu-

lating Resonant Frequency of an MDA in Free Space

4.2.1 Formula Derivation

This subsection introduces the derivation of the formula proposed by Endo et al. [14]

for calculating the resonant frequency of an MDA in free space, from its geometrical

parameters.

Page 63

4.2 Introduction and Validation of the Formula for Calculating ResonantFrequency of an MDA in Free Space

Figure 4.2 shows a dipole antenna with two meander lines. The two parallel vertical

lines are treated as twin lines with a short circuited termination. In addition, the bold

line and the dashed line are considered as a straight conducting wire with length s and

diameter b.

s

w

hb

Conductingline

Short circuitedterminal

Twin line

Figure 4.2. Meander line dipole antenna loaded with two meanders.

The derivation of the MDA’s resonant frequency proceeds as follows. The characteris-

tic impedance of twin lines can be expressed in the following form [37]:

Z0 =η

πlog

2wb

(4.1)

where, η is the wave impedance in free space, w is the distance between twin lines, b is

diameter of the conducting wire, and log is the natural logarithm operation.

Zin is the input impedance of twin lines, which is given by (4.2) [37]:

Zin = Z0ZL + jZ0 tan βhZ0 + jZL tan βh

(4.2)

where β is equal to 2π/λ. Now suppose that all twin lines are terminated in a short

circuit. Thus, the load impedance of the twin lines is zero (ZL = 0), and (4.2) becomes

(4.3):

Zin = jZ0 tan βh (4.3)

where h is the height of twin lines. Following Endo et al. [14], tan βh can be expanded

into three orders on condition that βh ¿ 1:

tan βh ≈ βh +13(βh)3 (4.4)

Then a new expression of input impedance is obtained:

Zin = jωL = jZ0[βh +13(βh)3] (4.5)

Page 64


If we insert (4.1) into (4.5), the inductance formed by each twin line can be shown to be

(4.6):

L =µ0 · h

π[1 +

13(βh)2] log

2wb

(4.6)

On the assumptions that the number of meanders is m and all the meander lines are

identical, the total inductance obtained by the twin lines should be Lp = m× L. The

straight conducting wire, which length is s, also results in a self-inductance. It is given

by the following equation [14]:

Ls =µ0

2πs(log

4sb− 1) (4.7)

where µ0 is the vacuum permeability. Then (4.6) and (4.7) can be solved to obtain the

total inductance of the MDA.

LT = Ls + m× L (4.8)

Thus

LT =µ0

2πs(log

4sb− 1) + m · µ0 · h

π[1 +

13(βh)2] log

2wb

(4.9)

Since the half wavelength dipole is composed by a straight wire which length is about

λ/2, its self-inductance can thus be derived by (4.7):

LH =µ0

π

14

λ(log2λ

b− 1) (4.10)

Following Endo et al. [14] we suppose that the inductance of MDA and half wave-

length dipole antenna is the same when they resonate at the same frequency, thus

LH = LT.

µ0λ

4π(log

2λ

b− 1) =

µ0

2πs(log

4sb− 1) + m · µ0 · h

π[1 +

13(βh)2] log

2wb

(4.11)

As predicted by (4.11), the wavelength λ of the resonant frequency is decided by the

physical dimension of an MDA. In detail, the resonant frequency declines as some

features increase such as the meander lines’ height h, number of folds m, the ratio

w/b of the meander lines’ width to the conducting wire’s diameter and the conducting

wire length s. Moreover, the loaded position of the meander lines on the dipole does

not affect the resonant performance.

Page 65

4.2 Introduction and Validation of the Formula for Calculating ResonantFrequency of an MDA in Free Space

4.2.2 Validation of Equation (4.11)

For examining the validity of the method introduced in Subsection 4.2.1, the simulation

software Ansoft HFSS is employed. MDA loaded with different numbers of meanders

are modeled by HFSS as shown in Figure 4.3, where the number of meanders m=2, 8,

14, the length of the MDA s=129mm, and the gap between dipole arms is 3mm.

In order to establish the influence brought by each parameter, the following method-

ology is used. First, the length of MDA s and diameter b remain 129mm and 1mm

respectively in all the following cases. Secondly, two of the three geometrical param-

eters of an MDA, (i) m, (ii) h, (iii) w, are fixed. Then, the unfixed one is varied over a

range. Therefore, the resonant frequencies of MDA for different shapes can be obtained

by (4.11) and simulation software Ansoft HFSS respectively as shown in Figure 4.4 in

which all the green curves are derived by (4.11) and all the blue curves are derived by

simulation.

s

m=8

m=2

m=14

b

D

Figure 4.3. Three models of MDA with various numbers of meander lines.

The number of meander lines m is varied from 2 to 14 on condition that each meander

line width w=6mm, height h=10mm. The resonant frequencies calculated by (4.11)

and HFSS are derived, as shown in Figure 4.4(a). The resonant frequency of MDA as

a function of the meander line height h is shown in Figure 4.4(b) when the meander

line width w is 6mm and the number of the meander lines m is equal to 2, 8 and 14

respectively. Similarly, the resonant frequency of MDA as a function of the meander

line width w is shown in Figure 4.4(c) when the meander line height h is 10mm and the

Page 66


number of the meander lines m is equal to 2, 8 and 14 respectively. The meander lines

discussed above are loaded in the middle of the dipole.

The influence on resonant frequency resulting from loaded position of meander lines

can be studied by moving the meander lines close to the end of the dipole step by step

which means the distance D in Figure 4.3 is diminished step by step. Additionally,

the values of other features remain unchanged (w=6mm, h=10mm, m=2, 8 or 14). The

resonant properties analysed by (4.11) and HFSS are illustrated in Figure 4.4.

2 4 6 8 10 12 14600

650

700

750

800

850

900

950

1000

1050

Number of meander lines m

Res

onan

tF

req

uen

cy(M

hz)

HFSS

Equation (4.11)

Meander line height 10mmh=

Meander line width 6mmw=

(a)

2 4 6 8 10 12500

600

700

800

900

1000

1100

1200

Meander line height (mm)h

Res

on

ant

Fre

qu

ency

(M

Hz)

m=2, HFSS

m=2, Equation (4.11)

m=8, HFSS


m=14, HFSS

m=14, Equation (4.11)Meander line width = 6mmw

m=2

m=8

m=14

(b)

2 2.5 3 3.5 4 4.5 5 5.5 6500

600

700

800

900

1000

1100

Meander line width (mm)w

Res

onan

t F

requen

cy (

MH

z)

m=2, HFSS


m=8, HFSS


m=14, HFSS



m=2

m=8

m=14

(c)

5 10 15 20 25 30400

500

600

700

800

900

1000

1100

Res

on

ant

Fre

qu

ency

(MH

z)

m=2, HFSS


m=8, HFSS


m=14, HFSS


The distance between the meanders and the end of MDA (mm)D


Meander line width 6mmw=

m=2

m=8

m=14

(d)

Figure 4.4. The resonant frequency of MDA as a function of its physical parameters.

Page 67

4.3 Modifications on Equation (4.11) for RFID Tag Antenna Design

All in all, the resonant characteristics of MDA predicted by (4.11) have a qualitative

agreement with simulation results. The meander line’s width w, height h and number

m are the features which contribute to the resonant characteristics and the meander

lines’ loaded position does not but this factor affects the radiation efficiency which will

be introduced in Section 4.5. Interestingly, the simulation results and the calculated

results are getting closer with the increase of the number of meander lines. Clearly,

two curves almost superpose each other, when the number is fourteen. As a result,

this method can be applied to calculate the MDA’s resonant frequency well when the

number of the meander lines is large. Compared to HFSS, this method does not need

to build models for analysis. Moreover the calculation efficiency is much higher than

the traditional methodology such as Moment of Method (MOM) and Finite Element

Method (FEM), so that the designers can consider the shape of an MDA before design-

ing it and making it resonant at a desired frequency.

4.3 Modifications on Equation (4.11) for RFID Tag An-

tenna Design

4.3.1 Limitations of Equation (4.11) in RFID Tag Antenna Design

The introduction in Section 4.2 demonstrates that (4.11) can derive a reasonably ac-

curate resonant frequency from the geometrical parameters of an MDA or in other

words, (4.11) can estimate an MDA’s shape at a known resonant frequency. However,

(4.11) is not suitable to estimate the MDA’s shape on a dielectric board for RFID tag

antenna design, because 1) when the MDA is placed or manufactured on a dielectric

substrate, the effects of the substrate should be considered. 2) An MDA on a board is

composed of metal strip instead of the wire assumed in (4.11), so an electrical equiv-

alent diameter of a dipole made from a strip should be calculated from b = 0.5a [37],

where b is the equivalent wire diameter and a is the width of the strip. 3) The size of

each meander could be different, whereas the meanders in (4.11) are uniform. 4) Com-

pared with antennas employed in other areas, the RFID tag antenna design is special

in its impedance matching condition, in that the tag antenna input impedance is not

Page 68


required to be real but complex, since the RFID tag antenna should be connected to

a chip, of which the impedance is complex (a small real component and a ten times

larger imaginary impedance, approximately 12-j130Ω around 1GHz). Hence, a needed

inductive reactance Xa should be taken into account in designing a tag antenna. After

considering all the four issues, (4.11) is modified to be (4.12).

µ0λ′

4π(log

4λ′

a− 1) + La =

µ0s2π

(log8sa− 1)

+µ0h1

π[1 +

13(β

′h1)2] log

4w1

a+ · · ·+

µ0hn

π[1 +

13(β

′hn)2] log

4wn

an = (1, 2, 3, · · · , n) (4.12)

where λ′

= λεre f f

, β′

= 2πλ′ , εre f f is the relative effective permittivity representing the

effects of the dielectric substrate to the resonant frequency of an MDA. hn, wn are the

nth meander line’s height and width respectively and La = Xa/ω is the inductance

brought by the extra needed inductive reactance Xa, and ω is angular frequency. The

meander line height and width are measured between mid-lines of the strips. If the

geometrical parameters of the MDA on a dielectric substrate and the impedance of the

chip which is going to be mounted on an MDA are known, the only unknown factor

in (4.12) is εre f f which is discussed in the next subsection.

4.3.2 Method for Calculating Relative Effective Permittivity of an

MDA on a Dielectric Substrate

Before introducing the method for calculating the εre f f of an MDA on a dielectric sub-

strate, the method for calculating the εre f f of the coplanar strip (CPS) transmission line

structure is given, since the former is developed according to the latter.

Page 69


(a) Relative effective permittivity of CPS on board

The two coplanar strips (CPS) lying on a dielectric substrate are shown in Figure 4.5.

The yellow rectangle denotes an infinite substrate with dielectric constant εr and thick-

ness t. The two brown rectangles represent the metal strips, with width d−c2 and a gap

c between them.

dct er

x

yz

Figure 4.5. Two coplanar strips on a dielectric substrate.

The transverse electric field configurations of the CPS for quasi-static approximation

are shown in Figure 4.6 in which the electric field travels from one strip into the air and

ends at the other strip.

er

x

z

Figure 4.6. The transverse electric field distribution in the cross section of a CPS on board.

The determination of the distributed capacitance between the strips or the effective

dielectric constant requires a solution of Laplace’s equation. Hanna [66] proposed an

equation for estimating the εre f f of the CPS shown in Figure 4.5 by a conformal map-

ping technique. A conformal mapping is a function that transforms curves in one

complex plane to other curves in another complex plane, preserving angles between

intersecting curves as it does so. It has the property that if a potential satisfies Laplace’s

Page 70


equation in the original coordinates it will continue to satisfy it in the transformed co-

ordinates. Conformal mapping is useful for solving problems in physics involving

inconvenient geometries. By making use of a conformal mapping technique, the origi-

nal coordinate system shown in Figure 4.5 may be transformed to a new one in which

the electric field or potential is parallel to one of the axes, and the solution to Laplace’s

equation is easily found. The Hanna paper does not disclose much of this technique;

only the final equations which are expressed as follows.

εre f f = 1 +εr − 1

2K(k

′)K(k1)

K(k)K(k′1)(4.13)

where

k =cd

(4.14)

k′=

√1− k2 (4.15)

k1 =sinh(πc

4t )sinh(πd

4t )(4.16)

k′1 =

√1− k2

1 (4.17)

and K(k) is the complete elliptic integral of the first kind.

(b) Relative effective permittivity of MDA on board

We firstly think the method for calculating the relative effective permittivity of a CPS

may be useful for calculating the counterpart of an MDA because of their shape resem-

blance. As mentioned previously, each meander line on an MDA can be regarded as

a CPS transmission line shortened at the end. Hence, it is assumed that the εre f f of an

MDA on a dielectric substrate as a whole is approximately equal to the εre f f of each

meander line on the same dielectric substrate which is actually a CPS model. Then,

the former can be calculated according to the latter. For example, an MDA loaded with

four identical meander lines on a dielectric substrate is shown in Figure 4.7. The yellow

rectangle denotes the FR4 substrate which has dielectric constant εr equal to 4.4 and

thickness t equal to 1.6mm. The brown strips represent the copper tape consisting the

meander lines. The strip width is 1mm and the distance between the adjacent strips is

1.7mm. Hence, by inserting the geometrical parameters of the meander line into (4.13),

i.e. c=1.7mm, d=3.7mm, t=1.6mm, εr=4.4, the εre f f is derived as 2.4.

Page 71


52.6

11

3.7

2 1

L

W

Cross Section

Unit: mm

Figure 4.7. An MDA loaded with four identical meander lines.

In order to test whether this calculation is valid, simulation was conducted on the MDA

shown in Figure 4.7 by HFSS. The method for deriving the εre f f by HFSS is described

as follows. First, the model in Figure 4.7 with the dielectric substrate, which thickness

is 1.6mm and dielectric constant is 4.4, is simulated and its resonant frequency with

the board, defined as the lowest frequency at which the reactance is equal to zero, is

obtained and is denoted by fb. Then, the substrate is removed from the model, so

that the MDA is placed in free space, and the resonant frequency f f of the MDA in

free space is obtained. The relative effective permittivity brought by the board is then

derived by (4.18).

εre f f = (f f

fb)2 (4.18)

For the model shown in Figure 4.7, W and L are set to be large enough in the simula-

tion so that they can be regarded as an infinite board, because we intend to compare the

simulated results with the calculated results derived by (4.13) which assumes the di-

electric substrate is infinite. fb and f f obtained from HFSS are 1047MHz and 1279MHz

respectively, so that εre f f of the MDA is obtained by (4.18) to be 1.49. This value is

much smaller than 2.4 previously derived by (4.13).

We then realised that the εre f f was not simply dependent on the shape resemblance

but ultimately dependent on the electric field distribution. If we look into the trans-

verse field distributions of an MDA and a CPS on a dielectric board, we will still find

the resemblance of the field distribution which makes (4.13) useful in calculating the

Page 72


εre f f of an MDA after making some modifications. The transverse electric field in the

cross section of a CPS has been shown in Figure 4.6. It is clear that the electric field

travels from one strip to the other. In terms of strength the field will be stronger if the

observing point is closer to the strip and the field attenuates if the observing point is

moved away from the strip.

The electric field magnitude distribution of the MDA shown in Figure 4.7 can also

be obtained by the simulation software Ansoft HFSS. When the dielectric substrate

thickness is 1.6mm and the relative permittivity is 4.4, the resonant frequency of the

MDA is about 1047MHz as mentioned previously. At this frequency, the electric field

distribution at the cross section marked in Figure 4.7 when the antenna is excited by a

1A r.m.s. current at the feed point is given in Figure 4.8.

Figure 4.8. Cross section view of electric field magnitude distribution of the MDA shown in

Figure 4.7.

According to Figure 4.8, the strong electric fields concentrate on the two strips at the

ends of the MDA and diminishes when the fields go further away from these two

strips. This field distribution is similar to that of CPS model. Based on this observation,

the other assumption can be made that the electric field radiated from an MDA can be

represented by the electric field radiated from a CPS model by treating the two strips at

the ends of the MDA as two coplanar strips (CPS) and ignoring the other components

of the MDA. In addition, the electric field distribution ultimately decides the εre f f , if

the electric field distributions are similar, the εre f f should be similar. The εre f f of the

Page 73


MDA can thus be calculated by Equation (4.13) after adjusting the values of c and d

in that equation according to the geometry of the MDA. In the case of the MDA in

Figure 4.7, d should be the whole width of the MDA which is 52.6mm and c is the

gap between the two strips at the ends of the MDA which is 50.6mm. Substituting

d=52.6mm, c=50.6mm, t=1.6mm and εr=4.4 into (4.13), the εre f f is derived as 1.61 which

is very close to the simulation result 1.49.

In order to further verify the method for calculating the εre f f of an MDA by (4.13)

as introduced above, more simulations have been done on the MDA in Figure 4.7 by

varying the thickness t and the relative permittivity εr of the dielectric substrate. The

εre f f based on the simulation results and derived by (4.13) are given in Figure 4.9. In

detail, the εre f f as a function of the εr is illustrated in Figure 4.9(a), when the substrate

thickness remains 1.6mm. Similarly, the εre f f as a function of the dielectric substrate

thickness t is illustrated in Figure 4.9(b), when the dielectric constant εr is held to be

constant at 4.4. In addition, the blue curves in Figure 4.9 represent the results derived

by (4.13) and the red curves represent the results derived by the simulation.

1 1.5 2 2.5 3 3.5 4 4.5 51

1.1

1.2

1.3

1.4

1.5

1.6

1.7

1.8

Relative permittivity

Rela

tive e

ffective p

erm

ittivity

(a)

1 1.5 2 2.5 3 3.5 4 4.5 51.3

1.4

1.5

1.6

1.7

1.8

1.9

2

Substrate thickness

Rela

tive e

ffective p

erm

ittivity

(b)

Figure 4.9. The relative effective permittivity of the MDA in Figure 4.7 as a function of

the dielectric constant εr for sub-figure(a) and as a function of the substrate

thickness t for sub-figure(b).

Page 74


The coincidence in Figure 4.9(a), at least for low values of the relative permittivity,

between the simulation results and the calculated results obtained by (4.13) demon-

strates the method of simplifying the MDA to the CPS according to their resemblance

in field distribution to calculate the MDA’s εre f f is feasible. But, it is also noted by Fig-

ure 4.9(a) that the higher the relative permittivity εr is, the less the agreement between

the simulation results and the calculated results.

We then investigated the effects of various relative permittivities (εr) on the electric

field distribution of the MDA. In order to observe the effects easily, the relative per-

mittivity εr of the substrate underneath the MDA shown in Figure 4.7 is varied over

a large range from 1 to 50. Meanwhile the thickness of the substrate remains 1.6mm.

Simulations were conducted on the MDA respectively when the εr are 1, 10, 20, 30,

40 and 50. The electric field magnitude of the MDA at its resonant frequency defined

earlier and depending on the εr was obtained by the simulation. The electric field dis-

tribution at the cross section marked in Figure 4.7 when the antenna is excited by a 1A

r.m.s. current at the feed point is shown in Figure 4.10. The subtitle under each figure

gives the values of the relative permittivity εr and the resonant frequency fb.

According to Figure 4.10, it is found that along with the increase of the εr, the electric

field magnitude near the two strips at the two ends of the MDA decreases more rapidly

than the electric field magnitude in other places does. Hence, when the εr becomes

large, besides the two strips at the ends, the effects of other components of the MDA

should be taken into account for calculating the εre f f . But, the method for calculating

εre f f of an MDA proposed here only considers the two strips at the ends. This explains

the divergence between the curves in Figure 4.9(a).

However, this does not impedes the usage of (4.13) in designing MDA in RFID appli-

cations, since the dielectric materials used in fabricating or packaging RFID tags are

usually low dielectric constant materials (1 < εr < 4), and the divergence between the

curves of Figure 4.9 is then acceptable.

Another notable aspect of this method is the dielectric substrate has been assumed

infinite but in most of the RFID applications, the dielectric substrate size usually is

equal to the footprint of the MDA or a little bit larger. However, simulation results

Page 75


(a) εr=1, fb=1279MHz (b) εr=10, fb=872MHz

(c) εr=20, fb=720MHz (d) εr=30, fb=633MHz

(e) εr=40, fb=570MHz (f) εr=50, fb=523MHz

Figure 4.10. The variation of the MDA’s electric field magnitude distribution at the resonant

frequency along with the variation of the εr.

not shown have confirmed that the variations of the substrate size does not affect the

results of εre f f significantly.

4.3.3 Further Validation of the Method for Calculating the εre f f of

an MDA on a Dielectric Substrate

In the last subsection, the method for calculating the εre f f of an MDA on a dielectric

substrate was proposed and this method was initially examined by simulations on the

MDA shown in Figure 4.7. In order to further confirm this method, we examine this

method by simulations on another type of MDA which is shown in Figure 4.11. It is

Page 76


noted that the shape of the MDA in Figure 4.11 is quite different from that in Figure 4.7

in terms of the number of the meander lines and the shape of each meander line.

According to the geometry of the MDA in Figure 4.11 and the method for calculating

the εre f f of an MDA by (4.13), introduced in Subsection 4.3.2, the values of c and d

which should be substituted in (4.13) are 39.8mm and 41.8mm as marked in Figure 4.11.

The values of the εre f f of the MDA on a dielectric substrate obtained by (4.13) and

simulations are shown in Figure 4.12. In detail, Figure 4.12(a) describes the variation

of the εre f f along with the variation of the εr at three different substrate thicknesses.

Similarly, Figure 4.12(b) describes the variation of the εre f f along with the variation of

the substrate thickness at three different εr.

In addition, the blue curves in Figure 4.12 represent the results derived by (4.13) and

the red curves represent the results derived by the simulation.

d=41.8

26.8 3.7

2 1

L

W

10

Unit mm

c=39.8

Figure 4.11. An MDA loaded with three different meander lines.

The coincidence in Figure 4.12 between the simulation results and the calculated results

demonstrates again that the method of simplifying the MDA to the CPS according to

their resemblance in electric field distribution to calculate the MDA’s εre f f is feasible.

Again, it is observed that with the increase of the εr, the divergence between the two

curves becomes large in Figure 4.12(a). The observation of this divergence has been

explained previously in Subsection 4.3.2.

Page 77

4.4 Experimental Validation of Equation (4.12)

1 1.5 2 2.5 3 3.5 4 4.5 51

1.2

1.4

1.6

1.8

2

t=3mm

t=1.6mm

t=1mm

Relative permittivity

Rel

ativ

e ef

fect

ive

per

mit

tiv

ity

(a)

1 1.5 2 2.5 3 3.5 4 4.5 51.1

1.2

1.3

1.4

1.5

1.6

1.7

1.8

1.9

2

2.1

er=4.4

Substrate thicknessR

elat

ive

effe

ctiv

e per

mit

tivit

y

er=3

er=2

(b)

Figure 4.12. The relative effective permittivity of the MDA in Figure 4.11 as a function of

the dielectric constant εr for sub-figure(a) and as a function of the substrate

thickness t for sub-figure(b).

The experimental validation of this method and Equation (4.12) is discussed in the next

section.

4.4 Experimental Validation of Equation (4.12)

In order to test the validity of (4.12), the MDA shown in Figure 4.11 was fabricated and

its input impedance was measured. The dielectric substrate of the MDA is FR4 with

overall size W × L=43.8mm×28.8mm, thickness t = 1.6mm and dielectric constant

εr = 4.4. The dimensions of the tested MDA have been optimised by the simulation

software HFSS in order to make its input impedance a conjugate match to the chip

impedance (13.6-j142Ω) at 923MHz. The frequency of 923MHz was chosen because it

is the centre frequency of a 6MHz band according to Australian RFID standards.

The input impedance is tested by the method which is shown in Figure 4.13. The un-

balanced version of the MDA is soldered on to an SMA connector which was mounted

on a ground plane. An image of the half MDA is created by the ground plane, which

can complete the whole MDA. A half unbalanced antenna on a ground plane will have

Page 78


half of the input impedance of a complete balanced antenna. The input impedance

of the half MDA on a ground plane at 923MHz measured by the network analyser

8714C was 5.7+j60Ω, as is shown in the Smith Chart in Figure 4.14. Hence, the input

impedance of the complete MDA is 11.4+j120Ω, which is very close to the target design

impedance. The complete MDA mounted with a Higgs-2 chip manufactured by Alien

Technology [55] is shown in Figure 4.15. The reading range under the 4W EIRP of the

tag is 5.2m. The details of the testing method, equipment and the details of the chip

can be found in Subsection 3.8.2.

(a) Photo of the tested half MDA

FR4

Copper

Ground plane

SMA

(b) Details and side view of the tested half MDA

Figure 4.13. The half MDA on a ground plane being tested.

Inputting the physical dimensions and εr equal to 4.4 of the MDA on the substrate

being examined into (4.13) to derive the εre f f gives the result 1.64. Then, substituting

εre f f = 1.64 into (4.12) with the physical dimension of the MDA to calculate the reso-

nant frequency of this MDA with a known complex chip impedance, gives the result

910MHz. Comparing the experimentally verified resonant frequency of 923MHz with

the calculated resonant frequency of 910MHz reveals that they are reasonably close to

each other which validates the feasibility of using (4.12) for design.

4.5 Radiation Pattern and Efficiency

As is verified by Nakano et al. [61], the radiation pattern of an MDA loaded with uni-

form meanders in free space is similar to that of the half wavelength dipole. We also

found by using the Ansoft HFSS simulation software that the radiation pattern of an

MDA on a dielectric board, either loaded with uniform meanders or different size me-

anders, is similar to that of the half wavelength dipole. Two factors which affect MDA’s

Page 79


Figure 4.14. Smith chart derived by the network analyser 8714C showing input impedance

of the half MDA on a ground plane. The mark is at the frequency of 923MHz.

Figure 4.15. A tag based on the MDA in Figure 4.11.

radiation efficiency are discussed in the two following subsections respectively. The

two factors are the physical dimension of the MDA and the properties of the dielectric

board underneath.

4.5.1 Physical Dimension of MDA

The radiation efficiency is defined as (4.19).

ηr =Pr

Pr + PL(4.19)

Page 80


where Pr is the radiation power and PL is the loss power.

Following Endo et al. [14] we assume that the distribution of the current along the

conductor forming an MDA is in sinusoidal form, and the peak value of the current

occurring in the middle of the MDA is I0. This current diminishes to zero at the two

ends of the MDA. The current can be expressed as in (4.20).

I(z) = I0 sin k′(l2− |z|) (4.20)

where z is the coordinate along the conductor and the MDA’s feed point sits on the

origin of this coordinate. In contrast with Endo et al., our l is the whole length of the

conductor either in the form of strip or wire, so our k′ which has the same value as

Endo et al. is equal to π/l.

The loss resistance per unit length of the conductor is expressed in (4.21).

RL =1P

√ωµ0

2σ(4.21)

where ω is the angular frequency, µ0 is the permeability of free space, σ is the conduc-

tivity, and P is the perimeter of the cross section of the conductor (P = πb for a circular

wire of diameter b; P = 2a for a very thin strip of width a).

Hence, the loss power of the MDA can be expressed in (4.22).

PL =∫ l

2

− l2

I2(z)RLdz (4.22)

Substituting (4.20) and (4.21) into (4.22), the loss power can be derived as:

PL =12

I20

lP

√ωµ0

2σ(4.23)

(4.23) states that the loss power of the MDA is proportional to the whole length of the

strip or wire comprising the MDA and the square root of the angular frequency, and it

is inversely proportional to the strip width and square root of conductivity.

The radiation power can be obtained by (4.24).

Pr =12

I20 Rr (4.24)

Page 81


where Rr is the radiation resistance defined at the feed point. The fact of 12 results from

our use of peak value phasors.

In [14], it is assumed and experimentally verified that the ratio of the horizontal length

of an MDA to the length of a half wavelength dipole is equal to the ratio of the radiation

resistance of the MDA to the radiation resistance of the half wavelength dipole, if the

two antennas are resonant at the same frequency. The equality is expressed in (4.25).

Rr

Rd=

sλ/2

(4.25)

where s is the horizontal length of the MDA as is shown in Figure 4.3. Rd is the radia-

tion resistance of the half wavelength dipole which is already known as 73Ω. Hence,

the radiation resistance of the MDA can be expressed as (4.26).

Rr =2sRd

λ(4.26)

Therefore, the radiation power becomes (4.27) after inserting (4.26) into (4.24).

Pr =12

I20

2sRdλ

(4.27)

Substituting (4.23) and (4.27) into (4.19), the other form of radiation efficiency of the

MDA can be obtained:

ηr =2sRd

2sRd + lP

√ωµ02σ

(4.28)

where l (the total length of the wire or strip comprising the MDA) is equal to s + 2mh,

if the MDA is loaded with m uniform meander lines which height is h. Generally

speaking, loading more and longer meander lines on a dipole will decrease the res-

onant frequency or decrease the ratio of the horizontal length of the antenna to the

wavelength according to the study in the previous sections. However, more or longer

meanders result in the increase of the whole length of the wire or strip comprising the

MDA (l goes up) and that increase according to (4.28) lowers the radiation efficiency.

Therefore, a tradeoff has to be made between the horizontal length of the MDA and

radiation efficiency at a known resonant frequency.

In addition, according to the assumption of the sinusoidal current distribution on an

MDA, the current is large in the middle and declines as the current flows further away

Page 82


from the centre of the MDA. Hence, the meander line loaded in the middle leads to

more losses than the same meander loaded at the end of the MDA. For example, in

Figure 4.16, there are two types of MDA. According to the study in Section 4.2 and Sec-

tion 4.3, the two MDAs will have similar resonant frequencies whether they are in free

space or on a dielectric substrate. However, for achieving better radiation efficiency,

type (a) is preferable. For the MDA loaded with uniform meanders, leaving more hor-

izontal segments near the feed port before loading the meander lines is preferable.

(a) (b)

Figure 4.16. Radiation efficiency comparison between two types of MDA.

4.5.2 Dielectric Substrate

The effects on an MDA’s radiation efficiency by a dielectric substrate have been studied

by Yamada et al. [65]. The radiation efficiency of two type MDAs is compared. Both

of them have the same footprint and are loaded with uniform meander lines. One of

them is placed in free space, εr = 1. The other MDA is sandwiched by two dielectric

boards of thickness t = 0.1mm and εr = 10. By adjusting the number of meander lines

m loaded on the MDA, these two antennas are made resonant at the same frequency.

It is found that the MDA sandwiched between high dielectric constant material needs

fewer meander lines than the other does and the smaller number of loaded meander

lines results in higher radiation efficiency as analysed above.

The statement in [65] that by deploying higher dielectric constant material will result

in fewer loaded meander lines can be explained by (4.12). By deploying higher εr

material, the factor on the left side of (4.12) becomes smaller, so that on the right side

of (4.12), fewer meander lines are needed to make the equation valid.

Page 83

4.6 Conclusion

4.6 Conclusion

In this chapter, a simple analytic Equation (4.11) from the literature [14] for calculating

the resonant frequency of an MDA in free space is introduced and verified by sim-

ulations. However, this equation has some limitations in analysing the MDA on a

dielectric substrate for RFID tag antenna design. In order to overcome those limita-

tions, Equation (4.12) is derived after making some modifications to (4.11). One signif-

icant modification is to add a new factor named as relative effective permittivity εre f f

in Equation (4.12) so that the equation can include the effects brought by the under-

neath dielectric board. The method of deriving the factor εre f f is given and verified by

simulations. In addition, the special needs of an impedance matching condition in de-

signing RFID tag antennas is also taken into account in Equation (4.12). The modified

Equation (4.12) is experimentally examined. Following that the radiation pattern and

efficiency of an MDA are discussed.

In addition, it should be noted that the method for obtaining the factor εre f f has its own

limitation that the method can only give reasonably accurate result when the relative

permittivity εr of the substrate is in the range from 1 to 4. However, this does not

impede the usage of the method in analysing the MDA made for RFID tags, since the

relative permittivity εr of the materials usually used for packaging and manufacturing

the RFID tags are within that range.

In summary the content in this chapter provides a thorough analysis of meander line

dipole antennas which are commonly used for designing RFID tag antennas because

of their size reduction and relatively high radiation efficiency properties. By making

use of the modified Equation (4.12), the shape of the MDA designed as an RFID tag

antenna can be approximately estimated. The process of the estimation is much more

efficient than that by simulation software based on numerical methods, but to some

extent at the expense of accuracy. As a result, we recommend that an antenna designer

should use Equation (4.12) in the primary design of the MDA for RFID tag and then

finalise the design by simulation software. The combination of the analytic equation

and the simulation can both shorten the design cycle and maintain the accuracy.

Page 84

Chapter 5

A Security Tag Design

AN electronic seal, based on a passive UHF RFID tag, and

called a “T-seal” because of its shape, is described in this

chapter. This seal is proposed by Wi-Protect and intended

to protect any container (large or small) that has either a) two sides that

join together where a sealing chamber is attached or b) a finger that slots

into a chamber, from being compromised. The author of this thesis partic-

ipated in this project by providing the tag antenna design which achieved

good reading range and a required off-on-off functionality. After consider-

able theoretical analysis and simulation work, a novel tag with the required

security function was designed and tested successfully.

Page 85

5.1 Introduction

5.1 Introduction

The concept of the T-seal as proposed by Wi-Protect, is an electronic seal incorporat-

ing a passive UHF RFID tag, attached to a two piece sealing chamber, to detect any

tampering with the seal. The author of this thesis took part in the project to design

the required tag antenna which provides good reading range and a particular off-on-

off functionality. The tag antenna is the most critical part in achieving that required

function. The author of the thesis produced a satisfactory design. The research on and

design of the security tag antenna is described in this chapter.

Following this brief introduction, Section 5.2 describes the concept of the T-seal. That

section seeks to provide an understanding to the function of the seal, its general shape,

and its requirement for an off-on-off operation. The emphasis is put on how this T-seal

can achieve outstanding security protection by the off-on-off function realised in the

context of a UHF RFID system. Section 5.3 describes the chip and general properties

of the antenna to be used in the security tag design. In Section 5.4, the realisation of

the off-on-off function or the security protection by the RFID tag design is introduced.

During that introduction, the previous work in Chapter 4 is made use of as theoreti-

cal background. The chapter ends with Section 5.5 which provides some conclusions

about the T-seal project and suggestions for further work thereon.

5.2 T-Seal Concept

In this section, the concept of the T-seal is introduced. As noted before, the T-seal

actually is a passive UHF RFID tag which is attached to the sealing chamber of any

container (large or small) that has either a) two sides that join together where a sealing

chamber is attached or b) a finger that slots into a chamber. An example is provided by

a shipping container. By orchestrating the design, the T-seal aims to achieve an off-on-

off operation, where the first “off” means before the attachment of the T-seal, the RFID

tag in the seal cannot be detected by a reader, no matter how close the tag is to the

reader antenna, because some parts of the tag antenna are missing. The middle “on”

means once the missing parts of the tag antenna are completed by the seal attachment

Page 86

Chapter 5 A Security Tag Design

operation, the tag can be read at a good range by a reader. The second “off” operation

denotes that the seal has been compromised or broken. This happens when the T-seal

has been tampered with, and leads to a failed detection from the reader. This failure can

excite an alarm or log a record of this compromise or break. Therefore, the “off-on-off”

operation is divided into two sections, the “off-on” is named as “turning on” and the

“on-off” is named as “turning off”. Furthermore, the reason of needing the first “off”

is because we aim to avoid some people defeating the system by using an improper

seal attachment operation. For instance, if the tag can be read before it is sealed on a

container, some people might just put or stick the tag on the container but not attach

it in the chamber to seal the container, which the owner of the container assumes has

been done. Consequently, the tag’s status in the view of the reader is always “on” even

though the container might have been opened. The structure of the T-seal is shown in

Figure 5.1.

As shown in Figure 5.1, the T-seal is composed of three main components. These are

the body shown in Figures 5.1(a), (b), (c), the cover shown in Figures 5.1(e), (f), and the

RFID tag which will be placed between the body and cover as shown in Figure 5.1(d).

Hence, the RFID tag must be designed to fit in a particular space, the dimensions of

which are marked in Figure 5.1 (e). Once the body, cover and an RFID tag are packaged

as a unit, the so-called T-seal, to protect valuable items in a container, the seal is locked

into a chamber consisting of a latch by upward locking spurs. Hence, the T-seal serves

as a one-directional lock. Anybody who intends to open the seal has to destroy the seal

forcibly.

To understand the T-seal, and in particular the scheme of off-on-off operation we ex-

pand the description as follows. Two pieces of good conductor which are named as

terminals and shown in Figure 5.1(g) have been manufactured beforehand as part of

the chamber. When the T-seal is plugged into the chamber, the two terminals will be

connected to the tag antenna through the two holes at the bottom of the body shown

in Figure 5.1(d) serving as parts of the tag antenna. In other words, before the T-seal

is plugged in the chamber, the RFID tag antenna is not complete so that the RFID tag

cannot be detected and it is thus in “off” operation. Once the T-seal is plugged in the

Page 87

5.2 T-Seal Concept

RF

TA

G F

ITS

BE

TW

EE

N

BO

DY

AN

D C

OV

ER

BO

DY

AN

D C

OV

ER

WE

LD

ED

TO

GE

TH

ER

9.4

3.0

25.04.9 4.0

OP

EN

ING

S F

OR

TE

RM

INA

LC

ON

TA

CT

S

ME

TA

LT

ER

MIN

AL

IS 3

MM

WID

E.

TR

AC

KS

SH

OU

LD

BE

SP

AC

ED

NO

WID

ER

TH

AN

0.8

mm

AP

AR

T

CH

IPA

LIG

NE

D C

EN

TR

ALLY

IN

TH

E H

OR

IZO

NTA

LP

OS

ITIO

N

AN

DA

S C

LO

SE

TO

TH

IST

OP

LIN

EA

S P

OS

SIB

LE

TE

RM

INA

LS

41

.8

13

.2

14

.8

26.85.6

8.0

8.0

26.8

36.5

63.3

CO

VE

R

RF

SE

AL

MU

ST

FIT

WIT

HIN

TH

ES

E

DIM

EN

SIO

NS

. SE

EA

SS

ME

BLY

AB

OV

E F

OR

PO

SIT

ION

OF

RF

TE

RM

INA

LS

AN

D C

HIP

AA

TE

RM

INA

L

~2.4

SE

CT

ION

A-A

SC

ALE

2 : 1

TE

RM

INA

L

SE

AL

BO

DY

SE

AL

CO

VE

R

RF

TA

G

1.0

62.3

(a)

(b)

(c)

(d)

(e)

(f)(g

)(h

)(i)

BR

IDG

E

Fig

ure

5.1

.T

-sealstru

cture.

(a)is

the

insid

esu

rfaceview

ofth

ebody.

(b)

isth

esid

eview

ofth

ebody.

(c)is

the

outsid

esu

rfaceview

ofth

e

body.

The

vuln

erable

bridge

ism

arkedbetw

een(b

)an

d(c).

(d)

isth

epacked

figu

reof

the

T-seal.

(e)an

d(f)

areth

eou

tside

and

insid

e

surface

viewof

the

cover.(g)

and

(h)

show

the

termin

alsan

dth

eirattach

edlo

cation.

(i)is

the

2:1scaled

side

viewof

sectionA

-A

show

nin

(h).

The

concep

tof

insid

ean

dou

tside

isdistin

guish

ed:

the

insid

esu

rfaceis

defi

edas

the

surface

which

isagain

stth

eRF

tag

and

the

other

plan

arsu

rfaceof

the

same

itemis

defi

ned

asou

tside

surface.

Page 88


chamber, the tag antenna is completed by the two terminals and the tag can be de-

tected by the interrogator. This is the so-called “on” operation. However, the RFID

tag is powered by RF energy rather than DC power supply, so that this switch off and

on operation is more difficult to achieve than a light switch at home. The gaps on

the RFID tag antenna formed by the absence of the terminals may not sufficiently dis-

connect the RF coupling power, if the tag antenna and its associated structure is not

carefully designed to do so. This is the main challenge of this project.

After the attachment of the T-seal into the chamber, the sealed container is always

under the supervision of the reader. If somebody attempts to open the container sealed

by the T-seal, the only vulnerable component is the bridge between the up-wide part

and the down-narrow part of the T-seal, which is shown in Figures 5.1(b), (c). As we

can see, it is deliberately designed thinner than the other components which makes the

bridge vulnerable. Therefore an intruder will, in breaking the seal, break the bridge.

Since the chip is placed to sit on the bridge, the breaking action leads to a disconnection

between the tag antenna and the chip and provides the second “off” operation. This

operation results in an alarm or record of the intrusion.

The following designs and tests are based on the Australian standards and regulations

for passive UHF RFID systems, which are noted in Section 2.4. In addition, some con-

tainers usually are composed of metals. The metal could affect the performance of the

tag, if the tag is placed in close proximity to the container. In this design, the T-seal is

assumed to be deployed sufficiently far away from the metal component of a container

so that the metal parts do not affect, or does positively affect, the performance of the

tag antenna.

5.3 Chip and Antenna Selection

Before the introduction of the design processes, the chip and the desired antenna pat-

tern should be selected so that for a given interrogator power density, maximum power

is transferred to the chip for its excitation. Only when the chip is selected can the tag

antenna’s input impedance be designed to conjugate match the chip impedance. In

Page 89

5.3 Chip and Antenna Selection

terms of the antenna selection, there are plenty of antenna pattern candidates. How-

ever, because of the special requirement of this particular T-seal, the antenna pattern

has to be carefully considered.

5.3.1 Chip Selection

The chip Higgs-2 manufactured by Alien Technology with parallel resistance and ca-

pacitance of 1.5kΩ and 1.2pF is used again here. As indicated in Chapter 4, the chip

conforms to the EPCglobal Class 1 Gen 2 specifications. It is implemented in a CMOS

process and uses EEPROM memory. Usually, the input impedance of a tag antenna

is presented in series. Hence, in order to simplify the analysis, the chip impedance

should be transformed into a series representation. According to the calculation in

Subsection 3.8.2, at 923MHz which is the centre frequency of the UHF RFID band in

Australia, the chip impedance in series is about 13.6-j142Ω. Typically the threshold

power of this chip is 40µW (about -14dBm). More details of the chip can be found

in [55].

The chip is installed on a tag antenna by electrically conductive adhesive transfer tape

manufactured by 3M (Model 9703) [56]. Currents can pass perpendicularly through the

sticky tape. The tape and adhesive material on it will bring losses and chip impedance

changes. However, previous experiences with this tape reported by other colleagues in

our laboratory indicated that these losses and impedance changes are negligible [8]. It

has to be noted that there is a life span to the optimum performance of the conductive

tape. The recommended shelf-life of the tape is 24 months from the date the tape is

manufactured. Hence, it is good when used for tag prototypes for short term testing

and measurement, but is not recommended for long term commercial purposes.

5.3.2 Antenna Selection

According to the previous discussion, there are three obstacles in designing the tag

antenna in the T-seal: (i) the tag antenna’s shape has many limitations as shown in

Figure 5.1(e) and the chip’s position has been identified, (ii) the tag has to be initially

Page 90


in the “off” state before it is plugged into the chamber, (iii) once the tag is plugged into

the chamber, it should maintain good radiation characteristics.

Because of these obstacles in designing the tag antenna in the T-seal, the meander line

dipole antenna (MDA) is selected to be the candidate antenna. A simple example of an

MDA with two meanders is shown in Figure 5.2. The MDA possesses a size reduction

property, resulting from the added meanders, since they provide more inductance than

a straight wire in the same horizontal length. By loading more or larger meanders

on a dipole, the antenna can obtain more size reduction at the expense of radiation

efficiency. Moreover, the meander’s height h and width w can be adjusted as needed,

which ensures flexibility for the antenna shape. The details of the analysis and design

of meander line dipole antenna (MDA) can be found in Chapter 4. The MDA has been

commonly and commercially used as a transponder antenna such as in the Alien ALN-

9540 [67] and ALN-9562 [68] tags. However these commercial MDA RFID tags neither

fit into the T-seal tag nor satisfy the security requirement, so that the tag antenna in the

T-seal is designed as described in Section 5.4. The antenna is intended to be fabricated

on a 1.6mm FR4 board. The dielectric constant of the material in the board is 4.41.

Meander line

Dipole

w

h

Figure 5.2. A regular sample of MDA with two meander lines.

5.4 The Security Tag Antenna Design

As noted before in Section 5.2, the T-seal requires an off-on-off operation to achieve

the security function. The “on-off” operation, also called “turning off” operation can

be easily achieved just by placing the chip on the vulnerable bridge and designing an

MDA which can be fit in the T-seal. However, the “off-on” operation, also called “turn-

ing on” operation, is the main challenge of this design, because residual RF power may1The reason why this FR4 board (thickness 1.6mm and dielectric constant 4.4) is used here, because

of its immediate availability in out laboratory. The tag is preferable to be fabricated on a 0.2mm board.

In the case of the thinner board, the antenna on the board should be redesigned correspondingly.

Page 91


reach the chip and turn the tag on when it should be off. Therefore, in order to sim-

plify the whole design, this section is divided into two parts. First, a semi finished tag

antenna is designed just for the “turning off” operation. Secondly, based on the under-

standing of the semi finished tag antenna, a final tag antenna is designed to accomplish

both “turning on” and “turning off” operations.

5.4.1 Semi Finished Tag Design

A regular MDA with “turning off” function is designed to fit in the T-seal and be res-

onant at 923MHz. This tag is so called the semi finished tag. This semi finished tag

design is divided into three parts: Firstly, a theoretical analysis based on the material

in Chapter 4 is made. Secondly, the semi finished tag antenna designed according to

that theory is modeled in the HFSS simulation software. Some modifications to the

design are made in the light of the simulation results. Finally, a sample tag is made

and tested.

(a) Theoretical analysis

According to the analysis in Chapter 4, (4.12) is derived to estimate MDA’s shape for

RFID applications at a known resonant frequency. That frequency is 923MHz here.

(4.12) is repeated as follows.

µ0λ′

4π(log

4λ′

a− 1) + La =

µ0s2π

(log8sa− 1)

+µ0h1

π[1 +

13(β

′h1)2] log

4w1

a+ · · ·+

µ0hn

π[1 +

13(β

′hn)2] log

4wn

an = (1, 2, 3, · · · , n)

where λ′

= λεre f f

(the method for calculating relative effective permittivity of MDA

on dielectric substrate has been given in Chapter 4), β′

= 2πλ′ , hn and wn are the nth

meander line’s height and width respectively. The meander line height and width are

measured between mid-lines of the strips, La = Xa/ω is the inductance brought by the

extra needed inductive reactance Xa for canceling the connected chip capacitance.

Page 92


As noted before, in order to satisfy the “turning off” operation, the chip has to be in-

stalled on the vulnerable bridge. According to this special requirement and (4.12), the

shape of MDA fabricated on FR4 board and resonant at 923MHz, is approximately es-

timated and shown in Figure 5.3(a). The brown strip is the antenna element composed

of copper. The gap in the middle is reserved for placing a chip where the vulnerable

bridge is. The yellow board is the substrate of the MDA and it can be fitted into the

T-seal.

41.8

26

.8

32

.4

3.7

2 1

15

3.7

14.8

Unit: mm

13.2

(a) Derived by Equation (4.12)

41.8

26

.8

32

.43.7

2 1

8

3.7

14.8

Unit: mm

13.2

(b) Derived by simulation

Figure 5.3. The semi finished tag design.

(b) Comparison with simulation results

The semi-finished tag shown in Figure 5.3(a) is built in HFSS. By means of several HFSS

simulations, a practical semi-finished tag is derived and shown in Figure 5.3(b). It

provides a 24+j136Ω input impedance at 923MHz, which can approximately conjugate

match the impedance of the chip. The comparison between Figures 5.3(a) and 5.3(b)

demonstrates again that (4.12) can predict the shape of MDA to some extent. Moreover,

the gain pattern of this antenna in the yz plane and in the form of a three dimensional

polar plot are also obtained by the simulation, and are shown in Figure 5.4. The MDA

Page 93


lies in the xy plane. The maximum gain is a little less than that for a lossless dipole

because of losses in the antenna.

z

y

q

(b)

0.20.4

0.60.81

30

210

60

240

90 270

120

300

150

330

(a)

Figure 5.4. Simulated gain pattern of the semi finished tag antenna by HFSS. Gain pattern

was plotted in the form of: (a) three dimensional polar plot (The reference direction for

the “Theta”, θ is the z-axis and the reference direction for the “Phi”, φ coordinate is

the x-axis) and (b) yz-plane, where φ = 900, (the antenna lies on the xy plane in the

same coordinate).

(c) Test on the semi finished tag

After the theoretical analysis and simulation, a sample of the tag shown in Figure 5.3(b)

is fabricated and is shown in Figure 5.5.

As noted before, this tag is fabricated on an FR4 board which thickness is 1.6mm and

the dielectric constant of the material in the board is 4.4. The chip is connected to

the open terminals of the antenna by the method introduced in Subsection 5.3.1. The

board is slightly different from the model shown in Figure 5.3(b). First, the board in

Figure 5.5 has a one millimeter margin outside of the antenna which does not exist in

Figure 5.3(b). This is required by the fabrication process of the workshop. Secondly,

this board in Figure 5.5 does not have a nib at the bottom. It is believed that these two

slight differences can be negligible to the antenna’s performance. The board can be

reshaped when it is going to be manufactured. These small differences are shared by

the following fabricated tags as well, and will not be noted again. A reading range test

Page 94


Figure 5.5. A fabricated sample of semi finished tag.

of this tag has been performed and the details of this test can be found in Appendix A.

Here, only the reading range of this tag which is 3695mm is given.

5.4.2 Completely Finished Tag Design

The analysis in Subsection 5.4.1 aims to understand the resonant property of MDA on

FR4 board and accomplish the “turning off” function. However, the antenna design

in Figure 5.3(b) only achieves the “turning off” operation rather than achieving both

“turning on” and “turning off” operations. To accomplish the full security function in

one design, it is natural to consider adding a simple linking-loop in the lower part of

the board which can link the main antenna element and the terminals.

An approximate model is shown in Figure 5.6. The absence and presence of the ter-

minals lead to the disconnection and connection at the bottom of the loop, which are

presented by the grey areas in Figure 5.6. The tag antenna in these two conditions will

have different performances in input impedance and gain pattern, which may serve as

“turning on” action.

A simulation is made on both semi finished tag with a complete loop and with an

incomplete loop. The width of the loop Wl is 10mm in the simulation. The gain patterns

Page 95


41.8

26

.8

32

.4

3.7

14.8

21

Unit: mm

7W

l

26

Figure 5.6. Semi finished tag with a loop in the down-narrow part of the board. The loop’s

length is mandatory to be 26mm to link the terminals. Wl is the width of this loop.

The grey areas on the loop are the position of the terminals.

of the two situations are shown in Figures 5.7 and 5.8 respectively. It may be noted

there is not much difference.

0.20.4

0.60.81

30

210

60

240

90 270

120

300

150

330

z

y

q

(b)(a)

Figure 5.7. Simulated gain pattern of the semi finished tag antenna with a complete loop.

Gain pattern was plotted in the form of: (a) three dimensional polar plot (The reference

direction for the “Theta”, θ is the z-axis and the reference direction for the “Phi”, φ

coordinate is the x-axis), and (b) yz-plane, where φ = 900, (the antenna lies on the xy

plane in the same coordinate).

Page 96


0.20.4

0.60.8

1

30

210

60

240

90 270

120

300

150

330

(a)

z

y

q

(b)

Figure 5.8. Simulated gain pattern of the semi finished tag antenna with a incomplete loop.

Gain pattern was plotted in the form of: (a) three dimensional polar plot (The reference

direction for the “Theta”, θ is the z-axis and the reference direction for the “Phi”, φ

coordinate is the x-axis), and (b) yz-plane, where φ = 900, (the antenna lies on the xy

plane in the same coordinate).

The input impedance of the semi finished tag antenna with a complete loop at 923MHz

is 12+j118Ω. The input impedance of the other one is 62+j223Ω. This difference is

responsible for the modest change in reading range noted later in this section.

To evaluate the operating range of the tags, there are mainly three methods: (i) by the

Friis equation introduced in Section 3.7, (ii) by a scattering matrix given in Section 3.8

and (iii) by experiments. At the early stage, we would use the first two methods to

estimate the operating range to avoid the tag fabrication.

• The Friis equation method

In Section 3.7, the equation of evaluating the operating range of a UHF RFID

systems has been given by (3.32) which is rewritten as follows.

Pchipr = PEIRP(1− |θ|2)gtag 1

plep (5.1)

where

θ =Zchip−Z∗tantZchip+Ztant

, Zchip and Ztant are the impedances of the chip and the tag antenna

respectively,

Page 97


Pchipr is the power received by the chip,

PEIRP is the transmitted power from the reader antenna in terms of EIRP,

gtag is the gain of the tag antenna,

1pl is the path gain which is assumed to be ( λ

4πr )2, λ is the free space wavelength

at the resonant frequency,

ep is the polarisation efficiency which is assumed to be 1 by using a linearly po-

larised reader antenna.

First, substituting the input impedance of the tag antenna derived by the simu-

lation and and the chip impedance into the expression of θ =Zchip−Z∗tantZchip+Ztant

to obtain

the θ factor, then substituting Pchipr = 40µW (about -14dBm) which is the thresh-

old power of the chip Higgs-2 used here, PEIRP = 4W which is the maximum

power allowed to be radiated from the reader antenna in Australia, 1pl = ( λ

4πr )2,

the θ factor derived previously and the value of gtag derived from the simulation

into (5.1), the reading range r can be calculated.

In particular, the reading range of the semi finished tag antenna with a complete

loop is about 5.6m and the reading range of the semi finished tag antenna with

an incomplete loop is about 2.6m.

• The scattering matrix method

In Section 3.8, a novel method for evaluating the operating range of a tag is pro-

posed by making use of a scattering matrix. The scattering matrix can be derived

by the simulation software Ansoft HFSS after building the tag antenna and reader

antenna model. We would not use this method in evaluating the reading range of

the two tags, since in this situation, (a) the design obviously does not satisfy the

requirement of the security tag, (b) the reading range of the both tags evaluated

by the Friis equation are over 2m. By setting the tag antenna and the reader an-

tenna at a distance over 2m, the simulation is memory expensive and encounters

the limitation of the computer used by the author of this thesis. But this method

will be used in the later analysis when the reading range is relatively small.

Page 98


Obviously, because the reading range of the semi finished tag antenna with an incom-

plete loop is about 2.6m, it is far away from the ultimate goal of initial “off” function.

This design approach was thus abandoned.

Because of the failure of the added loop scheme, another approach shown in Fig-

ure 5.9(a) was investigated. Here the linking loop is replaced by two separated and

symmetrical meander lines in the down-narrow part of the board. With this approach

it is possible to accomplish the “turning on” operation, because when the terminals at

the bottom are not connected to the tag antenna, all that is left connected to the chip are

two close vertically parallel strips and the direction of current on the two strips is op-

posite, and thus does not contribute to the radiation but still brings losses. Hence, the

gain is degraded dramatically. Moreover, the broken link leads to a serious impedance

mismatch, which worsens the power transmission to the chip. Since the sketch in Fig-

ure 5.9(a) is also one MDA on board, the theoretical interpretation in Subsection 5.4.1

allows us to estimate the antenna dimension to some extent, and then HFSS is utilised

to make an accurate model. The process of the theoretical analysis is not repeated here.

Only the accurate model obtained by HFSS is shown in Figure 5.9(b). Then, simula-

tions were conducted on the two models based on the antenna in Figure 5.9(b). One

imitates the T-seal when it is attached into the chamber which means the two vertical

copper elements from the chip are connected to the rest of the meanders according to

the attachment of the two terminals. This tag is named as Tag1. The other one is for

when the meanders are broken in the grey areas at the bottom by the absence of the

two terminals. This tag is named as Tag2. The gain patterns of Tag1 and Tag2 are shown

in Figure 5.10 and Figure 5.11 respectively. The input impedances of Tag1 and Tag2 are

16.2+j136Ω and 3.3-j149Ω respectively. As described above, three methods: (i) the Friis

equation method, (ii) the scattering matrix method and (iii) the experimental method

can be used to estimate the reading range of a tag which are discussed as follows.

• The Friis equation method

By substituting the gain and input impedance derived from the simulation into

the Friis equation (5.1) as discussed, the reading range of Tag1 and Tag2 can be

derived which are 7m and 70mm. However, we suppose that the Friis equation

Page 99


41.826.8

33.5

14.8

18

13.2

(a)

41.8

26

.8

33

.5

3.7

14.8

2 1

Unit: mm

5

27

16

13.2

(b)

Figure 5.9. The final design of the security tag. The white gap in the middle is reserved for

placing a chip where the vulnerable bridge is. The terminals will be attached in the grey

areas.

(b)(a)

z

y

q

0.20.4

0.60.8

30

210

60

240

90 270

120

300

150

330

Figure 5.10. Simulated gain pattern of Tag1 . Gain pattern was plotted in the form of: (a) three

dimensional polar plot (The reference direction for the “Theta”, θ is the z-axis and

the reference direction for the “Phi”, φ coordinate is the x-axis), and (b) yz-plane,

where φ = 900, (the antenna lies on the xy plane in the same coordinate).

Page 100


0.020.04

0.060.08

30

210

60

240

90 270

120

300

150

330

(b)(a)

z

y

q

Figure 5.11. Simulated gain pattern of Tag2 . Gain pattern was plotted in the form of: (a) three

dimensional polar plot (The reference direction for the “Theta”, θ is the z-axis and

the reference direction for the “Phi”, φ coordinate is the x-axis), and (b) yz-plane,

where φ = 900, (the antenna lies on the xy plane in the same coordinate).

is not suitable to analyse Tag2 since by using the Friis equation, the distance be-

tween the two communicating antennas should be larger than 2D2/λ, where D

is the largest dimension of either antenna, which is not satisfied by the situation

of Tag2. That is to make Tag2 work, it would have to be brought so close to the

reader that the condition on the separation would be violated. Hence the scat-

tering matrix method is made use of to analyse the reading range of Tag2, and is

discussed in the next itemisation.

• The scattering matrix method

In Section 3.8, a novel method for evaluating the operating range of a tag is pro-

posed by making use of a scattering matrix. The scattering matrix can be de-

rived by the simulation software Ansoft HFSS after building the tag antenna and

reader antenna model. In this case, we build Tag2 and a linearly polarised patch

antenna with gain 8dBi as a reader antenna into the simulation. By varying the

distance between the tag antenna and the reader antenna and accomplishing the

simulation, the power received by the chip of Tag2 can be obtained by (3.60). Ac-

cording to the scattering matrix method, the conclusion is drawn that no matter

Page 101

5.5 Conclusion

how close Tag2 is to the reader antenna, the power received by it cannot be larger

than 40µW (-14dBm) which is the threshold power to excite the chip.

• The experimental method

Tests on Tag1 and Tag2 have been done which details can be found in Appendix A.

The reading range of Tag1 is 580mm and Tag2 cannot be read. Therefore, the final

security tag design can satisfy the off-on-off operation. Apparently, the reading

range of Tag1 is much smaller than that evaluated by the Friis equation. That

may be caused by the following two factors: firstly, the reader antenna used in

the test is a circularly polarised reader antenna which brings in a polarisation

loss, secondly and most significantly, the attachment of the two terminals may

bring some impedance mismatch between the chip and the tag antenna which is

not considered in the simulation.

5.5 Conclusion

A novel tag antenna has been successfully designed to be embedded into the mechan-

ical layout of the T-seal. The carefully designed RFID tag together with the mechanical

layout can serve as an electronic seal to determine whether the security of a container

has been compromised. Although the tag shown here is fabricated on a 1.6mm thick

FR4 board, which may not be the ideal scheme for mass production and too thick to be

inserted into the T-seal, it still demonstrates the feasibility of the T-seal concept. Exam-

ination of the effects of using different and thinner substrates could be undertaken in

future work.

Page 102

Chapter 6

Solutions for the Antennaon Metal Problem

ONE of the problems faced by UHF RFID systems is that

the reading of RFID tags placed on products incorporating

metal is difficult. This difficulty is known as the antenna on

metal problem. In this chapter, the fundamental reasons for this problem

are explained, the analysis of antenna parameters and measurements of the

reading ranges of some commercial tags in proximity to metal are given.

The existing solutions and their limitations to the antenna on metal prob-

lem in UHF RFID applications are summarised.

Page 103

6.1 Introduction and Outline


The practical world is not uniform. It contains hundreds of thousands of materials

with different electromagnetic characteristics. Hence, electromagnetic boundary con-

ditions in various forms should be considered. One of the major concerns of this thesis

is to provide a solution to the problem of poor tag performance when a tag antenna is

placed on or near metal. In Section 6.2, firstly the metallic boundary conditions and the

method of images are used to give the reasons for the problem. Then existing research

on antenna parameters in proximity to metal is described to illustrate how metal affects

antennas in this situation. Following this, some measurements on the reading ranges

of various commercial tags on metal are reported. In Section 6.3, some previous solu-

tions to this problem are described. These include the one quarter wavelength isolator

solution, antenna selection solutions and some artificial magnetic conductor (AMC)

solutions. Finally, in Section 6.4, the limitations of the existing solutions in UHF RFID

applications are summarised. The conclusion is drawn that in UHF RFID applications,

a simple structure, low profile, low cost and compact size solution to the antenna on

metal problem is still needed.

6.2 The Antenna on Metal Problem

6.2.1 Metallic Boundary Conditions

The general boundary conditions at the interface of two different materials according

to Maxwell’s equations are:

n× (E2 − E1) = 0 (6.1)

n · (D2 −D1) = ρs (6.2)

n× (H2 −H1) = Js (6.3)

n · (B2 − B1) = 0 (6.4)

where n is the unit vector which is perpendicular to the interface formed by the two

media and is directed from medium 1 to medium 2. E and H are electric and magnetic

Page 104

Chapter 6 Solutions for the Antenna on Metal Problem

+

+

+

+

+

+

+

+

+

H

E

nJ

s

rs

PerfectConductor Free space

Figure 6.1. Boundary conditions at a perfect conductor surface. The surface current points

out the paper and the positive symbols denote surface charge density ρs

field respectively. D and B are electric and magnetic flux densities respectively. ρs

and Js are the surface charge density and surface current density respectively. The

subscripts 1 and 2 are used to indicate two adjacent media. In linear media, the media

may be characterised differently in dielectric permittivity ε, magnetic permeability µ

and electric conductivity σ.

When medium 1 is assumed to be a perfect electric conductor, the electric field in this

medium becomes zero (E1 = 0). As a result, in a linear medium, D1 = 0, H1 = 0 and

B1 = 0. The statements about H1 and B1 assume a non-zero frequency.

The metallic boundary conditions are then found to be:

n× E2 = 0 (6.5)

n ·D2 = ρs (6.6)

n×H2 = Js (6.7)

n ·B2 = 0 (6.8)

Apparently, a conclusion can be drawn that there is a normal component of electric

field and there are tangential components of magnetic field on the boundary. A simpli-

fied illustration of electric fields and magnetic fields in the interface between free space

and perfect conductor is shown in Figure 6.1.

Additionally, we will illustrate the metallic boundary conditions from another per-

spective, making use of the method of images. This method provides a direct way to

analyse the effect of a perfect conductor on an antenna.

Page 105


+ q

- q

+q

Perfect Conductor

(a) (b)

D

D

Figure 6.2. Electric field when a positive charge is put above the perfect conductor. (a)

Single positive charge, (b) positive charge and its image.

When a single positive charge is above a perfect conductor at a distance D, there is

as shown in Figure 6.2(a) no tangential electric field just above the conductor. If the

perfect conductor is absent, in order to obtain the same electric field distribution on

the original plane where the conductor used to be, a negative charge should be placed

underneath the plane at an equal distance D. The two charges have a symmetry. To

sum up, the ground plane acts as a mirror and we get an image below it.

If a straight wire, carrying current, is placed vertically to a perfect conductor, the image

of this wire will enhance the radiation, as is shown in Figure 6.3(a). However, one

thing should be noted that a vertical straight wire above a perfect conductor produces

the same field as a dipole in free space. If the dipole is made of a resonant length λ/2,

the straight wire would be resonant at a length of λ/4.

However, when the wire is parallel to a perfect conductor surface as shown in Fig-

ure 6.3(b), the direction of the current in the image is opposite to that of the original

current and if the original wire is close to the conductor, the current will be cancelled

and there is no field radiated outwards.

Therefore, when a dipole antenna, or an antenna design based on dipole antennas, e.g.

meander line dipole antenna, folded dipole antenna, is placed parallel to the surface of

a perfect conductor and is close to it, the antenna cannot radiate because of the current

cancellation between the antenna and its image.

Page 106


(a) (b)

Figure 6.3. A straight wire carrying current and its image underneath the ground plane.

(a) Vertical placement (b) Parallel placement. The solid arrow represents the subject

current and the dashed arrow represents the image current.

6.2.2 Antenna Parameters in Proximity to Metal

Most tag antenna designs are based on dipole antennas, so the performance of these

RFID tag antennas may be degraded because of the metallic boundary conditions, as

described previously.

Current research has shown that this degradation is caused by the three factors listed

below [49] [50] [69].

1. Variation in impedance

The input impedance of an antenna varies with different distances between the

antenna and a metal plate. If the antenna is a receiving antenna, the impedance

mismatch impedes the optimum power transfer from the antenna to the load

as described in Subsection 3.2.1. This analysis has been proven experimentally

by previously published research [49] [50]. There the reading range and input

impedance of several commercial tags were measured at varies distance above

a metal plate [49]. None of them could be read in near proximity (< 2mm) to

a metal plate [49]. The measurements also demonstrate that the power transfer

efficiency to the load is degraded since the input impedance of the tag is varied

when it is placed closely to the metal plate. The variation in impedance is found

to be dependent on the antenna geometry. Another result disclosed by Prothro

et al. [50] also supports this conclusion. In their work [50], two folded dipole

antennas are fabricated either by using a narrow silver paste strip or by using of

a broad silver paste strip. It was found that both the real and imaginary parts of

Page 107


the input impedance of the narrow strip one are less sensitive to the presence of

neighboring metals than those of the broad strip antenna. However, this advan-

tage is derived at the expense of the antenna gain. This reduction in antenna gain

is attributed to the higher loss resistance of the narrow strip.

2. Radiation pattern

The radiation pattern should also be considered when a tag is placed near to

metal. Raumonen et al. [69] stated that the presence of the metal plate splits the

main lobe of a folded dipole’s radiation pattern into many sharp, narrow, side

lobes [69] and with the decrease of the distance between the metal plate and the

antenna, the number of the side lobes declines. The changes in radiation pattern

definitely affect the reading range of a tag. For example, most forms of dipole

antenna are omnidirectional in the equatorial plane. However, when they are

deployed in proximity to metal, the radiation pattern changes, so that tags cannot

be operated in a certain direction.

3. Lack of tangential electric field

Last but not least, as required by the metal boundary conditions introduced in

Subsection 6.2.1, there is only a normal component of electric field. However,

most of the RFID tag antenna designs are based on dipole antennas, which are

excited by tangential electric field. If these antennas lie on a metal plate or are

placed parallel to the plate at a small distance, they are not able to be exited

because of the lack of sufficient tangential electric field.

6.2.3 The Performance of Commercial Tags Above a Metal Plate

From the analysis in Subsection 6.2.1 and Subsection 6.2.2, it is known that the metal

can degrade the performance of some antennas when they are just placed above the

metal. In order to determine how effectively the metal can affect commercial tags,

the reading ranges of several commercial tags were measured in two different envi-

ronments: firstly, these commercial tags were placed in free space, and secondly, they

were placed in close proximity to a metal plate. This work mainly follows that reported

Page 108


in [8] and [49]. The methods and results of these experiments in the two environments

have been described in the following two itemisations respectively.

• Commercial tags in free space

Six passive UHF RFID commercial tags have been chosen for this experiment.

The make and models of these tags will not be disclosed for confidentiality rea-

sons. Some of the tag antenna designs, for example the meander line dipole an-

tenna (MDA) or the folded dipole antenna (FDA) are based on half-wavelength

dipole antenna. Some of the tag antenna designs are based on the simple bow-tie

antenna (BTA). One of the tag antennas is the combination of dipole antenna and

bow-tie antenna. In order to distinguish them, they are labelled with the num-

bers from “1” to “6”. The RFID reader (a refurbished early model ALR-9780-EA)

and 6 dBi gain circularly polarised reader antenna (Model ALR-9610-BC) both by

Alien Technology, operating over the somewhat limited Australian UHF RFID

band, were employed here to detect these commercial tags. The experiment was

operated under the Australian UHF RFID standards and regulations, which has

been introduced in Section 2.4. A shielding tunnel shown in Figure 6.4 was used

again to isolate this experiment from the outside environment. The details of the

absorbing foams inside the tunnel can be found in Subsection 3.8.2. When the

tags were put in this tunnel, it was assumed that they were effectively in free

space. The reader antenna is pointed into the tunnel.

The reading ranges of the labelled commercial tags in free space obtained by the

experiment and some information of those tags, such as their footprints, shape

patterns, are listed in Table 6.1.

• Commercial tags above a metal plate

This experiment tested the reading ranges of each selected commercial tag when

it is above a metal plate by various distances. The commercial tags, reader and

reader antenna were the same as in the previous experiment. A 650mm×500mm

aluminium plate was used. Tags “1” to “6” were placed above the plate one at

a time. To separate the aluminium plate from each commercial tag by various

Page 109



Table 6.1. Reading ranges of commercial tags in free space. In the pattern column, FDA means

folded dipole antenna, MDA denotes Meander line dipole antenna and BTA represents

bow-tie antenna. In protocol column C1G1 or C1G2 denotes Class 1, Generation 1 or

Class 1 Generation 2.

Commercial Tags Overall Size (mm) Pattern Reading range (m) Protocol

1 148× 10 FDA 1.54 C1G1

2 95× 8 MDA 0.98 C1G1

3 95× 9 FDA 1.50 C1G2

4 94× 28 BTA 1.40 C1G2

5 88× 25 BTA+Dipole 2.63 C1G2

6 95× 28 BTA 1.31 C1G1

distances, various numbers of cardboard sheets or a single bubble wrap were in-

serted between the plate and the tag. The thickness of each cardboard and the

bubble wrap are 0.6mm and 5.6mm respectively, so that the number of the card-

board sheets and the bubble wrap inserted can determine the distance between

the plate and the tag. Moreover, the permittivity of these cardboard sheets and

the bubble wrap was assumed (with perhaps some inaccuracy in the case of the

cardboard) to be that of air. The reading ranges of each commercial tag when it is

placed various distances above the metallic plate obtained in this experiment are

shown in Figure 6.5.

Page 110


0 1 2 3 4 5 60

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Rea

din

g r

ange

(m)

Tag 1

Tag 2

Tag 3

Tag 4

Tag 5

Tag 6

Distance between tag and metal plate (mm)

Figure 6.5. Reading ranges of labelled commercial tags when they are placed above the

aluminium plate at various distances.

As shown in Figure 6.5, all the commercial tags’ reading ranges are degraded dra-

matically in proximity to the aluminium plate, compared with the reading ranges

in free space given in Table 6.1. When the tags are moved further away from the

metal plate, the reading range is increased. Moreover, the tag antennas (Tags 1, 2,

3) based on the dipole antenna are more sensitive to the variation of the distance

away from the metal plate than the tag antennas (Tags 4, 6) based on the bow-tie

antenna. That is because the current on the dipole can be cancelled by the image

current underneath the metal and with the decrease of the distance between the

metal and the dipole antenna, the cancellation becomes more obvious. As a re-

sult, the dipole on metal cannot radiate effectively. However, for the tag antennas

derived from the bow-tie antenna, when they are placed on the aluminium plate,

the whole structure (the antenna with the underneath metal) serves as a poorly

radiating patch antenna, hence, it is found that tags “4” and “6” have longer read-

ing range than other tags, when the tags are put in close proximity to the plate

and the distance between tag and the plate is less than 3mm.

Page 111

6.3 Previous Solutions to the Problem


6.3.1 One Quarter Wavelength Isolator Solution

Inserting a quarter wavelength space between an antenna and metal is a common solu-

tion to address the antenna on metal problem. The reason for this is that when an elec-

tromagnetic wave is reflected by metal, one π phase change occurs due to the metallic

boundary condition, so that the tangential electric field of the incident wave at the

metal surface will be totally cancelled by the reflected wave, and an antenna which

depends on this tangential electric field will not couple well when it is put close to

the metal surface. If there is one quarter wavelength space between the antenna and

the metal, the incident wave travels a λ4 path from the antenna to the reflecting plane

(where it suffers a phase change) and travels the other λ4 path from the plane to the

antenna after reflection. In total, it is a λ2 path which provides another π phase change

to cancel the previous π phase change caused by the metal. Hence the electric field

of the traveling waves and reflecting waves meeting at the antenna is in phase rather

than out of phase. A model of this solution is shown in Figure 6.6. The solution only

works for illumination from a direction perpendicular to the metal sheet.

Dielectric Material

Metal

Antenna

l /4r

Figure 6.6. Side view of an antenna placed at one quarter wavelength distance above a metal

plate. The space between the antenna and metal is full filled by dielectric material. λr

is the wavelength in the dielectric material.

This method derives from work by Dallenbach and Kleinsteuber [4]. By placing a quar-

ter wavelength dielectric slab between the antenna and the metal surface, the thick-

ness of the material layer is less than one quarter wavelength in air at the operating

frequency. This method can work well without a dielectric material or with a low

dielectric constant material in the SHF (super high frequency) band (3GHz-30GHz),

Page 112


especially at the higher frequency, because the free space wavelength is then 10mm

(30GHz) and one quarter of that can be acceptably thin. Pinho et al. [70] reported a

1.5mm thick wave dielectric isolator fabricated to work in X band (8-12GHz) and Ku

band (12-18GHz). Additionally, Tanaka [5] reported a thickness of 1mm for another

one quarter wavelength dielectric isolator resonant at 50GHz.

However, as mentioned in Subsection 2.3.2, in the USA, Australia and Europe, the

UHF band centres are 915MHz, 923MHz and 866MHz respectively. If some low di-

electric constant materials are employed to be the substrate, one quarter wavelength is

about 50-75mm thick. Obviously this arrangement suffers from being bulky and eas-

ily damaged. High permittivity materials might then be introduced to obtain a low

profile. Kim et al. [6] made the isolator resonant at 2GHz by using high permittivity

materials such as BaTiO3 (BT). They achieved a thickness of less than 4mm. However,

manufacturing these high permittivity materials at a desired value of dielectric con-

stant is complicated. Numerous procedures such as controlling sintering temperature,

pressure, and the ratio of composite materials are involved. Moreover, a one quarter

wavelength isolator can also be made from ferrites because of their high permittivity;

however, these materials are accompanied by low radiation efficiency and significant

weight.

6.3.2 Antenna Selection Solutions

Not all types of antennas suffer from the negative effects brought by metal. Some

antennas, such as patch antennas and inverted F antennas (IFA), can utilise the ground

plane as part of the antenna element, and some other antennas such as loop antennas

depend on tangential magnetic field rather than tangential electric field. These types of

antennas above metal can work equally well or even better than they do in free space.

In [7], one patch antenna is designed to be an RFID tag antenna and a regular ground

plane is replaced by an electromagnetic band gap (EBG) ground plane in order to in-

crease the antenna gain and decrease the side lobe. A simple patch antenna designed

as a UHF tag antenna for the purpose of achieving balance among tag size, costs and

performance is reported in [8].

Page 113


The inverted F antennas (IFA) were originally three dimensional antennas, which means

the thickness of IFA is too large to be ignored. However, recently some coplanar IFA

have been reported [9], [10] to achieve low profile. In the coplanar IFA the radiat-

ing element is in the same plane as the ground plane but is off to one side. Tag an-

tennas designs based on the coplanar IFA have been proven to radiate effectively on

metal [9], [10].

As discussed in Subsection 6.2.1, while there is no tangential component of electric

field in proximity to metal, the magnetic field there is doubled. Hence, it is possible

to make use of this doubled field to excite a tag by putting a loop antenna in close

proximity to metal. A small loop tag antenna designed for metallic item identification

is reported in [11].

6.3.3 Artificial Magnetic Conductor Solutions

An artificial magnetic conductor (AMC) is usually composed of a periodically dupli-

cated metal surface, a dielectric layer and a ground plane. According to its name, it

does not exist in nature (artificial) and it performs as a magnetic conductor, which

means on its periodically duplicated metal surface, there are tangential electric fields

and no magnetic fields. An AMC performs just oppositely to an electric conductor.

Because of its special field distribution property, the reflection coefficient on an AMC’s

surface is +1 at a certain frequency and remains positive along a frequency band. In

addition, the surface impedance is very high, therefore, some researchers [12] [13] call

it high impedance surface.

Besides high impedance surface, various terminologies such as frequency selective sur-

faces (FSS) [71], photonic crystals [72], electromagnetic band gaps (EBG) [73] and pho-

tonic band gaps (PBG) [74] have been used for describing the AMC depending on the

domain of the applications and the way of understanding the structure. In some parts

of this thesis, an AMC is also called a decoupler.

Because the tangential electric field exists on the surface of an AMC, antenna designs

based on dipole antennas can radiate well if placed above an AMC. Therefore an AMC

Page 114


(a)

(b)

Figure 6.7. Sievenpiper high impedance electromagnetic surface. (a) Cross-section of the high

impedance electromagnetic surface. (b) Top view of the high impedance electromag-

netic surface.

which is put between the attached metal object and an tag antenna, can isolate the

tag antenna from the metal and the tag antenna on metal problem can be solved by

this placement. Two existing AMCs, not suitable for RFID applications, are described

further in the following two subsubsections.

(a) Sievenpiper high impedance electromagnetic surface

Sievenpiper et al. [12] presented one kind of AMC to cope with the antenna on metal

problem, which is shown in Figure 6.7. As noted above, the inventor calls this AMC a

high impedance electromagnetic surface.

The AMC is composed of a ground plane, a dielectric layer and a metal surface con-

sisting of many discrete cells. All the cells are hexagonal metal patches of the same

size. A ground plane is at the bottom. Some vias are used to connect each patch on

the top to the bottom ground plane. Dielectric materials fill the space between these

two layers. Cells of this structure can be analysed as lumped-circuits as long as the

dimensions of each patch are small compared to the operating wavelength. Thus each

cell can be drawn as in Figure 6.8. Obviously, the loop formed by the two conductor

layers and the connectors can be taken as inductance. Meanwhile, the edges of two

surrounding patches form the capacitance. Therefore, this structure can be resonant

Page 115


at a specific angular frequency: ω0 = 1/√

LC, which is around 15GHz in [12] for the

dimensions quoted there and a relative dielectric constant of 2.2. As noted above, if the

AMC is operated at the resonant frequency, a high impedance surface is created near

the top. The resonant frequency can be adjusted by changing the physical dimensions

of the cells and the dielectric constant.

A three conductive layer structure, shown in Figure 6.9, was also described by Sieven-

piper [12], in order to enable the AMC to work at lower frequency (2.2GHz-2.5GHz).

This kind of AMC may be made very thin (less than several millimeters). However,

the structure is complicated, especially when it is resonant in the UHF RFID band.

+ -C

L

Figure 6.8. Origin of the capacitance and inductance in each cell.

(b) Space-filling curve AMC

The first space-filling curve was presented in 1890 by an Italian mathematician Giuseppe

Peano and one year later another similar curve was introduced by German mathemati-

cian named David Hilbert [75]. With the increasing of the iteration orders, the Hilbert

curve begins to occupy the two dimension square space which has been shown in Fig-

ure 6.10.

These curves with appropriate feed points can be used as antenna structures due to

their special self-duplication and resonant properties [76]. This property allows an-

tennas with large numbers of radiating elements to be made within an available area.

Layer 2Layer 2 LayerLayer 1

Layer 3

Figure 6.9. Three conductive layer high impedance electromagnetic surface.

Page 116


order1 order2 order3 order4

Figure 6.10. Hilbert curve in various orders.

Vinoy et al. [77] reported that the resonant frequency of an antenna composed of a

Hilbert curve would decrease as long as the iteration orders increase in the same unit

area.

In an alternative use of the Hilbert curve, a dielectric layer sandwiched between a

ground plane and a Hilbert curve is reported to be another kind of AMC [13], and is

shown in Figure 6.11. Analysis of this structure has been investigated by the IE3D code

that uses the Method of Moments with periodic boundary conditions. At its design

frequency, this structure can reflect the illuminating wave in phase instead of out of

phase, and its operating frequency can be controlled by setting the number of orders

in a unit area. Similarly, the Peano curve also obtains the same radiation characteristic

if the iterations are set appropriately [78].

Six parameters govern the behaviour of the structure. They are:

1. d, the length of the smallest element in the pattern.

2. N, the order of iterations in the pattern.

3. L, which is both the overall width and overall length of the pattern.

4. Height, the thickness of the space between the pattern and the ground plane.

5. Spacing, the separation distance between one instance of the Hilbert curve of a

give order and another instance obtained by translating the curve in either the x

or y direction.

6. εr, the relative dielectric permittivity of the material between the Hilbert and

the ground plane. It is always assumed by McVay et al. [13] to be unity, but

Page 117

6.4 Conclusion

those authors have stated that if one were used the resonant frequencies would

be lowered.

Dimensions d, L, Height, and Spacing are marked in Figure 6.11. A PEC wire of radius

0.01mm is used in the modeling by McVay et al. [13]. The footprint of the curve is

1.2mm by 1.2mm.

A definition of centre frequency is introduced as the frequency for which the reflection

coefficient is 1, and a definition of bandwidth is introduced as the frequency range

for which the phase of the reflection coefficient lies between plus and minus ninety

degrees.

According to the simulation results in [13], the six parameters just discussed affect the

centre frequency and bandwidth differently. In detail, the variation of spacing has

only minor effect on the behavior of the structure. The function of dielectric constant

is easily understood. The most significant factors are N, the order of iterations in the

pattern; L, which is both the overall width and overall length of the pattern (d, the

length of the smallest element in the pattern, is fixed once the above two factors are

given); and height, the thickness of the space between the pattern and the ground

plane. The increase of those three parameters (N, L and height) will be beneficial to

the performance of the AMC in terms of obtaining low centre frequency and wide

bandwidth. However, that will also increase the structure size or its complexity.

So we may ask what are the prospects for obtaining a low centre frequency with con-

venient parameters? After studying the variation of centre frequency and bandwidth

with respect to available parameters it may be concluded that obtaining good opera-

tions in the UHF RFID band with only a small height above the ground plane is infea-

sible.

6.4 Conclusion

As described above, a large number of solutions to the antenna on metal problem have

been found. However, they all have their own weaknesses. For example, the one quar-

ter wavelength isolator suffers from being bulky or being of high cost in fabricating

Page 118


y

x

z

d

Height

Spacing

L

Figure 6.11. Hilbert curve AMC based on order 4 Hilbert curves. The order 4 curves are

replicated, without interconnection, twice in the x direction and twice in the y direction.

high relative permittivity material. When employing the antenna selection solution,

very careful antenna design is needed for each particular application to achieve the

maximum power transmission. The Sievenpiper AMC structure is complicated and it

is also very difficult to lower the resonant frequency to the RFID UHF band of around

1GHz. The space-filling curve AMCs suffer from having a complex structure and po-

tential high cost to manufacture. In addition, it does not appear to be feasible to lower

the operating frequency to the UHF RFID band while maintaining a thin structure.

In the light of the above shortcomings we are led to consider yet another promising

structure, that of the slitted decoupler, which is the subject of the next chapter.

Page 119

Chapter 7

The Slitted DecouplerDesign for Metallic Item

Detection

THE slitted decoupler design for metallic item detection in UHF

RFID systems is introduced in this chapter. The slitted decou-

pler possesses some significant advantages in dealing with the

antenna on metal problem described in Chapter 6. Theoretical analysis and

simulation results for the slitted decoupler are presented, comprehensive

operational principle of this structure is given. Design principles of the slit-

ted decoupler are proposed for the size minimisation and performance op-

timisation. Experiments are conducted to validate the analysis and design

principles.

Page 121



As discussed in Chapter 6, the previous solutions dealing with the antenna on metal

problem within the UHF RFID band (860MHz-960MHz) suffer from their own weak-

nesses in terms of thickness, complexity of structure or university for most applica-

tions. Hence, the solution with low profile, simple structure and superior universality

for most RFID applications is desirable.

The slitted decoupler discussed in this chapter is regarded as a potential solution with

all above described merits. The slitted decoupler is also one kind of artificial magnetic

conductor (AMC) (we will give the reason in the following discussion). Therefore, it is

inserted between a tag and a metallic item. The original idea of the slitted decoupler,

defined in the next section, comes from the patent “Electromagnetic radiation decou-

pler” [15]. In this patent, many results showing the performance of one tag on a slitted

decoupler have been given. Some simple design principles in terms of the physical di-

mensions and material selection of the decoupler were proposed. According to these

results and design principles in the patent, the slitted decoupler is found to have a

strong team of talents in thickness and simplicity. The thickness of itself together with

a tag is less than 2mm and the structure is extremely simple which the reader will find

out in Section 7.2. However, there is in this patent no explanation of these results. Fur-

thermore, the disadvantage of the decoupler proposed in [15] is also obvious that its

size is too large particularly its large width. The interaction occurring between the tag

antenna and the slitted decoupler underneath is not carefully considered and the ef-

fects brought by the variation of the size of the attached metallic item to the resonance

of the slitted decoupler is absent.

To remedy these shortcomings, this chapter provides a quantified and comprehensive

analysis of the slitted decoupler and its interaction with the tag above and the metal-

lic item underneath by the electromagnetic method, and in the light of that analysis

the size of the slitted decoupler is meant to be reduced meanwhile its performance is

enhanced.

Following this introduction, Sections 7.2, 7.3 and 7.4 give the brief introduction to the

slitted decoupler proposed in [15] for two purposes: (i) making the basic functionality

Page 122

Chapter 7 The Slitted Decoupler Design for Metallic Item Detection

of the slitted decoupler understandable; (ii) setting a reference for comparing with the

work done by the author of the thesis in other sections in this chapter. In detail, Sec-

tion 7.2 provides a definition of the structure of the slitted decoupler and establishes

pertinent dimensions. Some design principles proposed in [15] for the slitted decou-

pler are listed in Section 7.3. A description of the simulation methods that will be used

in study of the decoupler is provided in Section 7.4. Then a sample of the slitted decou-

pler made according to the above design principles was built in the HFSS simulation

software and some results are derived.

Sections 7.5, 7.6 and 7.7 contain the main results of this chapter obtained by theoreti-

cal method and simulation method. Based on these results, improvements are made

in designing the slitted decoupler compared with the design in [15]. In detail, the

analysis of the slitted decoupler begins in Section 7.5 with an analysis of a rectangular

patch antenna because of their resemblance in structure. The knowledge in under-

standing rectangular patch antenna is expected to be made use of in understanding

the slitted decoupler. In Section 7.6, comprehensive simulations of the effects of vary-

ing the dimensional and electrical parameters of slitted decoupler are made and the

new principles for designing the slitted decoupler are summarised. In a further stage

of simulation in Section 7.7, a simple half wavelength dipole is placed upon the slitted

decoupler instead of a tag antenna for the purpose of simplicity to observe the inter-

action between the dipole and the decoupler. The interaction is investigated firstly by

varying the slit width and the separation distances between the dipole and the slit-

ted decoupler; and then observing the influences of these variations on the induced

voltage, input impedance and transmitted RF power of the dipole.

In Section 7.8, various slitted decouplers differing in size were fabricated, a commercial

tag was deployed thereon and the reading range of the tag on different decouplers was

obtained by measurement. By comparing the experimental results, the conclusions and

expectations made above were validated.

Finally, conclusions are drawn and possible further work on the slitted decoupler is

described.

Page 123

7.2 Structure of the Slitted Decoupler

7.2 Structure of the Slitted Decoupler

The basic structure of a slitted decoupler, introduced in [15], is shown in Figure 7.1. It

is composed of two separated patches with a slit in the middle, a dielectric layer, and

a ground plane shared by the two patches.

The inventors [15] also provided other types of slitted decoupler for the purpose of

achieving multiple resonant frequencies or making the decoupler work despite the po-

larisation of the irradiating waves. Those derivations are all based on the understand-

ing of the basic structure as shown in Figure 7.1. This chapter only aims to analyse the

basic structure and improve its performance. But we believe the performance of other

derivations can also be improved based on the understanding in this chapter.

Dielectric layer

Ground plane

h

Top patchSlit

y

z

(a) Side view

W

L

s

y

x

(b) Top view

Figure 7.1. The structure of the slitted decoupler.

The dimensional parameters contained in this model are the length L and width W of

each patch, the thickness h of the dielectric layer and the width s of the slit. The slitted

decoupler is supposed to be placed between a tag and a metal item with which the tag

is intended to be associated. The placement of the decoupler is shown in Figure 7.2.

Dielectric layer h

Tag

Dz

y

z

Metal item

(a) Side view

W

L

s

Tag

y

x

(b) Top view

Figure 7.2. Slitted decoupler placement illustration. A tag is placed above the slitted decoupler

in a distance Dz. Below the decoupler, can be seen the metal item.

Page 124


7.3 Design Principles

Some design principles for the slitted decoupler, proposed in [15], are now listed.

• Patch Length L

L ' λ0

2√

εrN (7.1)

where λ0 is the wavelength in the free space, εr is the relative permittivity of

the sandwiched dielectric material. N is an integer. The fundamental resonant

frequency is obtained when N = 1.

• Patch Width W

The patch width may be determined by the dimension of the selected RF tag.

Commonly, the patch width is 4 to 5 times than that of a tag on the slitted decou-

pler. A reduction in patch width is stated in [15] to diminish the read range of the

tag on it.

• Thickness h

The thickness of the dielectric material is stated in [15] to be preferably less than

a few λ0/1000.

• Slit Width s

The slit width is preferably less than λ0/100.

• The distance Dz between a tag and slitted decoupler

The distance is proposed to be a few hundred micrometers.

• Dielectric material

The dielectric core can be one of the commonly used dielectric materials such

as polystyrene, BOPP (Biaxially oriented polypropylene film) or polycarbonate.

Material with low loss is preferable.

In Subsection 7.6.7, some new principles for designing a slitted decoupler are proposed

by the author of this thesis. These are based on the extensive theoretical analyses

Page 125

7.4 Simulation

in Subsection 7.5.1 and simulations in Subsection 7.5.2. Some principles in Subsec-

tion 7.6.7 agree with the principles proposed in [15], while those principles which do

not so agree can be used to improve the understanding and performance of the slitted

decoupler.

7.4 Simulation

7.4.1 Construction of the Simulated Devices

An example of the slitted decoupler constructed according to the design principles

given in Section 7.3 and intended to be resonant at 923MHz was simulated in HFSS.

Some parameters of this device are listed below and are also shown in Figure 7.3. The

rectangular (x, y, z) coordinate system shown in Figure 7.3 is used for discussion. The

slitted decoupler is excited by an incident uniform wave propagating as shown in Fig-

ure 7.3 towards the decoupler.

λr =λ√εr

h = 1mm <λr

180

L = 90.5mm ' λr

2W = 70mm

s = 0.5mm <λr

360εr = 3.2

δ = 0.003

where δ is the dielectric loss tangent. The dielectric material used here is polyester.

7.4.2 Simulation Results

The magnitudes of the r.m.s. phasors representing the simulated electric fields are

shown in Figure 7.4. Simulation results not shown illustrate that the electric fields in

the planes parallel to yz plane and cutting the decoupler are similar. Hence, only the

Page 126


er=3.2 1mm

Slit

k

E=1V/m

Uniform plane wave

y

zTop patch

Ground plane

Origin

(a) Side view

Top patch 70mm

90.5mm

0.5mm

y

x

Origin

(b) Top view

Figure 7.3. The simulated slitted decoupler. A rectangular coordinate system is defined in this

figure. The red dot is the origin. The decoupler lies parallel to the xy plane and the

slit is parallel to x axis. A uniform plane wave, in which the electric field of 1V/m is

directed along the positive y axis, illuminates the decoupler.

magnitude of the simulated electric field distribution in plane cutting the middle of the

decoupler (the plane is x = 35mm according to the coordinates in Figure 7.3) is shown

in Figure 7.4(a). In particular, it shows the large electric field in the slit, and that the

strong field is dramatically attenuated as we move along the y axis away from the slit

or along the z axis away from (above) the slit. The field direction is also marked in

Figure 7.4(a) and the direction of electric field in the slit is found to be opposite to that

of the incident field.

Figure 7.4(b) is the magnitude of electric field underneath the top patch. It is evident

that the maximum electric field appears on the edge of each patch. The minimum elec-

tric field occurs at the centre of each patch. The magnetic field distribution is just op-

posite to that of the electric field according to simulation results which are not shown

here. In the slit, the area separating the two patch edges, the electric field is signifi-

cantly higher than that in other areas.

It is apparent that there is a tangential component of electric field on the top surface

of the decoupler. Moreover, the RFID tag antennas are usually linearly polarised and

the polarisation is parallel to the surface to which they are attached. The tangential

component can excite tag antennas if this component is large enough. For the purpose

of observing the values of these tangential electric fields in different positions above

the decoupler, Figure 7.5 is given.

Page 127

7.4 Simulation

4mm

s

z

y

(a) Side view

x

y

(b) Top view

Figure 7.4. Magnitude of the r.m.s. phasors representing the simulated electric fields of

the slitted decoupler. (a) shows the magnitude of electric field in the an yz plane

at x = 35mm (the middle of the patch). The warm-toned area is the position of slit,

of which the width is s=0.5mm. (b) shows the electric field in the transverse section,

underneath the top patch. The yellow strip, which color denotes high electric field, is

the place where the slit is.

Figure 7.5(a) indicates that the peak value of the y-directed fields above the decoupler

occurs at y=0mm, where the slit is. It attenuates sharply as we move away from the

slit in z direction. Although as mentioned above when we move away from the slit

the peak value of the y-directed fields drops dramatically, a relatively large y value

(≥2V/m) of the E-field can be found in a wide region above the decoupler.

All these simulation results demonstrate that the field reflected by the slitted decou-

pler is quite different from that of metal. Some tangential electric field is found to

exist rather than to be cancelled as happens in close proximity to metal. Moreover,

according to these results, the proper placement for a tag on the slitted decoupler can

be estimated. The tag’s centre should be put just above the decoupler’s centre and the

polarisation of the tag antenna should be perpendicular to the slit and parallel to the

top surface of the decoupler to collect more useful y-directed fields as it is shown in

Figure 7.2. The distance between the tag and the slitted decoupler, which is presented

as Dz in Figure 7.2 can be chosen flexibly between a few hundred micrometers to a few

Page 128


-10 -8 -6 -4 -2 0 2 4 6 8 10

0

20

40

60

80

100

120

140

y (mm)

z=1mm, x=35mmz=0.6mm, x=35mmz=0.4mm, x=35mmz=0.2mm, x=35mm

Ey(V

/m)

z=5mm, x=35mm

(a) Along y axis

0 10 20 30 40 50 60 700

20

40

60

80

100

120

140

160

180

x (mm)

Ey(V

/m)

z=0.2mm, y=0mm

z y=0.6mm, =0mm

z y=2mm, =0mm

(b) Along x axis

Figure 7.5. The magnitude of y-directed electric field variation along the y and x axes at

various heights. Ey is the magnitude of the r.m.s. phasors of the y-directed electric

fields. x, y, and z are the coordinates of the observing point in the rectangular coordinate

system built in Figure 7.3.

millimeters, since there is a relative high electric field (>2V/m) in the range. More de-

tails on this distance Dz have been discussed in Subsection 7.7.1. From another point

of view, the large electric field in the slit of the patch, which further simulations have

shown to be accompanied by very low magnetic field, is a property of the artificial

magnetic conductor (AMC) which was introduced in Chapter 6. Hence we view the

slitted decoupler as one kind of AMC. As one kind of AMCs introduced in Subsec-

tion 6.3.3, to obtain large y-directed electric fields or high impedance in the slit is the

main concern in designing a good slitted decoupler.

The structure of slitted decoupler inevitably makes people associate it with rectangu-

lar patch antennas. The slitted decoupler is similar to a patch antenna array which

contains two rectangular cells separated by a short distance. Additionally, the electric

field distribution of a slitted decoupler resembles that of a rectangular patch antenna.

In order to understand the operating scheme of this decoupler, it is assumed that the

slitted decoupler has resonance properties similar to those of the rectangular patch an-

tennas, so use can be made of knowledge of patch antenna properties. Although the

excitation modes between the decoupler and patch antenna are very different, (one is

Page 129

7.5 Patch Antenna Resonant Property Analysis

passive radiating element which is powered by an incident wave, the other one is a

driven antenna), this assumption is still reasonable by reciprocity theory. Particularly,

as introduced above, the high y-directed electric fields or high impedance in the slit

are the main goals in designing a well performing decoupler and the y-directed elec-

tric fields above the slit of the decoupler correspond to the y-directed electric fields

on the edge of rectangular patch antenna. Hence, to determine which factors play a

significant role in establishing the tangential electric fields or impedance on the patch

antenna edge is the task of Section 7.5.


Rectangular patch antennas have been widely used in recent years due to their low

profile, low costs and easy fabrication into linear or planar arrays. The structure of a

simple rectangular patch antenna without excitation part is shown in Figure 7.6.

z

y

Dielectric layer

Top patch

Ground plane

h

(a) Side view

x

y

Top patch

Grand plane

W

L

(b) Top view

Figure 7.6. The structure of a simple rectangular patch antenna without excitation. h in

(a) is the thickness of dielectric layer, W and L are width and length of the top patch

separately.

As we can see, the patch antenna is composed of three layers which are a top patch,

a ground plane and a sandwiched dielectric layer. Usually, the ground plane is much

larger than the top patch.

The description at the end of Section 7.4 suggests that to design a good decoupler

should start with the understanding of rectangular patch antenna, especially how to

Page 130


get the high y-directed electric fields or high impedance on the edge of rectangular

patch antenna. We believe the y-directed electric fields on the edge of patch antenna

is caused by fringing fields, which has been shown in Figure 7.7 and the magnitude of

the y-directed electric fields and fringing fields on the patch edge significantly depend

on the antenna resonance. Hence, the resonant properties of the rectangular patch

antenna are introduced in this section. In detail, several resonant properties of rectan-

gular patch antennas such as patch size, resonant input impedance and electric field

distribution are discussed below. All these properties are analysed using two main

methods: (a) in Subsection 7.5.1, a theoretical method mainly based on transmission

line modes, and (b) in Subsection 7.5.2, a simulation method using Ansoft HFSS.

z

y

Top patch

Ground plane

h

Fringing fields

Figure 7.7. Electric field distribution of a rectangular patch antenna.

The transmission line model means that a rectangular patch antenna can be consid-

ered as an array of two radiating slots, each of patch width W and dielectric material

thickness h, separated by a transmission line of length L.

7.5.1 Theoretical Analysis

(a) Patch width and length selection for resonance

Generally speaking, there is a rough relationship between the resonant frequency of a

rectangular patch antenna and the length of the rectangular patch. This relationship is

expressed in (7.2).

L =c

2 fr√

εr(7.2)

where L is the length of the rectangular patch, c is the light speed in free space, fr is the

resonant frequency, εr is the relative permittivity of the dielectric layer.

Page 131


(7.2) is based on an assumption that all the fields are in the dielectric material of which

dielectric constant is εr. However, this assumption is not quite accurate, because some

fields fringe out of the dielectric material into the air, as shown in Figure 7.7.

Therefore the relative permittivity εr and length L in (7.2) should be replaced by rela-

tive effective permittivity εre f f expressed in (7.3), which is only valid for W > h, but

this is generally the case, and effective length Le f f expressed in (7.4). Hence, (7.2) be-

comes (7.5) [79].

εre f f =εr + 1

2+

εr − 12

[1 + 12h

W]−1/2 (7.3)

Le f f = L + 2h× 0.412(εre f f + 0.3)(W

h + 0.264)

(εre f f − 0.258)(Wh + 0.8)

(7.4)

Le f f =c

2 fr√

εre f f(7.5)

By inserting (7.4) into (7.5), (7.6) for the length L can be obtained.

L =c

2 fr√

εre f f− 2h× 0.412

(εre f f + 0.3)(Wh + 0.264)

(εre f f − 0.258)(Wh + 0.8)

(7.6)

(7.6) states all the factors including the patch length L, patch width W, thickness h and

relative permittivity εr of the dielectric material affect the resonant frequency fr. If

a patch antenna is required to be resonant at 923MHz, and the thickness and relative

permittivity of the dielectric material are 1mm and 3.2 respectively, the combinations of

patch width and length which can lead to the required resonant frequency are shown

in Figure 7.8.

However, Figure 7.8 does not mean that the optimum radiation performance can be

achieved by adjusting the patch width and length along the curve arbitrarily, because

if the width is too small compared with the wavelength, the radiation efficiency will

be decreased and if the width is too large compared with the wavelength, high order

modes may appear and result in field distortions [79]. Even though Figure 7.8 does

not point to optimum conditions, it still indicates that the resonance of a patch antenna

mainly depends on the patch length rather than the width. A practical width Wp is

Page 132


0 20 40 60 80 100 120 140 160 180 20090

95

100

105

Patch width (mm)W

Pat

ch l

ength

(mm

)L

Figure 7.8. Patch width and length values making the antenna resonant at 923MHz. The

thickness of dielectric material h is 1mm. The relative permittivity is 3.2.

stated in [79] to be:

Wp =c

2 fr

√2

εr + 1(7.7)

Balanis using (7.3), (7.4), (7.5) and (7.7) proposes the patch antenna design procedure

which can be found on page 819 in [37]:

• Specify the required resonant frequency fr, the dielectric constant εr and thick-

ness of the dielectric material.

• A practical width is decided by (7.7).

• (7.3) gives the effective dielectric constant.

• The actual length of the patch can be calculated then by solving (7.6).

Following this procedure, if h=1mm, εr=3.21, fr=923MHz, the practical patch width of

112mm and length of 90.64mm are obtained. The simulation results by HFSS show that

the resonance at 923MHz can be obtained with a patch 89.5mm by 112mm, which is

very close to the theoretical result. The details of this simulation model and the results

are shown later in Subsection 7.5.2.1The material used here is polyester, which loss tangent is 0.003.

Page 133


(b) Thickness effects

According to (7.6), which is derived by the transmission line model, the dielectric layer

thickness affects the patch antenna’s resonance. However, (7.6) does not show the rela-

tionship between the value of the dielectric thickness and the magnitude of the electric

fields on the patch edge. The cavity model for analysing a rectangular patch antenna,

the description of which can be found on page 826 in [37], gives us qualitative analysis

on the relationship between the dielectric thickness and the electric fields on the patch

edge. When the ratio of thickness to the patch width is quite small, the energy tends to

be stored between the two layers instead of radiating outwards, which not only results

in high quality factor but also narrows bandwidth. This is because the directions of

currents flowing on the top patch and the ground plane are opposite. When the top

patch and the ground plane are in close proximity, the near field cancellation becomes

significant [80].

Top patch

Ground plane

+++

+++

+++++

- - -

- - -

- - - - -

Jb1

Jb2

Jt

Figure 7.9. Charge and current distribution in a rectangular patch antenna.

By the analysing method of cavity model, a side view of a rectangular patch antenna

is shown in Figure 7.9, wherein the distribution of charges and currents is presented.

When the antenna is energised, a charge distribution is established under and above

the top patch, as well as on the surface of the ground plane. “The charge distribution

is controlled by two mechanisms; an attractive and a repulsive mechanism” [81]. The

attractive mechanism provides forces on the charges distributed on the two opposite

edges of the two conductor layers. The movement of these charges forms current den-

sities Jb1 and Jb2. On the other hand, some charges, under the bottom of the top patch,

are pushed to go around the edge to its top surface. This charge flow creates the current

density Jt. However, Jt will tend to be zero, when the dielectric layer thickness-to-patch

width ratio becomes quite small, since the forces arising from the charges caused by

Page 134


currents Jb1 and Jb2 flowing on two adjacent conductors tend to cancel each other and

little current is sent by repulsion to the top surface. Hence there is almost no current

or in other words no tangential magnetic field component on the edges of the patch.

Moreover, a large amount of charge accumulates on the edge of the patch so that max-

imum electric fields appear there.

As a result, a conclusion can be drawn that a larger y-directed electric fields and

impedance on the patch edge can be obtained when the thickness of dielectric layer

is smaller on condition that other parameters of the patch antenna, such as the patch

length, width, dielectric constant etc. are adjusted according to the dielectric layer

thickness to make the antenna resonant.

(c) Resonant input impedance

According to the transmission line model, a rectangular patch antenna can be consid-

ered as an array of two radiating slots, each of patch width W and dielectric material

thickness h, separated by a transmission line of length L. The two slots are labelled

as slot #1 and slot #2. Assuming that the admittance of slot #1 is Y1 = G1 + jB1, the

conductance G1 can be obtained from (7.8) which is derived from Balanis and can be

found on page 822 in [37].

G1 =1

πη

∫ π

0[sin( k0W

2 cos θ)cos θ

]2 sin3 θdθ (7.8)

where k0 = 2πλ0

is the wave number in the free space and λ0 is the wavelength in free

space at the resonant frequency.

Equations giving the admittance at slot #2 are stated by Balanis [37] to be the same.

The total admittance at slot #1 is obtained by transforming the admittance at slot #2 to

the position of slot #1, and adding it to the admittance at slot #1.

In performing this transformation we take note (i) that to make the antenna resonant

in the presence of the fringing field we have had to reduce the physical length of the

antenna by an amount:

2∆L = 2h× 0.412(εre f f + 0.3)(W

h + 0.264)

(εre f f − 0.258)(Wh + 0.8)

(7.9)

Page 135


and (ii) the susceptance B1 or B2 at each end provided by the fringing field, and which

brings the patch to resonance, is the same as that of a strip transmission line of width

equal to the patch width and of length ∆L.

Then in transforming the susceptance at slot #2 to the position of slot #1 we note that

the transformation is over a transmission line length of λ/2− 2∆L. The result from

these observations is that we obtain for the transformed admittance approximately the

same conductance, and a susceptance of approximately the same magnitude but with

the sign changed. Thus the transformed admittance is:

Y2 = G2 + jB2 = G1 − jB1 (7.10)

Therefore the input admittance, which is the sum of Y1 and Y2, becomes 2G1 and the

resonant input impedance is, as shown by Balanis on page 823 in [37], expressed by:

Zin =1

Y1 + Y2=

12G1

(7.11)

However, (7.11) does not consider the mutual effects between the slots. Following

Balanis [37] we state that the mutual conductance can be obtained from:

G12 =1

ηπ

∫ π

0[sin( k0W

2 cos θ)cos θ

]2 J0(k0L sin θ) sin3 θdθ (7.12)

where J0 is the Bessel function of the first kind of order zero. For typical microstrip

antennas Balanis [37] states that the mutual conductance is small compared with the

self conductance. Therefore, the practical resonant input impedance becomes:

Zin =1

2(G1 ± G12)(7.13)

“where the plus (+) sign is used for modes with odd resonant voltage distribution

beneath the patch and between the slots while the minus (-) sign is used for modes

with even resonant voltage distribution” [37]. The Zin in (7.13) is the resonant input

impedance at the edge of the patch.

7.5.2 Simulation results

In this subsection, the details of the simulation result are provided in the following four

aspects: simulation model, resonant frequency, input impedance and electromagnetic

field distribution.

Page 136


• Simulation model

This rectangular patch antenna is fed by a coax cable with one watt of incident

power, and is meant to be resonant at 923MHz. The detail of the antenna dimen-

sion is shown in Figure 7.10. The resonant frequency 923MHz can be obtained

for various values of patch width and length by means of which have been intro-

duced in Subsection 7.5.1 and the expected results have been shown in Figure 7.8.

(a) Side view of a patch antenna fed by a coaxial cable

1

W

L

(b) Top view of a patch antenna fed by a coaxial cable

W+60

L+60

0.7 er=3.2

Unit: mm

Figure 7.10. A rectangular patch antenna fed by a coaxial cable. The red probe is the con-

nection between the top patch and the feed line. Its diameter is 0.7mm. The relative

permittivity of dielectric material in the middle is 3.2. The feed power is 1W.

The conductor used here is copper and its thickness is 10µm which is thicker than

the skin depth. The skin depth δ of good conductor can be calculated from:

δ =1√

π f µ0µrσ(7.14)

where σ is electrical conductivity. The dielectric material is polyester for which

εr = 3.2 and the loss tangent is 0.003.

All the simulation results following are obtained at the resonant frequency 923MHz.

• Resonant frequency and input impedance

Page 137


A typical input impedance of a rectangular patch antenna as a function of fre-

quency obtained from HFSS is shown in Figure 7.11.

910 915 920 925 930 935-60

-40

-20

0

20

40

60

80

100

120

Frequency (MHz)

Imp

edan

ce (

oh

m)

Reactance

Resistance

Xf

Figure 7.11. A typical input impedance of patch antenna as a function of frequency. The

red and blue curves correspond to the reactance and the resistance respectively. These

two curves are symmetrical about the green dashed line. The feed reactance X f here

is 20Ω.

Ideally, both the resistance and reactance are symmetrical about the resonant fre-

quency [37]. As shown in Figure 7.11, the reactance and resistance are symmetri-

cal along the green dashed line which indicates the resonant frequency (923MHz).

However, because of the feeding coax cable, the reactance at the resonant fre-

quency is not zero, but has a value which is called the feed reactance, and this

feed reactance can be ignored when the thickness of the patch antenna is much

smaller than the wavelength. The equivalent circuit of a patch antenna with a

feed reactance is drawn in Figure 7.12.

As illustrated by Figure 7.11, the input impedance of a rectangular patch antenna

is a function of frequency. Particularly, the maximum resistance can be obtained

at the resonant frequency. Moreover, this resonant resistance on the edge of the

patch can be calculated by (7.13) and its forerunners. Therefore, it is concluded

that the high input resistance can be obtained on the patch edge at the resonant

frequency and preferably be obtained by a small patch width antenna, since (7.13)

Page 138


Xf

R XL XC

Inputimpedacne

A

B

Figure 7.12. The equivalent circuit of a patch antenna which is fed by a coax cable. X f is

the feed reactance brought by the feed cable. R, XL and Xc compose the circuit of

patch antenna itself. The input impedance can be obtained at the left port AB.

and its forerunners state that the resonant resistance is inversely proportional to

the patch width.

As mentioned at the end of Subsection 7.5.1 and discussion of the working scheme

of AMCs in Subsection 6.3.3, the high y-directed electric fields on the patch edge

is always companied with high resistance there. Hence, we expect the high y-

directed electric fields on the patch edge are preferably obtained by a small patch

width antenna at its resonant frequency. This expectation will be verified in the

next itemisation.

In addition, the comparison between the simulation results from HFSS and the

theoretical results shown in Figure 7.8 in terms of top patch size at resonance is

shown in Figure 7.13. The two curves reasonably agree with each other.

In the red curve in Figure 7.13, various patch sizes which can make the patch an-

tenna resonant at 923MHz according to the simulation are shown. By inserting

these size values, h = 1 and εr = 3.2 into the formula (7.13) and its forerun-

ners, the theoretical resonant input impedance on the patch edge can be derived.

The resonant input impedance on the patch edge can also be derived from the

HFSS software. Hence, the simulation results and theoretical results in terms of

resonant input resistance on the patch edge are compared in Figure 7.14. One

thing should be noted here is that the resonant input impedance obtained from

(7.13) and its forerunners is real. However, the result obtained from HFSS is com-

plex. That is because the HFSS takes into account the feed reactance which can

add some reactance at the resonant frequency. This added feed reactance here is

Page 139


0 20 40 60 80 100 120 140 16089

90

91

92

93

94

95

96

Patch Width (mm)W

Pat

ch L

eng

th(m

m)

L

Theoretical results

Simulation results

Figure 7.13. The comparison between the simulation results and the theoretical results in

terms of patch size at resonance. The blue curve is the theoretical result which

was shown earlier in Figure 7.8. The red curve is the simulation result from HFSS.

much smaller than the resistance at the resonant frequency, so in this comparison,

the feed reactance is ignored.

From Figure 7.14, all the three curves demonstrate that high resonant resistance

preferably occurs at narrow patch width. However, the simulation results (b),

which consider the losses, do not agree well with theoretical results. This is at

least partially caused by the losses. With the decline of the patch width, the losses

in the antenna becomes considerable [79], which losses (7.13) and its forerunners

do not consider. Hence, the simulation results (a), which omit the losses in the

simulation model, and are drawn in the green curve in Figure 7.14, are seen to be-

come somewhat closer to the theoretical results in the blue curve. However, when

the patch width is less than λr/2 (90.8mm), there is still large disagreement be-

tween these two curves. A conclusion is made that (7.13) and its forerunners are

not suitable for calculating the resonant input resistance, when the patch width

is less than λr/2 (90.8mm), and are especially not suitable, when the losses are

large. Another conclusion that can be drawn from Figure 7.14 is that the losses

can reduce the resonant input resistance, especially when the patch width is less

Page 140


20 40 60 80 100 120 140 1600

1000

2000

3000

4000

5000

6000

7000

8000

9000

Res

onan

t In

put

Imped

ance

(ohm

)

Theoretical results

Simulation results (a)

Simulation results (b)

Patch Width (mm)W

Figure 7.14. The comparison between the simulation results and the theoretical results in

terms of the resonant input impedance. The blue curve is the theoretical result

which is calculated by (7.13) and its forerunners. The red curve is the simulation result

from HFSS taking losses into account. The green curve is the simulation result from

HFSS omitting all the losses.

than λr/2 (90.8mm). This also indicates that the losses could possibly reduce the

magnitude of the y-directed electric fields on the patch edge.

• Electromagnetic fields distribution

Once a rectangular patch antenna is resonant, the maximum electric field appears

on the edge of the top patch and the minimum appears in the middle of the top

patch. The distribution of magnetic field on the patch is just converse to that of

the electric field.

The electric field distribution for a patch antenna, which patch width W=112mm,

patch length L=89.4mm and resonant at 923MHz is shown in Figure 7.15. In

Figure 7.15, it is obvious that the E-fields on the edge of the top patch are not

vertical to the ground plane but inclines towards to the left. These are known

as the fringing electric fields, so that some tangential electric fields exist along y

axis.

Page 141


The y-directed electric fields on the edge of the patch antenna as a function of

the patch width at various patch length were obtained by HFSS and are shown

in Figure 7.162. By comparing the y-directed electric field values in each curve in

Figure 7.16, it is found that the peak value always occurs when the patch width

and length are matched to be resonant. Additionally, by comparing the peak val-

ues among the three curves in Figure 7.16, it can be stated that the smaller patch

width contributes to obtaining higher y-directed electric fields on the edge of the

patch. Both the conclusions made above comply with the previous analysis and

assumptions (second bullet point above) indicating that the high tangential elec-

tric fields occur with the patch dimensions that lead to resonance and a narrow

width gives high resonant impedance. It is also noticed from Figure 7.16 that the

larger the peak value of the curve is, the sharper the curve is. That fact demon-

strates when the patch width becomes narrow, the antenna is resonant at a high

Q factor which leads to a narrow bandwidth.

(a) Electric fields vector (b) Magnitude of the electric fields

Figure 7.15. The r.m.s phasor of the electric field distribution underneath top patch obtained

by HFSS. The electric field is shown on the top surface of the dielectric material,

which is just underneath the top patch. The input impedance of the antenna has

been matched to the feed port so that the antenna can receive the maximum available

power.

2The y-directed electric fields are not constant along the patch edge, but fluctuate slightly from an in-

termediate value, hence the values shown in Figure 7.16 are the intermediate values of the r.m.s phasors

of the y-directed fields along the patch edge.

Page 142


10 20 30 40 50 60 70 80 90 1000

1

2

3

4

5

6

7x 10

4

L=92mmL=90.5mmL=89.7mm

Ey(V

/m)

Patch Width (mm)W

Figure 7.16. The y-directed electric fields as a function of the patch width at various patch

length. The thickness of dielectric layer h=1mm, Ey is the magnitude of the r.m.s.

phasors of the y-directed electric fields on the patch edge.

In stating a conclusion for this section, the high y-directed electric field on the patch

edge is mainly determined by the antenna’s resonance. The top patch length, width,

the dielectric constant and thickness of the dielectric layer should be considered to-

gether to achieve the optimum resonance. The resonance of the rectangular patch an-

tenna mainly depends on the patch length (around half wavelength after considering

the effects of the dielectric material) but other mentioned parameters also play a sig-

nificant role in achieving resonance. Moreover, on condition that the rectangular patch

antenna is resonant, the high y-directed electric fields on the patch edge are obtained

preferably by narrow patch width, thin dielectric layer and low-loss materials. How-

ever, it is also found that the high tangential electric field is obtained at the expense

of bandwidth. As introduced at the end of Section 7.4, because of the resemblance

between the rectangular patch antenna and the slitted decoupler, it is useful to make

use of the knowledge of the former to understand and design the latter. In addition,

obtaining high y-directed electric fields in the slit of the decoupler is the main goal in

designing a good decoupler. It is therefore assumed here that the high y-directed elec-

tric fields in the slit of the decoupler can be obtained with a slitted decoupler which

Page 143

7.6 Slitted Decoupler Parameter Settings

has a narrow patch width, thin dielectric layer and low loss material and a condition

of resonance.


Because of the shape resemblance between the slitted decoupler and the rectangular

patch antenna, an assumption is made at the end of Section 7.5 that the high y-directed

electric fields in the slit of the decoupler share the properties of the corresponding

points on the edge of the rectangular patch antenna. In this section, simulation results

giving the y-directed electric fields in the slit of the slitted decoupler are presented. By

comparing these simulation results between the antenna mode and decoupler mode,

the assumption made at the end of Section 7.5 is validated in this section. Finally, and

by deriving from these simulation results for the slitted decoupler, some design prin-

ciples for the slitted decoupler are proposed for comparison with the design principles

in the patent [15], which were described in Section 7.3.

7.6.1 Simulation Model

As introduced in Section 7.2, a typical structure of the slitted decoupler is shown in Fig-

ure 7.1, and is shown again in Figure 7.17 as a reminder. This decoupler is illuminated

by a uniform plane wave which is also shown in Figure 7.17(a).

Dielectric layer

Ground plane

h

Top patchSlit

y

z

k

E=1V/m

Uniform plane wave

(a) Side view

W

L

s

y

x

(b) Top view

Figure 7.17. The structure of the slitted decoupler illuminated by a uniform plane wave. h

is the thickness of the dielectric layer, s is the width of slit, W and L are the width

and length of each patch.

Page 144


Several parameters should be considered in the design which can be divided into two

different types. First, these are the electrical parameters, such as conductivity of the

metal material, relative permittivity and loss tangent of the dielectric material. These

parameters mainly depend on the material selection. Secondly, these are dimensional

parameters, including the length L, width W of each top patch, the width of the slit s

and the thickness of the dielectric material h3. All the parameters mentioned above can

be explored by Ansoft HFSS. By varying these parameters, the electric fields which are

in and vertical to the slit (y-directed) can be derived.

Except for the simulations in Subsection 7.6.5 for examining the effects of the dielec-

tric material variation to the y directed field in the slit, the dielectric material used is

polyester which relative permittivity is 3.2 and its loss tangent is 0.003. We expect the

performance of this material to be similar to that of a lossless material. This material

is also used in the simulation model of patch antenna in Section 7.5. The metal above

and underneath the dielectric material is copper. Its thickness is 10µm, which is signif-

icantly larger than the skin depth at 923MHz.

The basic values of the decouplers’ physical parameters are provided as follows: L =

90.5mm, W = 32.5mm, h = 1mm, s = 0.5mm. They are fixed in building the model

in the simulation software unless they are varied for examining the effects to the y-

directed electric field in the slit from that variation separately. The values are inten-

tionally chosen since they are forming one combination of the resonance of decoupler.

Any change in a single parameter will destroy the resonance which leads to the re-

duction of the y-directed electric field in the slit (the reader will find out why in the

following discussion).

Moreover, the slitted decoupler is designed for working at 923MHz which is the centre

frequency of Australian UHF RFID band. The following simulation results are derived

at this frequency unless specified otherwise.

3The size of the ground plane remains the same as it shown in Figure 7.17. The effects brought by the

variation of the ground plane size is analysed in Subsection 7.6.6.

Page 145


7.6.2 Length and Width of Each Top Patch Selection

The parameter simulation tests begin with the size of each top patch, since it is the

most critical dimension to ensure the slitted decoupler’s resonance. The y-directed

electric fields in the slit as a function of the patch width at various patch length are

obtained by HFSS and shown in Figure 7.184. As mentioned above, except the param-

eters being examined, other parameters are fixed and their values can found at the end

of Subsection 7.6.1 or in the caption of Figure 7.18.

10 20 30 40 50 60 70 80100

150

200

250

300

350

400

450

500

550

600


Ey(V

/m)

Patch Width (mm)W

Figure 7.18. The y-directed electric fields in the slit as a function of the patch width at

various patch length. The thickness of dielectric layer h=1mm, the width of slit

s=0.5mm, Ey is the magnitudes of the y-directed electric fields in the slit.

By comparing Figure 7.18 and Figure 7.16, it is found that the y-directed electric fields

in Figure 7.16 are neither in the same order of magnitude nor proportional to the cor-

responding y-directed electric fields in Figure 7.18. This enormous difference is caused

by the different excitation modes. The values in Figure 7.16 are obtained for an an-

tenna driven from a source of available source power 1W. In contrast, the values in

Figure 7.18 are obtained by illuminating the slitted decoupler by a uniform plane wave

4The y-directed electric fields are not constant in the slit, but fluctuate slightly from an intermediate

value, hence the values shown in Figure 7.18 are the intermediate magnitudes of the r.m.s phasors of

the y-directed electric fields in the slit. This note is also applicable for the other y-directed electric fields

shown in Figure 7.19, 7.20, 7.21 and 7.23.

Page 146


of r.m.s. phasor 1V/m propagating along the -z axis. However a similarity is found

that the peak values of y-directed fields occur at the similar patch sizes in both antenna

mode and decoupler mode. The comparison tells us the resonant properties of the slit-

ted decoupler are similar to those of rectangular patch antennas in terms of patch size.

The high y-directed electric field is preferably to be obtained by a small patch width

decoupler. Furthermore, the gradients of the three curves in Figure 7.18 illustrate that

the resonance occurring with a narrow patch width is a high Q resonance. That high Q

resonance leads to narrow bandwidth of slitted decoupler which is in accordance with

what happens to patch antenna discussed in Subsection 7.5.2.

7.6.3 Dielectric Material Layer Thickness Selection

From the analysis of rectangular patch antennas, it is known that the thickness of the

middle dielectric material affects the y-directed electric fields on the patch edge to some

extent. For the slitted decoupler, this property is also found and shown in Figure 7.19.

0 2 4 6 8 10200

250

300

350

400

450

500

550

Thickness of dielectric layer (mm)t

L W=90.5mm, =32.5mm

Ey(V

/m)

Figure 7.19. The y-directed electric fields in the slit as a function of the dielectric layer thick-

ness at the particular patch size 90.5mm×32.5mm. The width of slit s=0.5mm.

From Figure 7.19, it is found that there is an optimum thickness to achieve the maxi-

mum y-directed electric fields in the slit. We believe this is because the resonant fre-

quency is in part dependent on thickness.

Page 147


7.6.4 Slit Width Selection

In this subsection, the y-directed electric fields in terms of the variation of the slit width

are examined and the results are shown in Figure 7.20.

s=1mm

s=0.25mm

s=2mm

s=5mms=10mm

s=0.5mm

Figure 7.20. The y-directed electric fields in the slit at various slit width. s represents the slit

width. The patch size is 90.5mm×32.5mm, the thickness of the dielectric material

remains 1mm. The magnitude of the y-directed electric fields can be interpreted by

the color column in the left side of each sub-figure.

Figure 7.20 illustrates that no matter how large or small the slit is, there is a high electric

field area (>100V/m) which is narrow and near the edge of each patch. When the slit

width is equal to or less than 2mm, the two adjacent areas join together. Once the

slit width is less than 2mm, the y-directed electric fields in the slit increase with the

reduction of the slit width. It does not mean that the less slit width the better, because

a tag antenna on the decoupler with a narrow slit may not receive as much power as

it does on the decoupler with a wide slit, even though, the y-directed electric fields

are larger in the narrow slit than that in the wide slit. Although we cannot decide the

optimum slit width for tag antennas on the decoupler to receive power just according

to Figure 7.20 (the method of figuring out the optimum slit width will be introduced

Page 148


in Section 7.7), a conclusion still can be drawn that the slit width has to be very small

compared to the wavelength (the slit width should be less than 2mm in this case). This

conclusion complies with the conclusion made in the patent [15].

7.6.5 Dielectric Material Selection

The dielectric material selection should be considered in two aspects: dielectric con-

stant and loss tangent. By changing the dielectric materials, the dielectric constant

may be increased or decreased, which leads to the shorter or longer patch length at

resonance. The effect brought by the variation of dielectric constant is relatively easy

to understand, therefore, we do not spend resources on that, instead, the variation of

loss tangent is the main factor which is concerned in this subsection.

10 20 30 40 50 60 70 8050

100

150

200

250

300

350


Ey(V

/m)

Patch Width (mm)W

Figure 7.21. The y-directed electric fields in the slit as a function of the patch width at

various patch lengths when the loss tangent is increased to 0.02. s=0.5mm,

h=1mm.

According to the analysis of rectangular patch antennas in Section 7.5, a lossless dielec-

tric material in the patch antenna is preferable to obtain high impedance on the edge

of the patch observed in Figure 7.14 and also results in high y-directed electric fields

there. In this subsection, a similar property is expected to be found for the slitted de-

coupler. In order to examine the influence of loss tangent, the loss tangent of polyester

Page 149


used before is changed from 0.003 to 0.025 and then Figure 7.18 is transformed to Fig-

ure 7.21.

By comparing Figure 7.18 and Figure 7.21, the lower y-directed electric fields are found

in Figure 7.21 than the counterpoints in Figure 7.18. Moreover, the comparison also

indicates that the losses shift the peak value of each curve into a narrow patch width,

except when patch length L=90.5mm. Hence, in order to obtain high y directed electric

fields in the slit, a low loss material is preferable.

7.6.6 The Ground Plane Size Selection

According to the simulation model shown in Figure 7.17. The length of the ground

plane is equal to two times of the top patch length plus the slit width and the width

of the ground plane is equal to the width of the top patch width. All the simulation

results before this subsection in Section 7.6 are obtained by the simulation model in

Figure 7.17. However, that model does not consider the effects brought by the variation

of the ground plane size. Hence, in this subsection, the effects of the ground plane size

to the y-directed electric fields in the slit are discussed. The new simulation model is


Dielectric layer

Ground plane

h

Top patchSlit

y

z

k

E=1V/m

Uniform plane wave

(a) Side view

W

L

s

y

x

Margin

(b) Top view

Figure 7.22. Slitted decoupler with a ground plane larger than the top layer.

In Figure 7.22, a new dimensional parameter “Margin” is defined. By varying the value

of the margin, the size of ground plane can be controlled. The y-directed electric fields

in the slit as a function of the margin are shown in Figure 7.23.

5Such a lossy polyester may not exist in the practical world, here we mean to assume the existence of

one material which shares other properties of the polyester but is more lossy.

Page 150


Figure 7.23 illustrates that the y-directed electric fields in the slit are maximum when

margin is zero, since all the electrical and dimensional parameters including the ground

plane with margin zero are optimised to form a resonance. The details of the parame-

ters can be found in the caption of Figure 7.23. With the increase of the ground plane

size, the y-directed electric fields do not stop decreasing until the margin is larger than

60mm. The reduction of the y-directed electric fields in the slit is caused by the de-

tuning of the ground plane variation. When the margin is 80mm, by expanding the

patch length from 90.5mm to 91.4mm and remaining the values of other parameters

to make the decoupler retune, the y-directed electric fields in the slit climb up to an

even higher value 620V/m at 923MHz, which demonstrates that the decoupler prefers

a large ground plane.

0 10 20 30 40 50 60 70 80200

250

300

350

400

450

500

550

Margin (mm)

E

(V/m

)y

Figure 7.23. The y-directed electric fields in the slit as a function of margin at 923MHz.

L ×W=90.5mm×32.5mm, dielectric layer is composed of low-loss polyester which

thickness is 1mm, εr=3.2. The slit width remains 0.5mm.

The size of the ground plane can also be considered as the size of the metallic item for

detection. Then, this results tell us that if the decoupler is required to be resonant at a

high level, the size of the detected item should be considered ahead.

Page 151


7.6.7 Design Principles for the Slitted Decoupler

According to the above simulation results of the slitted decoupler in Section 7.6, the

resonant properties of a slitted decoupler are similar to the resonant properties of a

rectangular patch antenna. Hence, the knowledge of the rectangular patch antenna

can be made use of to design the slitted decoupler. Some design principles for the

slitted decoupler are thus proposed for comparison with the design principles in the

patent [15], and introduced in Section 7.3.

• Resonance

The top patch length, width, the size of the ground plane, the dielectric constant

and thickness of the dielectric layer should be considered together to achieve the

optimum resonance.

The resonance of the slitted decoupler mainly depends on the patch length (around

half wavelength after considering the effects of the dielectric material) but other

mentioned parameters also play a significant role in achieving the resonance.

Moreover, on condition that the slitted decoupler is resonant, the high y-directed

electric fields in the slit are obtained preferably by narrow patch width, thin di-

electric layer and large ground plane.

• Dielectric material loss tangent

A low-loss dielectric material is preferable to be used in designing a slitted de-

coupler.

• Slit width

There is no certain conclusion of the slit width, but it should be much smaller than

operating wavelength, generally less than λ0/100, where λ0 is the wavelength of

the resonant frequency in free space. To obtain the optimum slit width, the tag

antenna on the slitted decoupler should be considered. This part has been done

in Subsection 7.7.1.

Page 152


7.7 A Dipole on the Slitted Decoupler

The analysis in the last section demonstrates that the slitted decoupler has the similar

properties to the rectangular patch antenna. Hence, using the analysis on patch anten-

nas, it is relatively easy to fix the size of the slitted decoupler to obtain high electric

fields in and perpendicular to the slit. However, there are still several questions need-

ing to be solved, such as: (i) whether the high electric fields in the slit are large enough

to excite the tag antenna on the decoupler; (ii) what is the optimum distance between

the tag and the decoupler; (iii) how to select the optimum slit width to make the tag re-

ceive the maximum power; and (iv) does the decoupler and the tag on it interact with

each other significantly.

Since there are abundant commercial RFID tags and most of them are structurally

complicated, it is hard to simulate and analyse the performance of a commercial tag

antenna on the slitted decoupler. Therefore, instead of a commercial tag antenna, a

simple half wavelength dipole is placed above the slitted decoupler for simulation.

The sketch of a dipole above the slitted decoupler illuminated by a uniform plane wave

is shown in Figure 7.24. Some dimensions of the slitted decoupler have been fixed and

marked in the figure. The material here is the low-loss polyester used before. Some

dimensions, such as Dz which is the distance between the dipole and the decoupler,

and the slit width s, are unknown. The half-wavelength dipole deployed is composed

of two round copper wires separated by a port in the middle. The wires are shown

in orange in Figure 7.24. The diameter of both wires is 0.4mm. The whole length

of the dipole is 153.4mm, the middle port length is 2mm. This dipole is resonant at

923MHz in free space. The input impedance in the middle port of the dipole is about

68Ω at 923MHz. The typical RFID tag antenna impedance matching condition is not

considered here because of the purpose of simplification.

For answering the previous questions, three radiation characteristics of the half wave-

length dipole on the slitted decoupler are examined in the following three subsections

respectively. The three characteristics are induced voltage, input impedance and avail-

able source power.

Page 153


Dielectric layer 1

Half-wavelength dipole

Dz

y

z

k

E=1V/m

Uniform plane wave

(a) Side view

32

.5

90.5

s

y

x

(b) Top view

Figure 7.24. A dipole on the slitted decoupler (unit:mm).

In the formulae and calculations to follow, a mixture of peak value phasors and r.m.s.

phasors are encountered. This comes about because it is a tradition in formulating

many equations of communications, peak value phasors are used. However, results

obtained from HFSS simulations are in terms of r.m.s. phasors. Care has been taken to

specify the correct units and to insert the correct values into formulae in each case.

7.7.1 Induced Voltage in the Middle Port of the Dipole on the De-

coupler

The open circuit terminal voltage, sometimes called the induced voltage, is one of the

characteristics for judging an antenna’s ability in collecting the incident wave. The

calculation method of the induced voltage of a receiving antenna illuminated by a

linearly polarised incident wave has been introduced in Subsection 3.2.3 and more

details about the induced voltage of half wavelength dipole incident by a uniform

plane wave (linearly polarised) can be found in Appendix B. However, neither the

method in Subsection 3.2.3 nor the method in Appendix B is capable in calculating

the induced voltage of the dipole on the slitted decoupler, since the electric field lines

around the slit are curved. Moreover, the simulation software HFSS does not provide

a direct function for obtaining the open circuit terminal voltage. The method we adopt

is to obtain the y-axis electric fields at a range of points within the gap between two

arms of the dipole by HFSS, check that there are no singularities, and then integrate

that field component across the gap. This method is called the integration method in

Page 154


this subsection. The induced voltage of the dipole in free space and above a metal plate

incident by a uniform plane wave can also be derived by the integration method.

Moreover, a theoretical method for calculating the induced voltage of the dipole in

free space and above an infinite ground plane incident by a uniform plane wave is

proposed in Appendix B for double check. Finally, the induced voltages are compared

among the dipole above the slitted decoupler at various distances, the dipole in free

space and the dipole above the metal at various distances. The comparison is shown

in Figure 7.25.

0 2 4 6 8 100

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Dz (mm)

Ind

uce

dV

olt

age

(V)

The dipole on the slitted decoupler (Integration method)

The dipole in free space (Integration method)

The dipole in free space (Theoretical method)

The dipole on metal (Integration method)

The dipole on metal (Theoretical method)

Figure 7.25. The induced voltage comparison among the dipole on the slitted decoupler, the

dipole on the metal and the dipole in free space. The distance Dz has different

meanings for the curves in three colors. For the blue curve, it is the distance between

the underside of the dipole and the slitted decoupler. For the red curves, it is also the

distance from the underside of the dipole to the metal. The green curves represent

the induced voltage of the dipole in free space and in this case there is no concept of

Dz, therefore, the green curve is totally flat. The slit width of the decoupler remains

0.5mm in all the cases.

Both the red curves with stars and squares denote the induced voltage of the dipole

above the metal at various distance. The red curve with stars is derived by the in-

tegration method. The metal underneath the dipole in simulation is not infinite but

a 180mm×90mm copper sheet. The red curve with squares is derived by theoretical

Page 155


method introduced in Appendix B, which involves an infinite ground plane under-

neath the dipole. Apparently, these two red curves agree with each other well. They

both indicate that once the dipole is placed in close proximity to the metal, the induced

voltage in the middle port of the dipole is nearly zero, since the y-directed electric fields

hardly exist there. With the increase of the distance between the dipole and the metal,

the induced voltage is increased as well but still much lower than the induced voltage

of the dipole in free space.

Similarly, the two green curves with stars and squares represent the induced voltage

of the dipole in free space, which are derived by the integration method and the the-

oretical method in Appendix B respectively. The values got by theoretical method are

slightly higher than those by integration method, since in the theoretical analysis, the

dipole length is exact half wavelength which is longer than the practically resonant

dipole in integration method and the longer dipole can induce more voltage.

The induced voltage of the dipole on the slitted decoupler derived by the integration

method and presented in the blue curve in Figure 7.25, drops with the increase of the

distance between the dipole and the decoupler, but it is always larger than that in free

space. From Figure 7.4, it is known that the electric fields around the slit attenuate

dramatically along both y and z directions. It seems that the dipole close to the slit

can induce much more voltage than the dipole which is further away from the slit.

However, we did not observe that in Figure 7.25. This is because although when the

dipole is moved upwards from the decoupler, the y directed electric field at the same

height and close to the centre of the dipole drops dramatically, the y directed electric

field at the same height of the dipole but further away from the centre of the dipole

will increase somehow as shown in Figure 7.5(a). Hence, the induced voltage of the

dipole on the decoupler is still somehow flat as shown in Figure 7.25.

According to the above analysis that demonstrates that the strong y directed electric

field only concentrate closely to the slit, it is assumed that a very short dipole (much

smaller than half wavelength) close to the slit can induce the same voltage as the half

wavelength dipole does there, but with the increase of the distance between the short

Page 156


dipole and the decoupler, the induced voltage of the short dipole will drop dramati-

cally. The assumption has been demonstrated in Figure 7.26 and this demonstration

complies with the introduction in patent [15] that a much smaller antenna than a com-

mercial tag antenna can be used in conjunction with a slitted decoupler. The total

length of the short dipole used here is 6mm containing two symmetrical copper wires

which diameter and length are 0.4mm and 2mm respectively. There is a 2mm gap be-

tween the two wires. Dz is the distance between the underside of the short dipole and

the decoupler.

0 1 2 3 4 5 6 7 80

0.05

0.1

0.15

0.2

0.25

The short dipole on the slitteddecoupler (Integration method)

Dz (mm)

Induce

dV

olt

age

(V)

Figure 7.26. The induced voltage of a short dipole on the slitted decoupler.

The slit width of the decoupler remains 0.5mm in the blue curve of Figure 7.25 and

Figure 7.26. In order to examine the induced voltage of the dipole on the slitted decou-

pler in terms of the variation of the slit width, Figure 7.27 is given and in this figure,

the distance between the half wavelength dipole and the decoupler remains 0.3mm.

Figure 7.27 states that with the increase of the slit width from 0.1mm to 2mm, the

induced voltage of the dipole on the decoupler climbs sharply to the peak value at the

slit width 0.4mm-0.5mm and after that drops dramatically as well. When the slit width

is over 2mm, the induced voltage declines at a relatively low rate.

Page 157


0 1 2 3 4 5 6 7 8 9

0.16

0.18

0.2

0.22

0.24

0.26

0.28

Slit width (mm)s

Ind

uce

d v

olt

age

(V)

Dz=0.3mm

Figure 7.27. The induced voltages of the half wavelength dipole on the slitted decoupler

as a function of the slit width. The distance between the dipole and the decoupler

remains 0.3mm.

7.7.2 Input Impedance of the Half Wavelength Dipole on the De-

coupler

If the available power from the half wavelength dipole on the decoupler is wanted,

knowing only the induced voltage is not enough. Knowledge of the output (or input)

impedance of the dipole above the slitted decoupler is also necessary. Therefore, in this

subsection, the input impedance of the half wavelength dipole on the slitted decoupler

at various distances is discussed. Only the input impedance of a driven antenna can

be obtained directly by HFSS. However, at the beginning of Section 7.7, it is noted that

the slitted decoupler with the dipole mounted above is treated as a passive element

and illuminated by a uniform plane wave. In this case, the output impedance of the

dipole cannot be derived directly by HFSS. However, by means of reciprocity theory,

the output impedance of an antenna above the decoupler will be the same as the input

impedance of the antenna above the decoupler when the antenna is driven by a voltage

source. As a result, the incident wave in Figure 7.24 is removed from the model, in-

stead, a lumped port with an available source power of 1W from a source impedance of

68Ω is added in the port in the middle of the half wavelength dipole. By this method,

Page 158


the input impedance of the dipole in various distances above the slitted decoupler is

obtained and shown in Figure 7.28.

0 2 4 6 8 100

50

100

150

200

250

300

350

400

450

Imped

ance

()

W

ResistanceReactance

Dz (mm)

Figure 7.28. The input impedance of the dipole in various distances above the slitted de-

coupler. The slit width in this case remains 0.5mm. The y axis Dz represents the

distance between the underside of the dipole and the slitted decoupler. The blue curve

represents the resistance in the middle port of the dipole. The red curve denotes the

reactance in the middle port of the dipole.

As noted above, the input impedance of the dipole in free space is about 68Ω. Ap-

parently, compared with the input impedance of the dipole in free space, the input

impedance of the same dipole above the decoupler varies dramatically, and there is

no simple relation between the impedance and the distance Dz. But, generally, the

input or output impedance of the dipole becomes less affected by the decoupler with

the increase of Dz. The general behavior shown in Figure 7.28 demonstrates that the

interaction between the decoupler and the dipole is too critical to be ignored.

7.7.3 Power Collected by the Half Wavelength Dipole on the De-

coupler

In Subsections 7.7.1 and 7.7.2, the induced voltage and input impedance of the dipole

on the slitted decoupler have been introduced. Supposing that the load impedance of

Page 159


the middle port is 68Ω. This port impedance can satisfy the maximum power transfer

when the half wavelength dipole is put in free space. The maximum available power

received by the load can be calculated by (7.15). The details of the derivation of (7.15)

are noted in Subsection 3.2.1.

PA =|V f

in|28R f

ant

(7.15)

where PA is the maximum available power received by the load of the dipole in free

space, V fin is the peak value phasor of induced voltage of the dipole in free space. Ac-

cording to the green curve with stars shown in Figure 7.25, the magnitude of the r.m.s

phasor of the induced voltage of the dipole is 0.099V when the dipole is illuminated by

a uniform plane wave of r.m.s. phasor 1V/m directed along the -z axis. The antenna

output or input impedance and the load impedance are 68Ω. The maximum available

power received by the load is 0.036mW by (7.15). Similarly, the power received by the

same load of the same dipole on the slitted decoupler can be calculated by (7.16). The

details of the derivation of (7.16) are noted in Subsection 3.2.1.

PL =|Vd

in|2RL

2|Rdant + RL + j(Xd

ant + XL)|2=

|Vdin|2RL

2[(Rdant + RL)2 + (Xd

ant + XL)2](7.16)

where PL is the power received by the load of the dipole on the decoupler. Vdin is the

peak value phasor of the induced voltage of the dipole on the decoupler. Rdant and Xd

ant

are the resistance and reactance of the antenna input or output impedance. RL (68Ω)

and XL (0Ω) is the resistance and reactance of the load respectively. In particular, when

the half wavelength dipole introduced at the beginning of Section 7.7 is put above the

slitted decoupler which patch length and width are 90.5mm and 32.5mm respectively,

dielectric material thickness is 1mm, the slit width is 0.5mm and Dz=0.3mm, the mag-

nitude of r.m.s phasor of induced voltage in the middle port of the dipole is 0.274V by

Figure 7.25. The input or output impedance of the dipole is 392+j110Ω by Figure 7.28.

Therefore, the power received by the load (68Ω) of the dipole on the slitted decoupler

is 0.0228mW by (7.16). When Dz=10mm, the input impedance of the dipole becomes

220+j65Ω by Figure 7.28. The magnitude of the r.m.s phasor of induced voltage be-

comes 0.215V by Figure 7.25. The power received by the load (68Ω) of the dipole

on the slitted decoupler becomes 0.0361mW by (7.16). That indicates the particular

load (68Ω) of the half wavelength dipole can receive more power when it is placed

Page 160


above the decoupler in a certain distance (Dz=10mm) than it is put in close proximity

(Dz=0.3mm) to the decoupler.

Furthermore, both the calculated powers received by the load of the dipole above

the decoupler, either 0.0228mW derived when Dz=0.3mm or 0.0361mW derived when

Dz=10mm, are comparable to the power 0.036mW when this dipole is matched in free

space.

7.7.4 Antenna Design Principles for the Slitted Decoupler

Based on Subsection 7.7.1, it is found that the half wavelength dipole can obtain similar

quantity of induced voltage, no matter whether it is very near to the decoupler or

placed in a certain distance (less than 10mm) above the decoupler. Additionally, a very

short dipole obtains nearly the same induced voltage as the half wavelength does,

when they are placed very close to the decoupler. Once the very short dipole is moved

further up, the induced voltage drops dramatically. According to Subsection 7.7.2,

although the input or output impedance variation of the half wavelength dipole above

the slitted decoupler does not have a clear relationship with the distance Dz between

the dipole and the decoupler, with increase of Dz , the input or output impedance of

the dipole becomes less affected by the decoupler underneath.

A large number of commercial tag antennas are designed based on dipole antenna pat-

tern and their length is usually λ0/3. The output impedance of these tag antennas is

deliberately designed to match the chip. It is assumed that the commercial tag anten-

nas on the slitted decoupler will be affected in a similar way to the half wavelength

dipole on the slitted decoupler. Based on this assumption, some design or placement

principles of the slitted decoupler with tag antenna are proposed following.

• Commercial tags placement above the slitted decoupler

Although in Subsection 7.7.3, the conclusion is drawn that the particular load

(68Ω) of the half wavelength dipole can receive more power when it is placed

above the decoupler in a certain distance (Dz=10mm) than it is put in close prox-

imity (Dz=0.3mm) to the decoupler. However, when it comes to the commercial

Page 161

7.8 Measurement

tags, it is hard to draw the same conclusion since the situation is quit different

from the half wavelength dipole: 1) the shape of the commercial tags are much

more complicated, 2) the impedance of the chip of commercial tags are complex,

therefore not 68Ω. Furthermore, both the calculated powers received by the load

of the dipole antenna above the decoupler either when Dz=0.3mm or Dz=10mm

are comparable with the available source power of the dipole in free space. That

indicates the dipole based tag antennas above the decoupler could work as well

as when they are in free space.

When it comes to the optimum distance between commercial tags and the slitted

decoupler, that optimum distance depends on the interaction between the tag

and the decoupler underneath. But that distance has to be within a few millime-

ters. How to find this optimum distance between commercial tags and the slitted

decoupler to achieve a long reading range could be a future research topic.

• Redesigned tag placement above the slitted decoupler

A new tag antenna can also be designed to make its input or output impedance

conjugate match the chip impedance when the antenna is close to the decoupler.

A low profile of the whole structure (tag on the decoupler) and long reading

range are expected to be achieved by this new tag antenna working with the

decoupler. The size of this antenna is expected to be designed based on a very

short dipole (much shorter than half wavelength) and it will be placed just above

the slit, since the y-directed electric fields concentrate in a small area around the

slit. How to design a suitable RFID tag to work in close proximity to the slitted

decoupler could be a topic for future work.

7.8 Measurement

This section means to examine the simulation results and the conclusions made based

on these results in Section 7.6 and Section 7.7. As shown in these two sections, the

magnitude of the electric field in the slit of the decoupler and the induced voltage in the

terminal of the half wavelength dipole on the decoupler are used as the two parameters

Page 162


to judge the resonance and the performance of the decoupler. However, practically,

these two parameters are difficult to measure. It is believed that the reading range

of the commercial tag on the decoupler is positively related to these two parameters.

Hence, the work in this section intends to measure the reading range of a commercial

tag on various decoupler differing in size.

Furthermore, because of the limitation of available materials, we do not use the low-

loss polyester which is used in the previous simulation. An FR4 board is available with

a dielectric constant of about 4.4, a loss tangent of about 0.02, a thickness of 1.6mm and

a copper thickness of 17µm. The effects brought by these fixed parameters cannot be

examined. However, there are still several physical parameters which can be controlled

during the fabrication and measurements, such as the top patch length and width, the

ground plane size and the distance between the decoupler and the tag. Those four

parameters are varied and tested by obtaining the reading range of a commercial tag

on slitted decouplers.

7.8.1 Measurement Facilities

A commercial tag is intentionally selected for the experiments in this section, which

is designed based on dipole antenna pattern. The overall size of this tag is 90mm×22mm. The tag adheres to EPC C1G2 protocol. In addition, the RFID reader Model

ID ISC.LRU 2000 and 5.7 dBi gain circularly polarised reader antenna both by FEIG

Electronics, were employed. All the following measurements were operated under

the Australian UHF RFID standards and regulations. A shielding tunnel shown in

Figure 6.4 was used again to isolate this experiment from the outside environment.

When the tag or the tag with slitted decoupler are put in this tunnel, it is assumed that

they are effectively in free space. The reader antenna is pointed into the tunnel. The tag

reading range in this tunnel without slitted decoupler underneath is about 5 meters.

The decoupler’s size was optimised by the simulation software Ansoft HFSS firstly

according to the introduction in Section 7.6. In the simulation, most of the physical

or electrical parameters of the decoupler, e.g. board thickness, dielectric constant, loss

tangent have been fixed as mentioned before, and the patch width could be either

Page 163

7.8 Measurement

35mm or 80mm for the purpose of investing the effects brought by the variation of the

patch width. The slit width is optimised to be 0.5mm according to the simulation. The

patch length is thus the only parameter of the decoupler left which we could adjust to

make the decoupler tune. In addition, the size of the ground plane in all the simula-

tions is just equal to the sum of the two patches and the slit. Hence, the effects brought

by the variation of the ground plane or in other words the size of the attached metallic

item is not considered in the optimisation. We will examine the degradation in the

decoupler’s performance caused by this thoughtlessness later.

This simulation based optimisation is not very accurate due to the inherent error of

the software but more importantly the inaccurate knowledge of the materials used in

fabrication. Therefore, the simulation can only provide an approximate range of the

optimised decoupler length. A few decouplers around this simulation based size were

fabricated. The reading range of the selected commercial tag is measured when it is

placed in near proximity above these fabricated decouplers separately. The decoupler

which makes the tag above it detected at the longest distance is defined as the really

optimised decoupler. In this section, not all of the testing results on the fabricated de-

couplers are given. We only give the results on the decouplers optimised by simulation

and those optimised by experiments.

As mentioned before, the patch width is fixed to be either 35mm or 80mm and cor-

responding to the comment made in the last paragraph that only the two kinds of

decouplers which patch length are optimised either by simulation or experiment will

be discussed, hence, four slitted decouplers fabricated are shown and marked with

numbers in Figure 7.29.

The geometrical parameters of the four decouplers can be found in Table 7.1. They

can be categorised by two ways. In terms of the patch width, decouplers 1 and 2 are

in the same category with the patch width of 35mm and similarly decouplers 3 and

4 are in the same category with the patch width of 80mm. In terms of the optimised

methods, decouplers 1 and 3 are in the same category in which the decouplers are

optimised by simulation and decouplers 2 and 4 are in the same category in which the

decouplers are optimised by experiments. It is found that the length of the decouplers

Page 164


1

3

2

4

Figure 7.29. The four fabricated slitted decouplers.

optimised by experiments is about 1mm longer than that of the decouplers optimised

by simulations.

Table 7.1. Geometrical parameters of the fabricated slitted decoupler.

Decoupler L (mm) W (mm) s (mm) h (mm)

Decoupler 1 75.5 35 0.5 1.566

Decoupler 2 76.6 35 0.5 1.566

Decoupler 3 74 80 0.5 1.566

Decoupler 4 75 80 0.5 1.566

7.8.2 Measurement Results and Comparison

The test results in terms of the tag’s reading range when the tag is placed above these

four decouplers were obtained separately and compared in this subsection, for the

purpose of validating the conclusions made in Section 7.6 and Section 7.7.

(a) Validation of the effects brought by the decoupler top patch size

In this subsubsection, the commercial tag is deployed in the near proximity above the

centre of the tested decoupler. The distance between the tag and the decoupler Dz

Page 165

7.8 Measurement

is 0.1mm. The reading ranges of this tag were obtained by varying the underneath

decoupler from number 1 to number 4 and these results are shown in Table 7.2.

According to the results in Table 7.2, the reading ranges of the tag above the decou-

plers which are optimised by experiments are longer than those optimised by simula-

tion which is easily understood. The other observation is that the reading range of the

tag on the decouplers with smaller patch width is more sensitive to the change of the

patch length. This is because the decoupler with small patch width resonates with high

Q factor and narrow bandwidth. Moreover, the maximum reading range of the tag oc-

curs on the decoupler 2 which is optimised by experiment and with the smaller patch

width 35mm. Those observations adhere to what is shown in Figure 7.18 in Section 7.6

that with the decrease of the patch width the curves of the electric field magnitude as

a function of the patch length become sharp, and the peak value of the curve becomes

large which indicates long reading range. Hence, to obtain long reading range, nar-

row patch width is preferable on condition that other parameters of the decoupler are

adjusted according to the narrow patch width to make the decoupler highly resonant.

Table 7.2. Reading ranges of the tag on the decouplers varied in size.

Decoupler Name (size) Reading range

Decoupler 1 (75.5mm×35mm) 2390mm


Decoupler 3 (74mm×80mm) 2810mm


(b) Validation of the effects brought by the decoupler ground plane size

In this subsubsection, the two experimentally optimised decouplers 2 and 4 are at-

tached on an aluminium plate separately. This plate can represent the ground plane of

the decoupler or the metallic item on which the decoupler is going to be attached. This

plate size is 260mm×260mm. The tag’s reading range is also tested when it is placed

in the near proximity (about 0.1mm) above the decouplers with the aluminium plate

underneath respectively. The placement of the three items are shown in Figure 7.30.

Page 166


Aluminum plate

Decoupler 2

Tag

(a) Decoupler 2

Aluminum plate

Decoupler 4

Tag

(b) Decoupler 4

Figure 7.30. The placement of the tag on both the decoupler and the plate.

Table 7.3. Reading ranges of the tag on both the decoupler and the aluminium plate.




By comparing the corresponding reading ranges in Table 7.2 which are obtained when

the tag is just above the decoupler and the reading ranges in Table 7.3 which are ob-

tained when the tag is above both the decoupler and the aluminium plate, the conclu-

sion can be drawn that the metallic item underneath can degrade the performance of

the combination of the tag and the decoupler which is optimised without the consid-

eration of the metallic item. In addition, the reading range of the tag on the decou-

pler with larger patch width is less degraded. Hence, the attached metallic item’s size

should be considered when the decoupler is designed. This conclusion complies with

that conclusion made in Subsection 7.6.6.

(c) Validation of the effects brought by the distance Dz

In the previous two subsubsections, the distance between the tag and the decoupler

Dz remains 0.1mm. In this subsubsection, this distance will be increased to 6.5mm by

inserting a bubble wrap between them. The effects brought by the variation of the

distance to the reading range of the tag will be investigated.

Page 167

7.9 Conclusion

The comparison between the corresponding data in Table 7.2 and those in Table 7.4

indicates that the reading range of the tag on decoupler 2 is diminished by enlarging

the distance between them, conversely, the reading range of the tag on decoupler 4 is

enhanced by enlarging the distance between them. It is hard to say that whether the

tag should be placed in near proximity above the decoupler or at a certain distance.

It all depends on the interaction between the tag used and the shape of the decoupler.

But one thing that can be said with certainty is that the tag can work fairly well in a few

millimeter range above the decoupler. These observations adhere to what were found

and concluded in Subsection 7.7.4.

Table 7.4. Reading ranges of the tag above the decouplers in a certain distance

(Dz=6.5mm).




7.9 Conclusion

The analysis in this chapter validates that the slitted decoupler originally proposed by

Brown et al. [15] can solve the antenna on metal problem, and the slitted decoupler re-

mains a low profile and simple structure within UHF RFID band (860MHz-960MHz).

Compared with the work in [15], the author of this thesis gives the comprehensive ex-

planation of the slitted decoupler’s operational principle by making use of the knowl-

edge of the rectangular patch antenna. New design principles are made differing from

those in [15] in order to further minimise the size and optimise the performance. In de-

tail, all the physical parameters should be optimised together to obtain the resonance

of the slitted decoupler. The size of the detected metallic item and the interaction

between the decoupler and the tag should be considered ahead to achieve premium

performance. More significantly, the patch width could be reduced dramatically com-

pared with that in [15]. Meanwhile, the reading range of the tag could be enhanced

along with the minimisation of the patch width. Although, the minimisation of the

Page 168


patch size and enhancement of the tag performance on the decoupler are obtained at

the expense of the bandwidth of the slitted decoupler. The expense is still acceptable,

since a UHF RFID system is a narrow bandwidth system. The fractional bandwidth of

most UHF RFID systems in the world is about 0.5%-2%.

Page 169

Chapter 8

Detection of MassiveNumbers of DVDs

THIS chapter aims to provide an investigation and a solution to

the problem of reading massive numbers of DVDs on a pallet

by UHF RFID systems. The components of a regular packaged

DVD are reported. The influence of a single DVD on the illuminating field

is investigated by theoretical analysis and simulation. Abundant experi-

ments have been conducted to define the labelling method of the tags on

the DVDs in the stack and to recommend the optimum configuration for

DVD stacking. Finally, a solution to a high level detection of massive num-

bers of DVDs is given. A perfect detection is also achieved by eliminating

the weak tags. The chapter also suggests avenues for improvement.

Page 171

8.1 Introduction and Motivation


This section begins with a description of the motivation for conducting this project.

Also described is an operational constraint which has shaped the conduct of the work.

Some general perspectives one of which could receive further study by others in the

future are mentioned. The relevance of some results that may be found in the literature

is assessed. The chapter outline is also discussed and the content of each section is

summarised.

8.1.1 Motivation

The identification of each packaged DVD, consisting of a DVD case with a disc inside,

among a large number of packaged DVDs in a stack is required by industry, and espe-

cially by the retail industry. For example, supermarket managers want to be in control

of the information and status of all DVDs in a stack in real time.

A single DVD is easily detected by a barcode system. However, when a large number

of packaged DVDs are piled up into a stack, the cost in time to scan them one by one

makes the barcode system totally impractical for addressing this task.

UHF RFID systems are thought to be potentially useful in solving this problem because

of the UHF radio wave’s long range propagation ability. However, when a large num-

ber of packaged DVDs are piled up into a stack, the detection of each DVD in the stack

is not as easy as the detection of a single packed DVD in free space, because of the wave

reflection, attenuation, absorption and diffraction which can occur in the stack. More-

over, DVDs contain a very thin metal layer (less than 100nm) working as a reflective

layer of a laser beam when they are being read. This metal layer and interference with

the propagating waves by a large number of tags make the prediction of the detectabil-

ity of packaged DVDs in stack difficult, so we here aim to provide an investigation and

if possible a solution to this problem. Since the wave in the stack is very complicated

due to the mutual coupling among the massive number of tag antennas on the DVDs

and interaction between the tags and the metal layer in DVDs, and since the ratio of

the minimum dimension (the metal layer in each disc which is less than 100nm) to the

Page 172

Chapter 8 Detection of Massive Numbers of DVDs

maximum dimension of the whole DVD stack is very large, neither theoretical analysis

nor use of simulation software provide very practical approaches to address the whole

issue. However, theoretical analysis and simulation are briefly employed in studying

the influence of a single DVD on the illuminating field. Despite this fact, direct experi-

ment is the main approach used to show the feasibility of detecting a large number of

packaged DVDs in stack by UHF RFID systems.

8.1.2 An Operational Constraint

This chapter contains the results of an investigation into the problem of reading mas-

sive numbers of DVD devices in a stack. For this purpose a massive number is consid-

ered to be in the region of 2000. However, funds supporting the project do not permit

the purchase or deployment of as many as 2000 packaged DVDs plus 2000 commercial

tags. Therefore a significant portion of the work has been done to identify stacking ar-

rangements for a limited number of packaged DVDs that will allow what is believed to

be an accurate prediction of the performance of a reading strategy for the full number

of 2000.

This investigation has led to the conduct of experiments not only in free space and on

typical wooden pallets but also in a reflection free environment created by stacking

pieces of UHF absorbing foam.

8.1.3 Some General Perspectives

We hope that, in addition to conducting experiments, the problem of reading 2000

DVDs on a pallet can be addressed from some general perspectives.

The first (and almost certainly misguided) perspective is the belief that by placing the

reader sufficiently far from the stack for all DVDs to fall within the main lobe of the

reader antenna radiation pattern, all tags will be read if there is sufficient power density

at the position of the furthest tag to do so. This calculation gives, with the commonly

used reader, a distance that the reader should stand back, and a power density at the

position of the furthest tag, which is adequate for reading. This analysis does not

Page 173


appear to give a reading limitation for the commonly used reader, i.e. there is in this

simple view, and with the reader we use, a reserve of power. Of course the limited

perspective above does not take into account the concept that the reading beam will

be weakened by passing through the tags closest to the reader, but such weakening

undoubtedly does occur.

In our studies we found neither theory nor observation to support the belief that it is

possible to place the reader at a sufficient distance from a dense DVD stack to place

all tags in its main lobe and then read all those tags. We recommend that, because

of the weaknesses of the assumptions, and the absence of empirical confirmation, this

perspective be abandoned in conceiving a solution to the problem.

The second perspective is that of identifying by how much the reader field is weakened

by its transit through the front DVDs. We have thought of two approaches to this

question. One is to apply the theory of scattering of reader fields by well matched

dipoles. The second seeks to take note of the effective area of the individual tags, and

the fact that there is substantial overlap between them.

We have found analysis according to these latter principles difficult and unconvincing

as to its relevance. Such analysis cannot replace direct experimental observation of

what is possible. For that reason, such direct experimental observation forms the major

content of this chapter.

8.1.4 Literature Treatments

The question of what are the interactions between the elements of an array of RFID

labels is beginning to receive some attention in the literature [82] [83]. However, the

former treatment makes the assumptions that: (i) there is substantial spacing, of the

order of a wavelength, between the labels; (ii) the labels are placed on empty cases;

(iii) the label antennas are straight; and (iv) non-replying labels have no backscattering

properties. We believe the first three assumptions are not suitable in our case, since

the DVDs, each with a metal layer inside, are densely stacked. We also believe the

fourth assumption is not true, because a tag antenna will backscatter as long as it is

Page 174


not opened. The latter treatment spaces the labels somewhat closer in some directions,

but generally not as closely as in our case. It considers only two planes of tags. It does

find some strong interactions, e.g. invisibility of the tags in the back plane except at an

optimum inter-plane spacing, and the readability of the front plane influenced by the

position of the back plane in a manner that resembles metallic reflection.

The assumptions above make those treatments inapplicable to the problem we are

studying in this chapter. The results emphasise the complexity of the problem. We

are thus confirmed in our view that a successful analytical treatment for closely spaced

folded label arrays in dense DVD stacks has yet to be developed, and our direct exper-

imental approach is appropriate.

8.1.5 Chapter Outline

This chapter can be divided into four parts. The content in each part is summarised as

follows.

The chapter begins in Section 8.1 which is part one, with a description of the motiva-

tion for applying RFID technology to the detection of DVDs in a stack. The difficulties

in and several general perspectives on the detection of massive numbers of DVDs in

a stack are discussed. Because of the difficulties, the experimental method is chosen

as the main approach to this problem. Theoretical analysis and simulation are briefly

employed in studying the influence of a single DVD on the illuminating field. In ad-

dition, an operational constraint relating to the number of DVDs available for test is

described. The literature on the interaction among elements of an array of RFID labels

is introduced but after careful study, the conclusion is drawn that it is not suitable for

analysing our problem.

Part two includes Section 8.2 and Section 8.3. Firstly, the physical and electrical pa-

rameters of a common packaged DVD are investigated in Section 8.2. It is reported

that the DVDs are mostly composed of low loss polymers which are not sensitive to

electromagnetic waves. The component that has most of effect is the metal layer in the

Page 175


middle of the disc and used for reflecting the laser, since this layer imposes a metal-

lic boundary condition that has a significant effect on the electromagnetic waves sur-

rounding it. Because of the small thickness of the layer (much smaller than the skin

depth), theoretical analysis and simulation of its effect is conducted in Section 8.3.

Based on the analysis, the feasibility of detecting a large number of packaged DVDs in a

stack is considered to depend mainly on two factors: (i) the labelling method of the tag

on the DVD, (ii) the penetrability of the interrogating waves of various polarisations.

The various combinations of different tag labelling methods and interrogating wave

types are tested by obtaining the reading range of the tagged DVDs in a three step pro-

cess. The three steps are described in the following sections respectively which in total

form part three. In Section 8.4 which is the first step, those combinations are examined

by obtaining the reading range of a single tagged DVD in free space. The combina-

tions remaining after filtering by step one are tested by obtaining the reading range of

a single and multiple tagged DVDs in a relatively small DVD stack (containing about

170 DVDs) surrounded by absorbing foams in Section 8.5. The final step discussed in

Section 8.6 considered the method of packaging and stacking DVDs in industry and

examines the effect of various stacking policies. The number of the tagged DVD is ex-

panded to 320. The readable ratio, defined as the ratio of the number of detected tags

to the number of totally tagged DVDs, is obtained. The optimum labelling and stack-

ing method, together with the appropriate interrogating wave, is identified according

to the value of the readable ratio.

The last section, Section 8.8, consists the part four. It contains a summary of the results

of those experiments and gives the possible future work on this project.

There are also two appendixes. Appendix C records the original test data for deter-

mining the reading ranges for tags attached at various positions on a DVD case. Ap-

pendix D contains a study of the reflecting properties of UHF absorbing foam used in

study of the identification of limited numbers of packaged DVDs.

Page 176


8.2 Parameters of a Packaged DVD Product

DVDs can be categorised into many kinds, such as DVD-ROM (“ROM” stands for

Read Only Memory), DVD R (“R” stands for recordable) etc. Since the DVD discs in

different categories have different inside structures [84], and the work in this chapter

only focuses on the identification of DVD-ROM, in the following description the term

“DVD” is meant to refer to a DVD-ROM.

In this section, the physical parameters and electrical parameters of a regular packaged

DVD are given. A packaged DVD is composed of a case and a disc. The dimensions

of a regular DVD case, which has the form of a rectangular prism, are shown in Fig-

ure 8.1(a), in which length is 190mm, width is 136mm and height is 14mm. For the

further discussion, two faces visible from the angle shown in the figure are named as

“cover” and “opening A” respectively, and one face invisible from this angle is named

as “spine”. The disc lies at the bottom of the case and the distance between the disc

and cover is about 10mm.

The material of the DVD case can be identified by the symbol on the case spine which

is shown in Figure 8.1(b). That symbol is a recycling triangle with a number 5 in it

and two letters “PP” under it. This symbol is called SPI Resin Identification Code

which is set to allow efficient separation of different polymer types for recycling. The

number and the letters in the symbol is variable depending on the type of the plastic.

Commonly, the symbol on a DVD case is that shown in Figure 8.1(b), which represents

the material polypropylene [85]. According to the study by Riddle [86], it is found that

at 9.4GHz, polypropylene’s dielectric constant is about 2.3 with a very low loss tangent.

We did not find any literature discussing the dielectric constant and loss tangent of

polypropylene in the UHF band, but we believe they would be close to the figures

shown above.

The top section view and the cross section view of a regular DVD are shown in Fig-

ures 8.2(a), and 8.2(b) respectively. The top view in Figure 8.2(a) is very familiar to

people, hence, it is not described in detail. Instead the cross section in Figure 8.2(b) is

what we are really concerned about. The whole thickness of the optical disc is 1.2mm

and the polycarbonate substrate makes up the majority of the disc material. The data

Page 177

8.2 Parameters of a Packaged DVD Product

14mm

190mm

136mm

Opening A

Spine

Cover

(a) A regular DVD case (b) SPI Code on the case spine

Figure 8.1. The structure of a regular DVD case and the SPI code on it.

layer takes the form of a grooved metal reflective layer. The grooved surface in the

polycarbonate is impressed by stamping in a moulding machine and then metal is

sputtered or condensed on to the surface as the reflective layer for a laser beam. The

thickness of the sputtered metal layer is about 30nm [87]. The metal material is usually

a good conductor such as gold, silver, aluminium or aluminium alloy. The last two

kinds of material are commonly used.

End o

f Data carry

ing area

118m

m

Disc o

uter ed

ge

120m

m

45mm

15mm

(a) Disc top view (b) Disc cross view [84]

Figure 8.2. The structure of a regular DVD disc.

As shown in Figure 8.2, the polycarbonate makes up the majority of the disc. Poly-

carbonate is one kind of plastic material commonly used as a packaging material. Sig-

nificant research has been done on the electromagnetic characteristics of this material.

Grosvenor et al. [88] reported that the real component of the relative permittivity of

polycarbonate is nearly constant from 1GHz to 6GHz (at 1GHz, it is about 2.88). The

imaginary component is positively related to the frequency and at 1GHz, it is about

Page 178

NOTE: This figure is included on page 178 of the print copy of the thesis held in the University of Adelaide Library.


0.035. Riddle et al. [86] stated that at 11GHz, both the real component of the relative

permittivity and the loss tangent are related to the temperature. However, in the range

of the temperature with which we are concerned (from -300C to +400C), the relative

permittivity is held somewhat constant.

Besides polycarbonate, DVDs consist of some very thin layers, such as label or print-

able surface, lacquer or adhesive layer and metal reflective layer. Since all except the

metal layer are dielectrics and very thin compared with the electromagnetic wave-

length in the UHF RFID frequency band, they are neglected. Although the thickness

of the metal layer is much less than a skin depth at the interrogation frequency, its

surface resistivity shows that it can still act as an effective metallic boundary and will

differently influence the propagation of vertically and horizontally polarised electro-

magnetic fields propagating tangentially to its surface. Hence, the effects from the

metal layer to the incident waves in various polarisations are worth studying, and are

discussed in Section 8.3.

8.3 Theoretical Analysis and Simulation Verification of

the Effect on a Uniform Plane Wave from a Thin

Metal Film

8.3.1 Surface Resistance of a Thin Metal Film

A thin metal film here is meant to be a film of which the thickness is much less than

the skin depth. The aluminium layer thickness in a DVD is only 30nm. The ratio of

30nm to the skin depth of aluminium at about 1GHz is approximately 1%. Hence,

the aluminium layer in a DVD disc is a very thin metal film at the UHF RFID oper-

ating frequency. It is worth calculating the surface resistance of this thin metal layer.

Knowing the surface resistance, the reflection coefficient when a uniform plane wave

is perpendicularly incident on the thin layer can be addressed. The electrical resistivity

ρ of one metal material viz. aluminum, is ρ=28.2nΩm. The surface resistance Rs of a

Page 179

8.3 Theoretical Analysis and Simulation Verification of the Effect on a UniformPlane Wave from a Thin Metal Film

metal material can be derived by (8.1).

Rs = ρ/t (8.1)

where t is the thickness of the metal. For the aluminium layer in a DVD with a layer

thickness of 30nm, the surface resistance derived by (8.1) is 0.94Ω.

The case that a uniform plane wave perpendicularly incident on an infinite 30nm alu-

minium film can be analysed by a transmission line model, as shown in Figure 8.3, in

which the surface resistance of this aluminium film is placed in shunt across the line.

RsZ0 Z0

Figure 8.3. Transmission line model of a uniform plane wave perpendicularly incident on an

infinite aluminium metal film. Z0 is the characteristic impedance of vacuum, which

is about 377Ω.

According to the transmission line model shown in Figure 8.3, the reflection coefficient

of the 30nm aluminium layer can be calculated by inserting the surface resistance of the

30nm aluminium layer Rs=0.94Ω and the wave impedance of free space Z0=377Ω into

(8.2). The reflection coefficient Γ derived is -0.995, which means there is a large amount

of reflection and only a small transmission occurring at the aluminium film. Both the

small surface resistance and large (in magnitude) reflection coefficient indicate that this

thin aluminium film will still perform effectively as a metal layer which is much thicker

than the skin depth. It is concluded that an incident wave in which the electric field

is tangential to the thin aluminium film will not propagate well along its propagation

direction when the wave meets the aluminium film, but an incident wave in which the

electric field is perpendicular to the thin film can propagate well.

Γ =Rs‖Z0 − Z0

Rs‖Z0 + Z0(8.2)

The above theoretical analysis has also been verified by the simulation software An-

soft HFSS. In HFSS, a very thin metal film can be assigned to be a layered impedance

boundary condition, which actually regards the thin film to be a surface impedance

Page 180


value. HFSS can not only give the surface resistance of the thin metal film but also

the surface reactance. A 30nm flat aluminium layer’s surface impedance at 923MHz1

derived by the software is 0.875+j7.3Ω. The frequency 923MHz is chosen here because

it is the centre of the 6MHz UHF RFID band specified in Australia. We are not com-

menting on the inequality of the real and imaginary parts. The real component of this

impedance is very close to the 0.94Ω calculated theoretically above.

The simulation method and results are discussed as follows. A cube radiation bound-

ary is built in a coordinate system as shown in Figure 8.4 in which the cube length

is 325mm (the wavelength of 923MHz wave in free space), the cube centre is set to

be origin of the coordinate system. The material inside the cube is set to be vacuum.

The radiation boundary here works as the boundary allowing the wave inside to pass

through the boundary and radiate infinitely far away from that surface or in other

words, there is no significant reflection from the radiation boundary. A square of which

the length is the same as the cube length and of which the centre is also at the origin

of the coordinate system is modeled on the xy plane. The boundary condition of this

square is assigned to be the layered impedance boundary as discussed above to imitate

a 30nm aluminium layer.

Figure 8.4. Simulation model of the square aluminium film.

1This surface impedance is held somewhat constant along a few GHz frequency band.

Page 181


Three forms of uniform plane waves at 923MHz are added at a 81.25mm distance (one

quarter wavelength) above the square respectively. The three forms of incident wave

are 1) the propagating direction k is perpendicular to the film and the electric field Ei is

tangential to the film, k=(0,0,-1), Ei=(1,0,0); 2) both the propagating direction k and the

electric field Ei are tangential to the film, k=(0,1,0), Ei=(1,0,0); and 3) the propagating

direction k is tangential to the film and the electric field Ei is perpendicular to the film,

k=(0,1,0), Ei=(0,0,-1). After the simulation processing, the magnitude of r.m.s phasor of

the total electric field (incident field plus scattered field) distribution in the xz plane at

923MHz are shown in Figure 8.5. Please note that the scales beside each sub-figure for

representing the magnitude of the electric field are different. Although the film is not

infinite, but a square of which the length is equal to the wavelength at 923MHz, we

still believe the field distribution above and underneath the infinite film will resemble

periodically that the square film used here.

k

E=1V/m

(a) k=(0,0,-1), Ei=(1,0,0)

kE=1V/m

(b) k=(0,1,0), Ei=(1,0,0)

k

E=1V/m

(c) k=(0,1,0), Ei=(0,0,-1)

Figure 8.5. Total electric field distribution shown in the xz plane of the simulation on the

square aluminium film. The aluminium film is represented by the black line in the

middle of each sub-figure.

According to Figure 8.5(a), it is found that most of the incident wave cannot penetrate

the aluminium film, instead it is reflected. That is the reason why the magnitude of

the electric field is nearly zero on the surface of the film and it is doubled in a quarter

wavelength above the film. Figure 8.5(b) indicates that the wave can propagate well if

it is far away from the film (the distance between the wave and the film should be larger

than one wavelength), however, when it gets close to the film, the magnitude of the

electric field drops dramatically. In the case shown in Figure 8.5(c), it is noted that the

Page 182


incident wave can propagate un-attenuated across the film. The film does not affect the

propagation of the incident wave. In conclusion, if the wave is expected to propagate

well, the electric field of the incident uniform plane wave should be orthogonal to the

conducting sheet. The conclusion made by simulation complies with the conclusion

by theory presented before.

8.3.2 Simulation on a DVD Disc

The effects on different types of incident waves by a 30nm thick aluminium square

are discussed in Subsection 8.3.1. However, the shape of the DVD disc as shown in

Figure 8.2(a) is different from the square sheet of which the length is one wavelength.

Hence, in this subsection, further simulation is done by replacing the square metal

sheet in Figure 8.4 by the real disc aluminium layer shown in Figure 8.6. The roughness

of the disc has been considered (the roughness comes from the grooved configuration

of the disc shown in Figure 8.2(b)), and this disc layer is also illuminated by the three

forms of incident plane wave applied in Subsection 8.3.1: 1) k=(0,0,-1), Ei=(1,0,0); 2)

k=(0,1,0), Ei=(1,0,0); and 3) k=(0,1,0), Ei=(0,0,-1). After the simulation processing, the

magnitude of r.m.s phasor of total electric field (incident field plus scattered field) dis-

tribution in the xz plane at 923MHz are shown in Figure 8.7. Please note that the scales

beside each sub-figure for representing the magnitude of the electric field are different.

For the third case shown in Figure 8.7(c), in which the electric fields of incident wave

are orthogonal to the disc, the simulation result is very close to that of the simulation of

the square sheet which indicates that no matter what the shape of the very thin metal

film is, it does not affect the propagation of the incident wave which electric field is

perpendicular to the film. However, for the first and second types of incident waves,

in which the electric fields are tangential to the disc, the simulation results are different

from those of the simulations on a square sheet. There occurs some resonance between

the incident wave and the disc film. The resonance is actually caused by the size of

the disc. As shown in Figure 8.2(a), the diameter of the metal layer in the disc is about

120mm which is very close to the half wavelength at 923MHz. The resemblance makes

the aluminium layer in the disc work as a very fat half wavelength dipole under the

Page 183


Figure 8.6. Simulation model of the aluminium film in the disc.

k

E=1V/m

(a) k=(0,0,-1), Ei=(1,0,0)

kE=1V/m

(b) k=(0,1,0), Ei=(1,0,0)

k

E=1V/m

(c) k=(0,1,0), Ei=(0,0,-1)

Figure 8.7. Total electric field distribution shown in the xz plane of the simulation on the

aluminium film in the disc. The disc film is represented by the black line in the middle

of each figure.

incidence of properly polarised waves. Fortunately, because the aluminium layer is in

its transverse directions very fat, we expect that the Q factor of this resonance is very

low. It is also verified by the simulation software HFSS that the maximum electric fields

on the edge of the disc excited by the incident wave are held somewhat constant along

a wide frequency band (400MHz centered at 923MHz). Therefore, this resonance can

be ignored in these two cases of incident waves in which the electric field is tangential

to the disc.

Page 184


8.4 Investigation of Tag Labelling Method

In this section, the method of a tag labelling on a packaged DVD is investigated by

experiments. The experiments are conducted by varying the position of the tag on the

packaged DVD and comparing the reading ranges of the tag in these positions. The

experimental results are interpreted by the theoretical analysis in Section 8.3.


Besides the variation of the labelling position of the tag on the DVD, the experiments

are also conducted to obtain reading ranges of the tag by varying the other two factors

which are the orientations of the DVD case in relation to propagating wave from the

reader antenna, and the reader antenna’s polarisation.

Before introducing the experimental procedures and the results, the facilities and pro-

tocols deployed in the measurement are given. The tag used here is a common com-

mercial UHF RFID tag, which overall size is 95mm×9mm. The protocol employed is

Class 1 Generation 2. The reading range of the tag in free space is about 6m under

Australian UHF RFID regulations. The tag is approximately an electric dipole with an

inductive loop placed near the connection to the chip to tune both the chip and dipole

capacitances.

The RFID reader (Model ID ISC.LRU2000) by FEIG Electronics and 8 dBi gain linearly

polarised reader antenna (Model S9028P) by Cushcraft Corporation, were employed

Page 185


here to detect the tag attached on a packaged DVD. All the following measurements

were operated under the Australian UHF RFID standards and regulations.

In order to simplify this investigation, instead of tagging multiple packaged DVDs and

piling up them into a stack, only one tag was attached on a single packaged DVD and

this packaged DVD is placed in the shielding tunnel, shown in Figure 8.8. The environ-

ment inside the shielding tunnel is regarded as equivalent to free space. Appropriate

experiments were conducted to obtain various reading ranges of the tag on the DVD

case by varying three factors. As indicated above, the three factors are: (i) the position

of the tag on the DVD case; (ii) the orientation of the DVD case in relation to the prop-

agating wave from the reader antenna; and (iii) the reader antenna’s polarisation. The

details of the experiments are listed in the following three subsections categorised by

the first variable: the position of the tag on the packaged DVD.

8.4.1 Tag Lying on the Case Cover

First, two situations when the tag lies on the case cover are tested. The two situations

are tag lying at the bottom of the case cover and tag lying in the middle of the case

cover as shown in Figure 8.9. The readability of the tag is given in Table 8.1.

Tag Opening A

Spine

Cover

(a)

Tag

Opening A

Spine

Cover

(b)

Figure 8.9. Tag lying on the case cover.

Necessary interpretation for understanding Table 8.1 is discussed as follows. As men-

tioned before, there are three variables in the reading range experiments. They are

the position of the tag on the DVD case, the orientations of the DVD case in relation

to propagating wave from the reader antenna and the reader antenna’s polarisation.

Page 186


Table 8.1. Reading range test results of the tag shown in Figure 8.9.

(a)↑ (a)→ (b)↑ (b)→Opening A >3m <0.2m 0.2m-1m <0.2m

Spine 0.2m-1m 0.2m-1m <0.2m <0.2m

These variables are all expressed in Table 8.1. The symbols “(a)” and “(b)” denote

the position where the tag is attached on the DVD case as shown in Figure 8.9(a) and

Figure 8.9(b) respectively. The symbols “↑” “→” represent the reader antenna’s polar-

isation; “↑” means the vertical polarisation. “→” means the horizontal polarisation.

In the first column of Table 8.1, some DVD case face names are listed. Those face

names are used to distinguish the orientation of the DVD case in relation to the direc-

tion of the propagating wave from the reader antenna. Particularly, the information

in a row beginning with a face name is obtained when this face is perpendicular to

the propagating wave from the reader antenna. The situation when the propagating

wave is perpendicular to the case cover is not investigated, since it was concluded in

Section 8.3 that the propagating wave hardly penetrates through the conducting sheet

of the DVD. The DVD case always stands on a piece of polystyrene foam when it is

tested. “Stands” means that no matter which case face is perpendicular to the propa-

gating wave, the DVD case is always placed to make the longer side of the face vertical.

The entries in the table stand for the approximate reading range of the tag. The original

testing results have been given in Appendix C. According to the above interpretation,

the cell identified by the intersection of row “Spine” and column “(a)↑” shown con-

tent “0.2m-1m” can be understood as that the reading range of the tag shown in Fig-

ure 8.9(a) is between 0.2m and 1m when the reader antenna’s polarisation is vertical

and the case spine is perpendicular to the propagating wave and the longer side of the

spine is vertical. The interpretation above is also suitable for understanding Tables 8.2

and 8.3 corresponding to the relevant figures.

Referring to Table 8.1, by observing row “Opening A”, it is found that the change in

polarisation affects the reading range dramatically. This is easy to understand because

both the tag antenna and the reader antenna are linearly polarised, and they have to

be matched well to obtain long reading range. The reading ranges in row “Spine”

Page 187


are all very limited, since when the case spine is perpendicular to the propagating

wave, the axis of the tag (also the polarisation of the tag) shown in Figure 8.9 is always

orthogonal to the reader’s polarisation no matter whether the reader’s polarisation is

vertical or horizontal. By comparing the columns with symbol “(a)” and the columns

with symbol “(b)”, it is found that the tag placement shown in Figure 8.9(a) performs

much better than (or at least equally with) the tag placement shown in Figure 8.9(b).

The degradation of the tag shown in Figure 8.9(b) is caused by the disc underneath,

which as is concluded in Section 8.3 can still be regarded as a good metal sheet even

though it is very thin.

8.4.2 Tag Lying on the Case Faces: Opening A and Spine

Besides the tag lying on the case cover, it can also lie on the case opening A and spine as

shown in Figure 8.10. The test results of the tag reading range lying on these two case

faces are given in Table 8.2. The differences in reading range are mainly caused by the

different polarisation matching conditions which were discussed in Subsection 8.4.1.

Tag Opening A

Spine

Cover

(a)

Tag

Opening A

Spine

Cover

(b)

Figure 8.10. Tag lying on the case faces: opening A and spine.


(a)↑ (a)→ (b)↑ (b)→Opening A >3m <0.2m 0.2m-1m 0.2m-1m

Spine <0.2m 0.2m-1m 2m-3m <0.2m

Page 188


8.4.3 Tag Folded on the Case Faces: Opening A and Spine

The tag folded on the DVD case instead of lying on the case is discussed here. There are

mainly two types for the tag folded on the DVD case which are shown in Figure 8.11.

The test results of the tag reading range are given in Table 8.3.

Tag Opening A

Spine

Cover

(a)

Tag Opening A

Spine

Cover

(b)

Figure 8.11. Tag folded on a DVD case.


(a)↑ (a)→ (b)↑ (b)→Opening A <0.2m 1m-2m <0.2m 1m-2m

Spine <0.2m 1m-2m <0.2m 1m-2m

By observing the results in Table 8.3, it can be seen that the two tag placements shown

in Figure 8.11 are only sensitive to the horizontal polarisation. This result can be un-

derstood by resolving the electric field along various sections of the tag. The tag part

on the case opening A in Figure 8.11(a) and the part on the case spine in Figure 8.11(b)

do couple to the horizontal component of electric field, with an effective length equal

to the minimum dimension of the case. The other parts of the tag on the front and back

cover faces are orthogonal to the horizontal electric field, so they do not couple to this

field. They indeed couple to the vertical electric field, however, when they are cou-

pled, the two sections couple with opposite senses, and furthermore the parallel parts

of the tag create large shunt capacitance which parallels the intentionally designed an-

tenna input impedance. The two facts make the tag almost unreadable as shown in

the columns with the symbol ↑ in Table 8.3 when the interrogating wave is vertically

polarised.

Page 189


This shunt capacitance can be reduced by staggering the two parallel components of

the tag to be not parallel as shown in Figure 8.12 so that the tag can couple not only

to the horizontal polarisation but can also couple to the vertical polarisation. We ex-

pect these labelling methods will obtain better performance when the interrogating

wave is circularly polarised, however, the experiments show that they are not sur-

prisingly good compared with the labelling methods shown in Figure 8.11. Moreover,

their labelling methods are inconvenient. Therefore, those two labelling methods are

abandoned at the beginning.

Tag Opening A

Spine

Cover

(a)

Opening ATag

Spine

Cover

(b)

Figure 8.12. Tag staggered on a DVD case.

The conclusions of this section are:

1. Once the tag is placed above the disc top surface, its reading range will be de-

graded because of the aluminium layer underneath in the disc. Hence, the tag

labelling method in Figure 8.9(b) is abandoned.

2. The whole tag shown in Figure 8.9(a) can couple to the incident uniform wave

when the electric field of the incident wave is parallel to the tag axis which also

indicates that the electric field is tangential to the disc. This tag obtained a very

long reading range (>3m) on a single DVD in free space as shown in Table 8.1.

However, when a large number of packaged DVDs are piled up in a stack, the

penetrability of this incident wave into the DVD stack should be considered. Ac-

cording to the analysis in Section 8.3, the 30nm aluminium film in a DVD disc is

still an effective metal layer which diminishes the tangential electric field near the

layer. As a result, it is predicted that this type of incident wave cannot transmit

well in a DVD stack and effectively couple to the tag.

Page 190


3. The tags shown in Figures 8.11(a) and 8.11(b) can couple to the incident uniform

plane wave in which the electric field is orthogonal to the DVD disc. This type of

incident wave is, as concluded in Section 8.3, considered to be capable of pene-

trating deeply into the DVD stack. However, the effective length coupling to the

orthogonal electric field of these two tag placements is only equal to the mini-

mum dimension of the DVD case which is much less than the effective length of

the tag shown in Figure 8.9(a). Hence, a tradeoff has to be made between the inci-

dent wave penetrability in the stack and the effective length of the tag placement

on the case. In addition, to prevent the tag from blocking the wave from transmit-

ting deeply into the stack, the case face where the coupling component of the tag

antenna is attached should not be placed so as to be orthogonal to the incident

wave. In other words, the area of the tag meeting with the incident wave should

be the minimum. Hence, the two tag placements in Figure 8.10 are abandoned.

All in all, once a large number of packaged DVDs is piled up, a tradeoff has to be made

between the wave penetrability related to the polarisation type and the effective length

of the tag which is related to the labelling method of the tag on the DVD. Three testing

schemes, which are shown in Figure 8.13, are selected to investigate this tradeoff. In

these testing schemes, each packaged DVD placement in a DVD stack, and the tag la-

belling method on it in relation to the interrogating wave are defined. Particularly, for

the testing scheme ”1” shown in Figure 8.13(a), the DVD is illuminated by a linearly or

circularly polarised propagating wave, in which the propagating direction is perpen-

dicular to the case opening A and the electric field of the linearly polarised antenna is

along the tag axis. Similarly, the other testing schemes can be understood by using of

Figures 8.13(b) and 8.13(c).

8.5 DVD Detection in a Stack

In this section, a large number of packaged DVDs is piled up into a stack according

to the requirements of each testing scheme introduced at the end of Section 8.4. The

number of the labelled DVDs in the stack is increased step by step to investigate the

Page 191


E

k

(a) Testing scheme 1

E

k

(b) Testing scheme 2

E

k

(c) Testing scheme 3

Figure 8.13. Three selected testing schemes. The red arrow represents the electric field of the

linear polarisation, the blue arrow denotes the incident direction.

probability of the tag detection of each testing scheme. Two testing schemes with high

probability of detection are identified for the further validation discussed in Section 8.6.

8.5.1 Testing Strategy and DVD Stack Description

In Subsection 8.1.3, some general perspectives about addressing the problem of read-

ing 2000 DVDs are discussed. One of the perspectives is that people cannot read all the

tags on 2000 DVDs in a stack by placing the reader sufficiently far from the stack for all

tagged DVDs to fall within the main lobe of the reader antenna radiation pattern. This

testing strategy is shown in Figure 8.14(a)

Instead, the testing strategy we adopt is to place a reader antenna against the surface of

the stack as shown in Figure 8.14(b). Certainly, the reader cannot read all the tags from

one position. But by moving the reader antenna along two dimensions of this surface

as shown by the green arrows in Figure 8.14(b), half of the tags in the stack could be

detected if the power carried by the wave is sufficient to excite the tags in the middle.

In addition, the other half could be identified if we were prepared to read from the

other face.

As mentioned at the beginning of this chapter, this project aims to detect 2000 DVDs

in a stack by a UHF RFID system. In the following, the shape of the 2000 DVD-stack is

described.

Page 192


DVD stack

ReaderAntenna

(a) Abandoned testing strategy

DVD stack

ReaderAntenna

(b) Adopted testing strategy

Figure 8.14. Testing strategy illustration.

As indicated by the dimensions of a packaged DVD shown in Figure 8.1(a), the pack-

aged DVD has a volume of 0.136m by 0.190m by 0.014m, i.e. 0.00036176 cubic meters.

2000 packaged DVDs have a volume of 0.724 cubic meters. A cube of side 0.898m

would have that volume. The distance 0.898m is approximately 64 times the DVD case

thickness (0.014m) and 5 times DVD case length (0.190m) and 7 times DVD case width

(0.136m). In total, the number in such a DVD stack would be 64 by 5 by 7 i.e. 2240

DVDs.

By implying the three testing schemes in Figure 8.13 to the 2240 DVD stack, three forms

of the testing strategy is obtained. For testing schemes “1” and “3”, the reader antenna

is set in front of the surface formed by all of the case openings, there are DVDs 5 deep

as shown in Figure 8.15(a) and Figure 8.15(c) respectively; for testing scheme “2”, the

reader antenna is set in front of the surface forming by all of the case spines, there are

DVDs 7 deep as shown in Figure 8.15(b). Only the tagged DVDs at the bottom of the

stack are shown in Figure 8.15. Hence if either a 3-deep DVD stack in Figure 8.15(a)

and Figure 8.15(c) or a 4-deep DVD stack Figure 8.15(b) is successfully read, we will

expect that all the tagged DVDs in the stack could be read if we were prepared to read

from both faces.

As analysed above, the depth dimension of the stack is the most significant dimension

to make all the tags in the stack readable. The labelled DVDs along the other dimen-

sions can be detected by moving the reader antenna along these dimensions. Hence,

it is not necessary to do an experiment on a 2240 packaged DVD stack. Instead, a

packaged DVD stack with a much smaller surface from the view of the reader antenna

Page 193


Interrogator

(a)

Interrogator

(b)

Interrogator

(c)

Figure 8.15. Three forms of testing a DVD stack in terms of the three testing schemes


but a greater depth than half the depth of the 2240 DVD cubic stack as described be-

fore is adequate to imitate part of the 2240 packaged DVD stack. If all the packaged

DVDs in the small stack can be read successfully, we expect 2240 packaged DVDs can

be read by moving the reader along the two dimensions shown by the green arrows in

Figure 8.14(b).

In order to ensure there is no reflection from metal surfaces in the lab and no diffrac-

tions occur to reach the tags at the back of the stack, the small DVD stack is placed

in the aperture surrounding by absorbing foams shown in Figure 8.16, so that if the

tag attached on a DVD at the end of the stack can be read, we believe the energy for

exciting the tag only comes from the incident wave after attenuation crossing DVDs in

front of and beside this tag.

The absorbing foam is manufactured by Emerson & Cuming company for the fre-

quency range from 600MHz to 4GHz. The absorbing foam can achieve a minimum

of 22dB return loss around 1GHz. The reflectivity performance can degrade for off-

normal incidence and at difference rates for different polarisations [89]. Except for the

foam at the end of the aperture, other foam pieces definitely cannot get 22dB reflection

loss since the incident angle on these foam pieces will be very large which results in

relatively large reflection. These reflections may cause some multiple path propagation

in this aperture which may cause some weak field positions and strong field positions

in the aperture. The reflections in the aperture are investigated in Appendix D. It is

concluded that the reflection caused by the absorbing foam can be ignored.

Page 194


300

49

6

610

61

0

114

11

411

4

(a) Schematic diagram of the aperture, unit: mm (b) The front view of the real aperture

Figure 8.16. Aperture structure illustration.

For the aperture shown in Figure 8.16, there are two types of DVD stack. First, for

testing scheme ”2”, there are 160 DVDs 20 wide by 2 high by 4 deep filled in the aper-

ture as shown in Figure 8.17(a). Secondly, for testing schemes “1” and “3”, there are

180 packaged DVDs 20 wide by 3 high by 3 deep filled in the aperture as shown in

Figure 8.17(b). The depth of the two types of the stack is deeper than half that of the

2240 DVD stack described before.

8.5.2 Single Tagged DVD Film in a DVD Stack

In this subsection, only one packaged DVD at the end of the DVD stack in the aperture

is labelled. The reading range of this tag labelled on the DVD is tested using the three

testing schemes selected at the end of Section 8.4 respectively. The shape of the DVD

stack, either the one shown in Figure 8.17(a) or the other shown in Figure 8.17(b), used

in the experiment depends on the requirement of each testing scheme as introduced in

Subsection 8.5.1.

The testing results are given in Table 8.4. The numbers in the first row of Table 8.4

represent the three selected testing schemes. In the first column, the reader antennas

Page 195


(a) DVD stack for testing scheme “2” (b) DVD stack for testing schemes “1”

and “3”

Figure 8.17. Two types of DVD stack in the aperture.

used in the experiment are given. They are the circularly polarised antenna with gain

5.7dBi and Model ISC.ANT.U250/250-FCC manufactured by FEIG Electronics com-

pany and the linearly polarised antenna with gain 8dBi (Model S9028P) manufactured

by Cushcraft Corporation. According to the experiments, it is found that all the three

testing schemes are capable of reading the tag on the DVD at the end of the stack and

the reader antenna does not need to be very close to the front side of the stack. There

can be a distance between them. The distance depends on the radiating power from

the reader and the testing scheme applied. Therefore, the reading range of the tag in

the stack consists of two parts. One is the reading range inside the DVD stack, the

other one is the reading range outside the DVD stack, for which propagation is in free

space.

No matter which testing scheme is deployed, once the tag at the end of the stack can

be read, we will conclude that it might be possible to identify all the DVDs in the stack

if they were all labelled, but this conjecture will have to be tested. Hence, the read-

ing ranges outside the stack become critical to judge this possibility, but the reading

ranges inside the stack, which are all approximately equal, are not. That is the reason

Page 196


why only the outside reading range of the tag tested under the three testing schemes

are given in Table 8.4. Please note that all the tests are under 1.26W EIRP radiation

power from the reader antenna. This radiation power is much lower than the maxi-

mum power limitation (4W EIRP) in Australia and the USA Some entries in the table

are intentionally left blank, because of the obvious polarisation mismatching condition

between the tag antenna and the reader antenna in the relevant testing scheme.

Table 8.4. Outside reading range of the tag at the end of the DVD stack.

1 2 3

5.7dBi Circular polarisation 205mm 1210mm 675mm

8dBi Linear polarisation ↑ 150mm

8dBi Linear polarisation → 840mm 570mm

By comparing the outside reading ranges obtained by the circularly polarised antenna

and these by the linearly polarised antenna, it is apparent that the former has a gener-

ally better performance than the latter. Theoretically, this superior performance should

not happen, since the tag placements in these schemes only couple to one kind of lin-

ear polarisation as concluded in Section 8.4. Possibly, that is because the circularly

polarised antenna is matched to the reader well since they are manufactured by the

same company. Based on the observation and assumption, in the following experi-

ments, only the circularly polarised antenna is deployed.

The comparison between the column “1” and the other two columns indicates that,

even though the effective length of the tag placement in testing scheme “1”, in which

the tag couples to vertical electric field, is much larger than those of the other tag place-

ments, which can couple to the horizontal electric field, the reading range of the tag in

testing scheme “1” is much less than the other testing schemes. It is concluded that the

penetrability of the incident wave, which we have concluded is related to the polari-

sation type of the interrogating wave, plays a more significant role than the effective

length of the receiving tag antenna in obtaining long reading range.

By comparing figures in Table 8.4, the readability of each testing scheme is ranked as

“2”, “3” and “1”. The experiment in this subsection was conducted by labelling one

DVD at the back of the stack. It does not take into account the negative effects when

Page 197


multiple DVD are labelled: (1) the mutual coupling among the tags, (2) the weakened

effect of the interrogating wave after passing through the tags in front. In the following

discussion, these three testing schemes are further investigated by labelling multiple

tags in the DVD stack. We would not drop off the testing scheme “1” at this stage just

because it has the worst performance among the three. We keep testing this scheme

since this is the best labelling method among the three in terms of attachment conve-

nience.

8.5.3 Multiple Tag Detection in a DVD Stack

In this subsection, multiple tags will be placed on DVDs in the stack and tested un-

der the testing schemes “1”, “2” and “3”, which details can be found at the end of

Section 8.4.

In order to distinguish multiple tags in the DVD stack, the code in each tag has to

be different. However, the commercial tags we are using are coded the same initially.

Hence, they have to be programmed before labelling the DVDs, which can be done by

the FEIG reader introduced before.

The tags comply with the EPC C1G2 protocol as mentioned before, so the code in the

tag is an Electronic Product Code (EPC). The EPC code is a globally unique identifi-

cation scheme designed to uniquely identify all physical objects and aggregations of

objects. In order to achieve that, the EPC should be sufficiently large and well organ-

ised to enumerate all the existing objects in the world. Now, the EPC is a 96 bit code

consisting of four distinct, hierarchical partitions: version number, domain manager,

object class code and serial number. More details of the EPC can be found in [90].

For the purpose of clear expression, several definitions are given. As shown in Fig-

ure 8.17, there are two types of DVD stack that will be used in the following experi-

ments depending on the testing scheme. The one shown in Figure 8.17(a) contains 160

DVDs 20 wide by 2 high by 4 deep. In terms of the depth, the stack can be divided

into four levels as shown in Figure 8.18(a) (the levels are counted from the front side of

the stack). Each level depth is equal to the width of the DVD case which is 136mm. In

Page 198


terms of the height, the stack can be divided into two floors as shown in Figure 8.18(b)

(the floors are counted from the bottom of the stack). Each floor height is equal to

the length of the DVD case which is 190mm. Hence, for the DVD stack shown in Fig-

ure 8.17(a), 160 DVDs composed of 4 levels (4×40=160) or 2 floors (2×80=160). This

DVD stack is used for examining the testing scheme “2”.

Level 2

Level 1

Level 4

Level 3

(a) Stack level division

Floor 2

Floor 1

(b) Stack floor division

Figure 8.18. The level and floor division of the stack shown in Figure 8.17(a).

Similarly, for the stack shown in Figure 8.17(b), there are 180 DVDs arranged into 3

levels (3×60=180) or 3 floors (3×60=180). This DVD stack is used for examining the

testing schemes “1” and “3”.

Only the circularly polarised reader antenna is used, since as observed in Subsec-

tion 8.5.2 this one has better performance. In addition, the reader antenna is placed

just against the front side of the stack and the radiation power is set to be the allowed

maximum power (4W EIRP).

For economy of label usage, and because the adjacent foam may disturb reading, and

because the circularly polarised antenna’s main lobe cannot cover the front surface of

the DVD stack once the antenna is close to the stack, not all the DVDs in the stack are

labelled. For the width dimension of the DVD stack, 20 DVDs in parallel can be fitted

in as shown in Figure 8.17, but only 14 DVDs in the middle are labelled and for the

height dimension, 2 floors of DVDs counting from the bottom of the stack are labelled.

Hence, in a level there are 28 DVDs that are labelled. In the following discussion, when

it is said that a level’s DVDs are labelled, that does not mean all of the 40 DVDs in the

Page 199


level of the stack shown in Figure 8.17(a) or the 60 DVDs in the level of the stack shown

in Figure 8.17(b) are labelled, but only 28 DVDs in this level are labelled.

The testing schemes “1”, “2” and “3” are all examined by the same method of which

the flow chart is given in Figure 8.19.

Label the last level DVDs by the testing scheme&

detect them by reader

N

Readable ratio>95%

N M<

Yes

NoTesting scheme fails

NoTesting scheme successes

Yes

N N= +1

Figure 8.19. Flow chart of the method examining the testing schemes “1”, “2” and “3”

respectively.

As shown in Figure 8.19, firstly the DVDs in the last N levels (N is given the value 1

initially) are labelled and stacked according to the requirement of the testing scheme

being examined. Hence, at the beginning only 28 DVDs in the level at the end of the

stack are tagged. Then, the reader is used to detect these tags. The ratio of the number

of readable tags to the total number of the tags in the stack is defined as the readable

ratio. If this ratio is larger than 95% which means most of the tags in the stack can

be detected, the last two level DVDs will be labelled and tested again. The reason

why 95% instead of 100% is chosen because we are seeking good but not yet perfect

performance. Perfect performance will be sought in Section 8.6. The process will not

stop until the readable ratio goes under 95% which indicates the testing scheme is not

qualified to be used or all the levels in the stack are labelled and most of the tags can

Page 200


be read which indicates the testing scheme is accepted to detect 2240 DVDs in a stack.

M in the flow chart denotes the total number of the levels in the stack. For the stack

shown in Figure 8.17(a), M=4 and for the stack shown in Figure 8.17(b), M=3.

The results are given in following three itemisations for the three testing schemes re-

spectively.

• Testing scheme 1

This testing scheme fails in the first round examination i.e. when N=1. Less than

half of the tags among the 28 labelled tags can be detected.


When N=1, all of the 28 tags can be detected, in other words the readable ratio

is 100%. When N=2 i.e. 56 DVDs in the last two levels in the stack are tagged,

the readable ratio is 98%. Only one tag placed on the edge of the stack is missed.

When N=3 i.e. 84 DVDs in all three levels of the stack are labelled (the testing

scheme 3 adopts the DVD stack shown in Figure 8.17(b) which contains three

levels in total.), the readable ratio is 97.6%. Only two tags in the last level of the

stack are missed.


When N=1 and 2, the readable ratio is 100%. When N=3 i.e. 84 DVDs in the

last three levels in the stack are tagged, the readable ratio is 98.8%. Only one tag

placed on the edge of the stack is missed. When N=4 i.e. 112 DVDs in all four

levels of the stack are labelled (the testing scheme 2 adopts the DVD stack shown

in Figure 8.17(a) which contains four levels in total.), the readable ratio is 98.2%.

Only two tags in the last level of the stack are missed.

According to the above testing results, the testing scheme “1” is totally abandoned and

the other two testing schemes are further verified in Section 8.6.

Page 201

8.6 Further Validation


The testing schemes “2” and ”3” have been verified as having a good probability of

detecting a large number of DVDs, up to 2000, in Section 8.5. However, the validation

in Section 8.5 is not very practical, since it did not consider the real requirement of the

industry in terms of the method of packaging and stacking a large number of DVDs,

and the number of tagged DVDs in the experiment in Section 8.5 is very limited, only

about 5% of the ultimate number of tagged DVDs (2000). Hence, in order to make

this validation more solid, the testing schemes are examined further in terms of four

aspects.

1. Taking the protocol capacity of anti-collision into account once the number of

DVDs in a stack goes up to 2000 or even more.

2. Expanding the number of DVDs in a testing stack.

3. Adopting a more realistic experimental environment.

4. Considering the method used by industry in stacking a large number of packaged

DVDs.

As a result, in Subsection 8.6.1, the Q parameter in EPC C1G2 protocol used for the pur-

pose of anti-collision is introduced. Then, in Subsection 8.6.2, the method of packaging

and stacking a large number of DVDs in industry is investigated. Then experiments

are conducted after expanding the number of DVDs in the testing stack and adapting

more realistic testing environment in Subsection 8.6.3.

8.6.1 Q Parameter in EPC C1G2 Protocol for Anti-Collision

Q is a parameter that an interrogator uses to regulate the probability of tag response.

An interrogator commands tags in an inventory round to load a Q-bit random (or

pseudo-random) number into their slot counters. The tags in their arbitrate state decre-

ment their slot counter every time when they receive a QueryRep command from the

Page 202


interrogator. The tags reply when the value in their slot counter is zero. Q is an integer

from 0 to 15 [32].

In our case, because of the limitation of the radiation power and the size of the antenna

main lobe, the number of the tags in the effective antenna field is much less than 2000

and is more likely to be around 400. Within the range of available Q factors, it may be

assured that the probability of collisions is low, and all tags may be read quickly.

8.6.2 Method of Packaging and Stacking DVDs in Industry

The method of packing and stacking a large number of DVDs in industry is investi-

gated in this subsection in order to make this project more practical. In the investiga-

tion, it is found that a large number of DVDs are usually distributed in cartons, and

the cartons with DVDs are stacked on a pallet for storing or shipping. In this sub-

section, the dimension of the carton and pallet which are used widely in industry are

introduced respectively.

The carton commonly used for distributing DVDs in industry is shown in Figure 8.20.

The carton is a cardboard box with length Lc = 300mm, width Wc = 280mm and height

Hc = 205mm. The carton is just suitable to fit in 40 DVDs as shown in Figure 8.20(b).

The spine of each DVD case is perpendicular to the bottom of the carton and the 40

DVDs are stacked into two rows, 20 for each row.

Carton opening

Hc

WcLc

(a) Packaged carton

Hc

WcLc

(b) Opened carton

Figure 8.20. DVD carton and its dimensions.

Page 203


A pallet (sometimes called a skid) is a flat structure that supports goods in a stable

fashion while being lifted by a forklift, pallet jack, front loader or other jacking device.

While most pallets are wooden, pallets also can be made of plastic, metal and paper.

The pallet in this chapter only refers to a wooden one.

A picture of a real pallet is shown in Figure 8.21. As the figure shows, the pallet is

generally made up by several stringers and deck boards and is hollow inside. The

dimensions with which we are really concerned are those of the top surface on which

the goods are loaded.

Figure 8.21. A sample of a real pallet.

There are two things to constrain the size of the top surface of a pallet: 1) the pallet

shape should be a square or nearly square surface, since that helps a load to resist

tipping; 2) pallet users want pallets to pass easily through the gates of buildings, to

stack and fit in racks, and to be able to be moved by a forklift or pallet jack in an auto-

mated warehouse. Even though there are some consensuses on the dimensions of the

pallet, no universally accepted standards for pallet dimensions exist. Companies and

organisations apply and publish hundreds of different pallet dimensions all of over

the world [91]. Here, six pallet top surface dimensions specified by the International

Organization for Standardization (ISO) in ISO Standard 6780 [92] are introduced in

Table 8.5. Each of them is widely used in particular country or region.

The data in Table 8.5 complies with the discussion above that the top surface should be

square or nearly square. The length of the square is approximate 1m. In the following

Page 204


discussion, pallet refers to the pallet commonly used in Australia. As stated in Table 8.5

its size is 1165mm×1165mm.

Table 8.5. Pallet top surface dimensions standardised by ISO.

Dimensions (mm) Region or Country used in

1219×1016 North America

1000×1200 Europe, Aisa

1165×1165 Australia

1067×1067 North America, Europe, Asia

1100×1100 Asia

800×1200 Europe

8.6.3 Experiments

In this section, the testing schemes “2” and “3” are further tested by expanding the

number of tagged DVDs and adapting the realistic testing environment of stacking

cartons on a pallet. The testing processes are discussed in the following two sub-

subsections.

Because, historically, the testing scheme “3” was investigated in these further experi-

ments before the testing scheme “2”, it is reported first below.

(a) Testing scheme “3”

For the testing scheme “3”, a large number of DVDs are piled up in cartons resting on

their broad sides on an Australian pallet as shown in Figure 8.22.

The width of each carton is placed to be vertical on the pallet. The carton opening is

towards to the interrogator. To describe the DVD stack, using the interrogator as the

position reference, we use the term height for a vertical dimension of the stack, depth

for a front to back dimension of the stack, and width for a side to side dimension of

the stack. Hence, there are three cartons along the stack height, three cartons along

the stack width, five cartons along the stack depth. In total there are 45 cartons which

Page 205


Pallet

Carton opening

Interrogator

Lc

Hc

Wc

Figure 8.22. The DVD stack structure for testing scheme “3”.

contain 1800 DVDs. If a 3-deep DVD carton is successfully read, we will expect that

all the tagged DVDs in the stack could be read if we were prepared to move the inter-

rogator along the height and width dimension in the front face of the stack, and then

read from the back face by the same method of mobilising the reader antenna. The

reader antenna has to be placed just against the stack. This testing strategy has been

described in Subsection 8.5.1 and shown in Figure 8.14(b). One of the cartons at the

right and bottom corner of the stack is made transparent in Figure 8.22. Two DVDs on

the edge of this transparent carton are drawn to show the method of DVDs stacking in

the carton and tag labelling on the DVDs. Each of them is labelled exactly in the same

way as shown in Figure 8.13(c). The tags are presented in the color orange.

Because of the limitation of the number of DVDs and tags as described in Subsec-

tion 8.1.2, the experiment is not conducted on the stack shown in Figure 8.22. The

number of the tagged DVDs used in the experiments of this section is 320 which can

be distributed into 8 cartons. These cartons are piled up as a certain shape on the pallet

shown in Figure 8.23 in which there are 8 cartons 1 wide by 2 high by 4 deep.

As mentioned before, we intend to read all the DVDs in the front three deep cartons.

To achieve this, the interrogator antenna has to be deployed just against the front face

of the stack and 4 positions of the reader antenna’s centre in relation to the front face

Page 206


Interrogator

Hc

Wc

Lc

Figure 8.23. The real DVD stack for testing scheme “3”.

are marked by the blue squares in Figure 8.24(a). Apparently, the four positions of the

reader antenna are all in the centre of the stack width, which has the dimension Lc in

this case, and the distance between adjacent positions is half carton’s width Wc. At the

bottom position the centre of the antenna was a little above where we know the tag to

be. The testing results are shown in Table 8.6.

Hc

Wc

Lc

(a) 4 position test

Hc

Wc

Lc

(b) 12 position test

Figure 8.24. The reader antenna’s positions in relation to the stack in terms of the testing

scheme “3”.

The first row of Table 8.6 denotes the depth of the tags in the stack. Hc represents the

carton’s height (we do not seek the detection of the tags in the two cartons at depth

Page 207


4 shown in Figure 8.23). Similarly, the first column of Table 8.6 denotes the height of

the tags in the stack. We can see better the position of the tags in Figure 8.22. The

dimensions Wc of the carton’s width, Hc of the carton’s height, and Lc of the carton’s

length, have been marked in Figure 8.20 and are further marked in Figure 8.23.

Table 8.6. Misreading tag distribution in the stack shown in Figure 8.24(a). For these results

the reader antenna occupied four positions.PPPPPPPPPPPPPP

Height

Depth0×Hc 1×Hc 2×Hc

0.5×Wc 0 4 10

1×Wc 0 5 5

1.5×Wc 0 4 10

2×Wc 0 4 13

Each cell of the table (except those in the first row or column) represents a line segment

in the stack on which segment there are 20 tags. The number in each cell illustrates the

number of the misreading tags among the 20 tags present in this cell. Apparently, with

the increase of the depth, the number of the misreading tags is increased. The misread-

ing tags are usually found on the side edge of the stack. Those misreading behaviors

are believed to be caused by the limitation of the reader antenna’s radiation pattern,

i.e. the relatively large radiation power from the reader antenna only concentrates in

a certain area in the stack, which is here named as the effective reading area. The ef-

fective area becomes narrow with the increase of the depth in the stack because of the

attenuation along the depth dimension, and the power absorption and consumption of

the tags in front. That is the reason why the number of the misreading tags is positively

related to the depth in the stack and most of them are on the edge of the deep section

of the stack.

However, a small portion of the misreading tags are not on the side edge of the stack.

This results from the tags’ weak performance since those misreading DVDs can be

detected by replacing them with other well reading tags. These misreading behaviors

are related to the manufacturer’s production reliability and the tag design.

Page 208


According to Table 8.6, the total number of the misreading tags is the sum of the num-

bers in the table, which is 55. The number of the tags in the stack except the tags in the

4-deep cartons is 240. Hence, the readable ratio is 77.1%. One thing should be noted

that the data shown in Table 8.6 is not obtained by just one inventory round in each in-

terrogating position of the reader antenna, but by 4 inventory rounds. The tags which

can be detected in any round of the four are regarded as the detected tags2.

There are three methods to increase the readable ratio according to the previous anal-

ysis of the reasons for misreading behaviors.

Firstly, for the limitation of the reader antenna’s radiation pattern can be compensated

by placing the reader antenna against the stack front face in more positions. For ex-

ample, the experiment is re-conducted by placing the reader antenna at 12 positions

in front of the stack to read the tags in different parts of the stack. The 12 positions of

the reader antenna’s centre in relation to the stack front face is shown by the 12 blue

squares in Figure 8.24(b). The vertical distance between the adjacent positions is half

carton’s width Wc. The horizontal distance between the adjacent positions is half of

carton’s length. At the bottom positions the centre of the antenna was a little above

where we know the tag to be. The testing results are shown in Table 8.7 which inter-

pretation is similar to that of Table 8.6. The readable ratio of this experiment is 96.25%.

Apparently, the increase of the readable ratio is only at the expense of time. It has

been established by DVD position swapping that the misreading tags are caused by

the tag’s weak performance. If there had been no weak tags there would have been no

misreads.

Secondly, other methods under testing scheme “3” for enhancing the readable ratio is

changing the type of tag attached on DVD packages, or designing a new type of tags

to obtain longer and more reliable performance.

Thirdly, as mentioned in Subsection 8.5.1, the reader antenna should be as close as

possible to the stack to reach the furthest tags inside. However, we notice that if the

reader antenna is too close to the stack, its input impedance is varied which results in

2This interpretation is also suitable for Tables 8.7, 8.8, 8.9.

Page 209


somewhat reflection loss, hence the distance between the reader antenna and the stack

should be optimised. The optimisation is discussed in detail in Section 8.7.

Table 8.7. Misreading tag distribution in the stack shown in Figure 8.24(b). For these

results the reader antenna occupied twelve positions.PPPPPPPPPPPPPP

Height

Depth0×Hc 1×Hc 2×Hc

0.5×Wc 0 0 1

1×Wc 0 1 1

1.5×Wc 0 0 2

2×Wc 0 0 4

(b) Testing scheme “2”

For the testing scheme “2”, a large number of DVDs are stacked in cartons resting on

their bases on an Australian pallet as shown in Figure 8.25.

Pallet

Carton opening

Interrogator

Lc

Hc

Wc

Figure 8.25. The DVD stack structure for testing scheme “2”.

Two DVDs on the edge of a transparent carton at the left edge of the stack are drawn to

show the method of DVD stacking in the carton and tag labelling on the DVDs. Each

Page 210


of them is labelled exactly in the same way as shown in Figure 8.13(b). The tags are

presented in the color orange. The height dimension of each carton is vertical on the

pallet. The face defined by the carton height and length is towards to the interrogator.

The real cartons fully filled with DVDs are stacked as the shape on the pallet shown in

Figure 8.26 in which there are 8 cartons (320 DVDs), 2 wide by 2 high by 2 deep. The

interrogator antenna has to be deployed just against the front face of the stack.

Interrogator

Hc

Wc Lc

Figure 8.26. The real DVD stack for testing scheme “2”.

The experiments for the testing scheme “2” were conducted similarly to those for the

testing scheme “3”. First, the experiment was conducted by placing the reader antenna

at 4 positions in front of the stack as shown by the blue squares in Figure 8.27(a). The

vertical distance between the adjacent positions is one carton’s height and the hori-

zontal distance between the adjacent positions is one carton’s length. The results of

this experiment have been given in Table 8.8. Then, the experiment is re-conducted by

increasing the testing position to be 10 as shown in Figure 8.27(b). The vertical dis-

tance between the adjacent positions remains one carton’s height and the horizontal

distance between the adjacent positions becomes half carton’s length. The results of

the this experiment have been given in Table 8.9.

The interpretation previously given for Table 8.6 can help to understand Tables 8.8

and 8.9. The first row of Tables 8.8 and 8.9 denotes the depth of the tags in the stack.

Page 211


Hc

Wc Lc

(a) 4 position test

Hc

Wc Lc

(b) 10 position test

Figure 8.27. The reader antenna’s positions in relation to the stack in terms of the testing

scheme “2”.

Table 8.8. Misreading tag distribution in the stack shown in Figure 8.27(a). For these results

the reader antenna occupied four positions.PPPPPPPPPPPPPP

Height

Depth0×Wc 0.5×Wc 1×Wc 1.5×Wc

1×Hc 3 6 13 18

2×Hc 2 7 7 19

Table 8.9. Misreading tag distribution in the stack shown in Figure 8.27(b). For these

results the reader antenna occupied ten positions.PPPPPPPPPPPPPP

Height

Depth0×Wc 0.5×Wc 1×Wc 1.5×Wc

1×Hc 0 0 0 1

2×Hc 0 0 0 3

Wc represents the carton’s width. Similarly, the first column of these two Tables de-

notes the height of the tags in the stack. We can see better the position of the tags in

Figure 8.25. Each cell of the table (except those in the first row or column) represents a

line segment in the stack on which segment there are 40 tags. The number in each cell

illustrates the number of the misreading tags among the 40 tags present in this cell.

Page 212


Hence, the readable ratios are 76.56% and 98.75% according to Tables 8.8 and 8.9 re-

spectively. It has been established by DVD position swapping that the misreading tags

in Table 8.9 are caused by the tag’s weak performance. If there had been no weak tags

there would have been no misreads.

However, as addressed before, the reader will be removed to the back side of the stack

to detect the tags in the third and forth deep cartons, as shown in Figure 8.28. Because

of the un-symmetrical structure of the tag labelling on DVDs, once the reader is moved

to the back side of the stack, the distance between the tags and the reader antenna

ranges from 0.5×Wc to 2×Wc. This is different from the situation when the reader

intends to detect the tags from the settled against the front side of the stack, i.e the

distance between the tags and the reader ranges from 0×Wc to 1.5×Wc. Hence, it is

worth investigating the readable ratio when the reader antenna is moved against the

back side of the stack.

Pallet

Carton opening

Interrogator

Figure 8.28. The DVD stack structure for testing scheme 2. The reader scans the back side

of the stack.

The experiment was still conducted based on the DVD stack shown in Figure 8.26, but

the stack was rotated by 180 degrees to contemplate the situation here. In addition, the

experiment was only conducted by placing the reader antenna at 10 positions which

Page 213


are similar to those shown in Figure 8.27(b). The results of the this experiment have

been given in Table 8.10.

Table 8.10. Misreading tag distribution when those tags are read from the back of the stack.

For these results the reader antenna occupied ten positions.PPPPPPPPPPPPPP

Height

Depth0.5×Wc 1×Wc 1.5×Wc 2×Wc

1×Hc 0 0 1 7

2×Hc 0 0 6 4

The readable ratio according to Table 8.10 is 94.37%. Again, it has been established by

DVD position swapping that the misreading tags are caused by the tag’s weak perfor-

mance. If there had been no weak tags there would have been no misreads. However,

the fact that the number of misreading tags in Table 8.10 is more than that in Table 8.9

indicates that when the reader antenna is moved to the back side of the stack, it makes

it difficult to detect all of the tags. This fact is easy to understand since the distance be-

tween the reader antenna and the tags is increased when the reader antenna is moved

to the back side.

The method to solve this problem is to stack the whole tagged DVDs symmetrically

about the centre plane of the stack depth, as shown in Figure 8.29.

Since the two deep cartons and the tags in these cartons in the front part of the stack

and those at the back part of the stack are symmetrical about the centre plane of the

stack depth, it is assumed that the reader will achieve the same high readable ratio

98.75% (or 100% without the weak tags) no matter whether the reader is settled in

the front side of the stack or in the other side. This assumption is sensible because

the interrogating waves are attenuated and absorbed dramatically during the first two

deep cartons, through which the wave propagates. The DVDs and the tags on them

in the third deep and forth deep cartons hardly reflect waves to influence the readable

behavior of the tags in front.

Page 214


Carton opening

Pallet

Figure 8.29. Illustrating reflection symmetry of tag positions about a vertical mid-plane.

8.7 The Optimisation of the Distance Between the Reader

Antenna and the DVD stack

As mentioned before, if the reader antenna is too close to the DVD stack, its input

impedance is changed from the originally designed 50Ω which causes wave reflection

and thus some power intended to be transmitted by the reader antenna is lost through

reflection. The power loss will somewhat degrade the readable ratio. However, if the

reader antenna is moved far enough from the DVD for the input impedance to remain

at 50Ω, there will be also some power loss in this extra propagation distance between

the reader antenna and the DVD stack. In this section, a tradeoff is made in terms of

this distance in order to improve the readable ratio.

The variation of the input impedance along with the variation of the distance de be-

tween the reader antenna and the DVD stack as measured by the network analyzer is

shown in Figure 8.30 in the form of Smith Chart.

Figure 8.30 shows the input impedance of the reader antenna from 920MHz to 926MHz

which is the UHF RFID frequency band regulated in Australia. The curves in differ-

ent colors denote the results when the reader antenna is placed at different distances

Page 215

8.7 The Optimisation of the Distance Between the Reader Antenna and the DVDstack

Stop 926.000 MHzStart 920.000 MHz Stop 926.000 MHzStart 920.000 MHz

-j250

-j100

-j50

-j25

-j10

j10

j25

j50

j100

j250

10 25 50 100 2500

de=0mm

de=35mm

de=55mm

de=15mm

Free space

Figure 8.30. The variation of the reader antenna input impedance in the form of Smith

Chart along with the variation of the distance between the reader antenna and

the DVD stack de measured by the network analyzer.

de from the stack. In addition, the curve in red represents the input impedance of the

reader antenna, when it is put further away from the DVD stack, which can be con-

sidered as placing the reader antenna in frees space. As we know, the Smith Chart is

formed by the concentric circles of which the radius represents the magnitude of the

reflection coefficient. The points in the Smith Chart on the same circle have the identit-

cal magnitude of reflection coefficient. Therefore, the smaller the radius of the circle is

the better the impedance is matched to the impedance marked in the centre, where it

is 50Ω. According to Figure 8.30, with the decrease of the distance de, the reflection be-

tween the reader generator and the reader antenna is increased. When de=0mm, which

means the reader antenna is just against the DVD stack, the magnitude of the reflection

coefficient (the radius of the blue circle) is about 0.33. Hence, the power loss caused

by the impedance mismatch is the square of this magnitude of the reflection coefficient

and is about 10%, because the reflection coefficient represents the wave reflection in

terms of voltage not power.

Page 216


To avoid this power loss, the reader antenna should be moved further away from the

DVD stack, but by doing this more power will be lost in propagation over this extra dis-

tance de. Hence, a tradeoff over de has to be made. According to Figure 8.30, when the

reader antenna is placed from the stack at a distance of 35mm, its impedance is similar

to that when it is placed in free space. We believe the power loss due to propagation

over an extra 35mm distance might possibly be negligible but the power compensation

because of the lesser reflection at the antenna input is somewhat significant, hence the

optimum distance would be about 35mm, because no significant further reduction in

reflection is available at greater distances.

The experiments on the testing scheme “2” were re-conducted by setting the distance

de to be 35mm. The results show that only 2 tags within the 320 tags are missed, and

as before these are attributed to weak tag performance, and thus the readable ratio is

increased from the previous 98.75% to 99.375%.

Although, we did not re-conduct the experiments on the testing scheme “3” by opti-

mising the distance de, it is believed that if we did, there would be a similar increase in

readable ratio compared with the previously obtained readable ratio 96.25% by placing

the reader antenna just against the DVD stack.

8.8 Conclusion

This section provides the conclusions for this chapter. It recommends the optimum

configuration for DVD stacking, and indicates what level of success can be expected in

and attempt to read all tags. It also suggests avenues for improvement.

8.8.1 Stacking Policies

In Subsection 8.6.2 the carton commonly used for distributing DVDs in industry was

shown in Figure 8.20. The carton is a cardboard box with length Lc = 300mm, width

Wc = 280mm and height Hc = 205mm. The carton is just suitable to fit in 40 DVDs as

shown in Figure 8.20(b). The spine of each DVD is perpendicular to the bottom of the

carton and the 40 DVDs are stacked into two rows, 20 for each row.

Page 217

8.8 Conclusion

With such cartons it is feasible to stack them (i) on their sides as shown in Figure 8.22

or (ii) on their bases as shown in Figure 8.25. The labelling methods of the tags on

the DVDs corresponding to the stacking policies are also shown in Figure 8.22 and

Figure 8.25 respectively. The results of both stacking policies have been investigated in

Subsection 8.6.3.

8.8.2 Results for Side and Base Stacking

In Subsection 8.6.3, 320 DVDs, which can be distributed in 8 cartons described above,

are used to imitate a significant portion of the 2000 DVD stack. Experiments for ex-

amining the stacking policies were conducted based on the 8 carton DVDs above a

wooden pallet. The results are discussed.

The readable ratio is 77.1 %, when such cartons are stacked on their sides and the reader

antenna is just moved vertically to occupy 4 positions in front of the stack as shown in

Figure 8.24(a), but 96.25% when the reader antenna is moved both vertically and from

side to side to occupy 12 positions in front of the stack as shown in Figure 8.24(b).

An investigation of the readable ratio when such cartons are stacked on their bases

gave a readable ratio of 76.56% when the reader antenna is moved vertically to occupy

two positions in front of the stack and for each such position is moved horizontally to

occupy two positions in front of the stack, hence there are 4 read positions to complete

the detection as shown in Figure 8.27(a), but 98.75% when the reader antenna is moved

both vertically and from side to side in front of the stack to occupy a total of 10 positions

as shown in Figure 8.27(b). As discussed in Subsection 8.6.3, it is not assured to obtain a

readable ratio as high as 98.75%, by just adopting the base stacking policy. To achieve

that goal, the tags have to be reflection symmetrical about the middle vertical plane

together with the base stacking policy as shown in Figure 8.29.

In Section 8.7, it is found that if the reader antenna is too close to the stack, its input

impedance will be varied from the originally designed value which causes some power

loss. The distance between the reader antenna and the DVD stack is so investigated

Page 218


and optimised. After the optimisation the readable ratio is increased from 98.75% to

99.375% for the base stacking policy.

For both of the side and base stacking, with the elimination of weak tags, a readable

ratio of 100% is expected.

We recommend people to use the latter method to solve the problem of reading mas-

sive numbers of DVD on a pallet for two reasons expanded as follows:

1. The base stacking policy corresponding with the tag symmetrical placement can

obtain a higher readable ratio than the side stacking policy without the elimina-

tion of the weak tags.

2. Base stacking is a natural way of stacking those cartons since their openings then

face upwards.

8.8.3 Further Work

Although the results for base stacking suggest that, with an appropriate arrangement

of cases including ensuring that the tags have reflection symmetry about the middle

vertical plane as shown in Figure 8.29, and without weak tags, all tags will be read, it is

still possible that these results could be further assured by not only eliminating weak

tags, but also by designing tags that are optimised for folding across the spine of the

DVD case.

Page 219

Chapter 9

Conclusions and FutureWork

THIS chapter concludes the thesis by reviewing the work done,

re-summarising the original contributions, and recommending

future work that could be undertaken by others.

Page 221

9.1 Review of and Conclusions from the Work in This Thesis

9.1 Review of and Conclusions from the Work in This

Thesis

The research work in this thesis focuses on passive Ultra High Frequency (UHF) Ra-

dio Frequency Identification (RFID) systems. Identification technology using radio

frequency waves has been in existence for a few decades, but it started developing

rapidly once the Electronic Product Code (EPC) concept was defined. The EPC con-

cept can provide every single item in a supply chain a unique identification at a low

cost, and can serve as a pointer to a large amount of data for each item.

Now, RFID systems are used widely in industry, such as in supply chains, airport lug-

gage management and animal tracking. However, there are still many categories of

items which are either very hard to detect by UHF RFID systems or omit important

functions, for example, physically small items, metallic items or items needing a secu-

rity function. This thesis aims to provide feasible solutions for these hard-to-tag items

or items with insufficient functionality in passive UHF RFID systems.

Chapter 1 and Chapter 2 provide the introduction to the thesis and introduction to

passive UHF RFID systems. In detail, Chapter 1 introduces the research area, motiva-

tions and contributions of this thesis. The thesis structure is discussed in this chapter

as well. In Chapter 2, more details of RFID systems’ background are described. These

include RFID history, classification, regulations and standards.

Chapter 3 evaluates the operating range of general UHF RFID systems thoroughly.

Some parameters which determine the performance (operating range) of RFID systems

are discussed. Those parameters include chip design, reader analysis, tag antenna de-

sign and working environment analysis. In addition, a novel method for calculating

the operating range of a UHF RFID system is proposed by making use of a scatter-

ing matrix which can be derived from simulation or experiments. The coincidence

between the experimental reading ranges of a self-made tag and the results calculated

by this method is found to validate this method. At the conclusion of this chapter, the

parameters which play a key role in extending the operating range in RFID systems

are identified. They are: (i) the selection of the parameter θ, the magnitude squared of

Page 222

Chapter 9 Conclusions and Future Work

which establishes the fraction of the available tag antenna power that is not delivered

to the tag chip, since it governs how much power would be delivered to power the

chip and how much will be backscattered to be sensed by reader; (ii) the rectifier de-

sign since the enhancement of the rectifier efficiency can lower the threshold power of

the chip; and (iii) the system deployment environment and in particular the attached

item especially when it is a metallic item or a liquid item.

Chapter 4 and Chapter 5 put emphasis on the analysis and applications of meander line

dipole antennas which provide one of the approaches to minimising the tag antenna

size. In detail, Chapter 4 gives the analysis of the meander line dipole antenna’s (MDA)

properties including resonant frequency, radiation pattern and efficiency, and the rea-

son why by loading meanders on a dipole can reduce its size. An analytic formula is

proposed to calculate the resonant frequency of a tag antenna based on an MDA above

a dielectric substrate. The simulations and experiments described in this chapter val-

idates this analytic formula. Chapter 5 provides an original security tag design based

on MDA pattern for any container (large or small) that has either a) two sides that join

together where a sealing chamber is attached or b) a finger that slots into a chamber. By

theoretical analysis, simulation and experimental demonstration, the security function

of this tag is shown to have been achieved by a combination of mechanical design and

tag antenna design. Appendix A contains some original experimental results on this

security tag and serves as supporting material for Chapter 5.

Chapter 6 and Chapter 7 aim to solve problems in detecting metallic items. Firstly,

Chapter 6 discusses the reason why metallic items are difficult to detect in UHF RFID

systems. The failure in detecting some commercial tags in close proximity to a metallic

plate is established by experiments. Existing solutions to this problem are summarised.

The conclusion is also drawn that those existing solutions either suffer in the UHF RFID

band from being bulky or high in manufacturing cost. Chapter 7 introduces a solution

from the literature [15] with significant advantages of low profile and simple structure

to detect metallic items by RFID systems. The solution is named as the slitted decou-

pler in this thesis. It is concluded that the slitted decoupler is an effective solution to the

metallic item detection problem. However, the inventors of this slitted decoupler [15]

did not give an explanation of the working scheme of the device. This thesis provides

Page 223

9.2 Recommendations on Future Work

the explanation by using theoretical analysis and extensive simulation results. From

that analysis and those simulation results, we have concluded some design principles

proposed by the inventors [15] need refinement. Hence, some revised design princi-

ples are proposed to minimise the size of the decoupler, and meanwhile improve its

performance. These principles are examined by both simulation and experiments.

Chapter 8 addresses the problem of detecting a large number, approaching 2000, of

packaged DVDs densely stacked. Since each DVD disc actually contains a very thin

metallic layer to reflect the laser beam, the detection of DVDs could be included in the

scope discussed in the previous paragraph which is the detection of metallic items. But

we choose to discuss this problem separately, because the difficulties in achieving the

detection are casued not only by the metallic component in the DVD but also by the

combination of that aspect and the large number. In order to achieve the goal of de-

tection of 2000 DVDs in a stack, the dimensional and electrical parameters of a general

packaged DVD are investigated. The labelling methods of a tag on this DVD and the

polarisation types of the interrogating wave are selected. The testing strategy which

denotes the relative positions of the reader antenna in front of the DVD stack and the

stacking policies which represent the stacking methods of the DVDs on a wooden pal-

let (usually used in industries in carrying those DVDs) are studied. The experiments

illustrate that with careful consideration of the issues mentioned above and elimina-

tion of weak tags, perfect reading can be achieved.

Chapter 9 is the final chapter of this thesis. It reviews and concludes the work in

this thesis, re-summarises the original contributions and recommends further research

topics that could be undertaken by others.


This thesis provides some feasible solutions for several categories of hard-to-tag items

in UHF RFID systems. However, due to either the time constraint imposed by the set

duration of this research or the current limitation of resources, or both, some ideas for

improving those solutions in performance or cost-efficiency were not able to be carried

out. They are listed as follows.

Page 224


1. Security tag design

• Low profile and low cost design

The security tag introduced in Chapter 5 is designed and fabricated on a

1.6mm FR4 board, which is too thick to be broken as is needed in the dis-

ablement process and it is also too thick to be placed between the seal body

and cover. Future work can take substrates with different dielectric constant

and small thickness into account. Achieving a low cost is also a significant

consideration for this tag to be commercially used.

• Tag placement on containers

Since the security tag is intentionally designed for protecting containers, for

example shipping containers, the effects brought by the materials (usually

metal) composing of containers and the shape of the container should be

considered. In this thesis, those effects are ignored because the distance be-

tween the security tag and the container is assumed to be sufficiently large

that the container does not affect, or could positively affect, the performance

of the security tag antenna. To avoid making this assumption, further re-

search on this design could locate the optimum placement of the security

tag on a detected container.

2. Slitted decoupler design

• Antenna design for slitted decoupler

As introduced in Chapter 7, it is predicted that the strong interaction be-

tween the slitted decoupler and commercial tag antennas near the decoupler

will shift the commercial tag antennas’ input impedance far away from the

intentionally designed impedance in free space. The shift in input impedance

of the tag antenna leads to a relatively short reading range compared to the

reading range of the same commercial tags are put in free space. In order

to get the long reading range and low profile of the whole structure (tag

above decoupler), a new type of tag antennas is needed to make its input

Page 225


impedance conjugate match to the chip impedance in close proximity to the

slitted decoupler.

• Chip design for slitted decoupler

By the analysis in Chapter 7, it is known that the induced voltage of a half

wavelength dipole on the decoupler is larger than that of the same dipole

in free space, when both dipoles are illuminated by the same uniform plane

wave. It is expected that a tag antenna on the slitted decoupler will induce

more voltage than that in free space, if the tag antenna is designed based on

dipole antenna pattern. Once the tag antenna collects the induced voltage

no matter whether it is on the decoupler or in free space, this AC voltage will

be rectified and multiplied into DC voltage by a voltage multiplier, so that

the interior circuits of the chip can be exited. The voltage multiplier contains

several stages to pump up the induced AC voltage to the needed operating

DC voltage. Each pumping stage is accomplished by a rectifier with certain

losses. If the tag antenna on slitted decoupler can induce more voltage than

it does in free space, then a voltage multiplier in the chip does not need

as many pumping stages as the multiplier needs in free space. Therefore,

if the voltage multiplier in chip is redesigned with fewer stages of rectifier

and the tag antenna on the slitted decoupler designed for the condition that

the antenna’s input impedance conjugate matches the chip impedance, the

efficiency of the voltage multiplier is expected to be be increased, and to

result in longer tag reading range.

• Genetic Algorithm application in designing slitted decoupler

Genetic Algorithms (GA) are now used widely in optimising designs which

need to consider many cross-impacting parameters. As noted in Chapter 7,

the performance of slitted decoupler is decided by several cross-impacting

parameters, such as, slit width, top patch length, width, dielectric layer

thickness and dielectric constant. Therefore, there is a potential to make

use of GA to optimise the slitted decoupler.

3. Cost analysis and reduction

Page 226


While the cost is always considered throughout the solutions presented in this

thesis, it is still worth estimating the actual cost of these solutions when they are

commercially adopted. One significant portion of the cost is the price of manu-

facturing each tag. A common manufacturing result is that the larger the number

of the items manufactured, the lower the cost. Therefore, the cost analysis begins

with considering how tag manufacturing cost varies with respect to the num-

ber of items (tags or decouplers) manufactured. Further, the cost of each item is

also critically dependent on the technology adopted in manufacturing process.

Adopting new technology can dramatically affect the interaction between the

market, the price, the protocol and the manufacturing process. We believe de-

velopment of printed electronics especially printed silicon electronics will reduce

the cost significantly. Hence, tags based on printed silicon electronics and new

protocols for these tags could be a potential research interest in the future.

4. Tag antenna design in the detection of a large number of DVDs

In Chapter 8, a very high readable ratio (over 99%) of DVDs in a stack is obtained

by labelling DVDs with one type of existing commercial tag after carefully op-

timising the labelling method, the type of interrogating waves, the testing strat-

egy and the stacking policy. Perfect reading can be achieved by eliminating the

weak tags. However, the unread tags are only relatively weak to the weakened

interrogating waves. They are randomly distributed among the supplied tags

and we could not tell them from the others before implementing them. In order

to achieve perfect reading without considering the weak tags, the tag antennas

could be redesigned so that they are optimised for folding across the spine of the

DVD case and thus have high performance in their deployed context.

5. RFID reader antennas in metallic environment

Also in Chapter 8, it is shown that if the reader antenna is too close to the DVD

stack, its input impedance will be varied significantly which result implies that

the effects of the deployed environment to the reader antenna’s performance can-

not be negligible. Moreover, a reader antenna may be required to be placed in

Page 227

9.3 Summary of Original Contributions to Knowledge

tight corners mostly surrounded by metallic surfaces, or placed on a metal fork-

lift. The existence of those metals may disturb the performance of the reader

antennas in terms of its input impedance, radiation pattern and radiation effi-

ciency. Hence, the disturbance is worth studying in order to improve the reader

antennas’ performance.

9.3 Summary of Original Contributions to Knowledge

The contribution to knowledge made in this thesis have previously been described in

Section 1.3. The contribution to the knowledge are re-summarised as follows.

1. Method for evaluating the operating range of a UHF RFID system

The factors e.g. chip design, reader design, tag antenna design, and deployed

environment analysis, which could affect the operating range of a UHF RFID

system are summarised. Key factors in deciding the reading range are identified.

The limitations of analysing the operating range by the Friis equation which is

commonly adopted are discussed. In order to overcome these limitations, a novel

method for evaluating the operating range of a UHF RFID system by making use

of a scattering matrix is proposed. By using this method, the operating range of

the UHF RFID system deployed in complex environments can be predicted. In

addition, the scattering matrix can be easily obtained by simulation software or

by experiments.

2. Relative effective permittivity of meander line dipole antenna on a dielectric substrate

An analytic method for calculating the relative effective permittivity of meander

line dipole antenna (MDA) on a dielectric substrate is proposed. As far as we

know, there is no published work on that topic. The method is examined ac-

cording to HFSS by varying the values of dielectric constant and thickness of the

substrate. The simulation results show that this original method can be used to

calculate the relative effective permittivity accurately and efficiently, when the

dielectric constant is in the range from 1 to 4 as is the case for materials which are

commonly used in manufacturing and packaging of RFID tags.

Page 228


3. Equation for designing tag antennas using meander line dipole antenna on a dielectric

substrate pattern

Equation (4.12) is proposed firstly for designing tag antennas using meander line

dipole antenna pattern. Published methods or equations [14] cannot evaluate

the effects of the underneath dielectric substrate and do not consider the special

impedance matching condition in RFID tag antenna design. Equation (4.12) deals

with those mentioned deficiencies. The simulation and experiments show that

Equation (4.12) can be used to design tag antennas using an MDA on a dielectric

substrate successfully. The design cycle using the combination of this equation

and a simulation software has been greatly shortened compared with the design

cycle by just using the commercial simulation software itself.

4. Security tag design

A novel tag with security function is designed as an electronic seal for protect-

ing any containers (large or small) that has either a) two sides that join together

where a sealing chamber is attached or b) a finger that slots into a chamber. The

tag antenna design is based on a meander line dipole antenna pattern. With the

combination of the mechanical design and tag antenna design, the tag can pro-

tect the containers from compromise successfully. The security function has been

verified by experiments.

5. Improvements in designing slitted decoupler

The slitted decoupler is proposed by the patent [15], as a solution for metallic

item detection by UHF RFID systems. Some of the given design principles of the

slitted decoupler by [15] do not accord with the theoretical analysis and simula-

tions described in this thesis. Hence, some improved design principles for not

only reducing the size of the decoupler but also improving its performance are

introduced in this thesis. The improved principles are confirmed by simulation

and experiment.

6. Detection of a large number of DVDs

Page 229

9.4 Conclusion

A solution for detecting a large number of packaged DVDs densely placed in a

stack is proposed. The number is in the region of 2000. To the best of our knowl-

edge, there is no publication describing similar solutions not only for detecting

DVDs but also for detecting other commodities on such a scale. Experiments on a

great portion of 2000 DVDs (320 DVDs) have been conducted after considereing

the needs of industry in terms of packaging and carrying. The results show that

by adopting the solution, a very high readable ratio (over 99%) can be achieved.

Perfect reading is also possible by eliminating the weak tags.

9.4 Conclusion

This chapter summarises the research carried out in the duration of the Ph.D study.

According to the discussion in Section 9.3, the research done in this thesis contributes

to knowledge in UHF RFID systems into two aspects. (i) The thesis provides general

methods for (a) designing a type of tag antenna (the meander line dipole antenna)

which is one of approaches to minimising a tag antenna’s size and (b) evaluating the

operating range of a UHF RFID system deployed in complex environments. (ii) The

thesis gives some feasible solutions for detecting various types of hard-to-tag items,

i.e. containers with security needs, fairly small metallic items and massive numbers of

packaged DVDs. The first aspect of the contributions could be used by other involved

researchers in their own studies and applications. The second aspect drives industry

further toward full deployment of RFID down to item level tagging. The work in this

thesis and the recommendations on future work in Section 9.2 will create more research

possibilities for improving UHF RFID systems.

Page 230

Appendix A

Tests of the Tags inChapter 5

THIS appendix contains the details of tests of the two security tags

designed in Chapter 5. One tag is the semi-finished tag while the

other is made according to the final design.

Page 231

A.1 Test Scheme

A.1 Test Scheme

Both the semi finished security tag and the final design of the security tag are tested

by the same test scheme. The details of the test scheme, including the test equipment,

process, regulations and standards are introduced here.

The RFID reader (Model ID ISC.LRU2000-FCC) and circularly polarised reader an-

tenna (Model ID ISC.ANT.U250/250-FCC), which gain is about 6dBi, both by FEIG

Electronics are employed here to detect the two tags designed in Chapter 5. A shielding

tunnel shown in Figure A.1 is used to isolate the tested tag from the outside environ-

ment. When a tag is put in this tunnel, we make the hopeful assumption that the tag is

in free space. The reader antenna faces into the tunnel, but from some distance outside

it. The experiment is operated under the Australian UHF RFID standards and regula-

tions, which have been introduced in Section 2.4 and are repeated here. The frequency

spectrum is assigned from 920MHz to 926MHz. The maximum transmitted power is

4W EIRP. In addition, the frequency hopping spread spectrum (FHSS) is chosen to be

the frequency channel selection mode. The EPC Class1 Generation2 standard is used

here, since the chip Higgs-2, installed on the tags, adopts the EPC Class1 Generation2.

Figure A.1. A shielding tunnel. The size of this tunnel inside is 1826mm×915mm×690mm.

Page 232

Appendix A Tests of the Tags in Chapter 5

A.2 Test Result on the Semi-finished Security Tag

In this section, the tested tag is the semi finished security tag, which is shown in Fig-

ure A.2(a). The reading range of the semi finished tag is given as 3695mm.

(a) Semi finished tag (b) Final designed security tag

Figure A.2. Two tested tags.

A.3 Test Result on the Final Design of the Security Tag

In this section, the final designed of the security tag shown in Figure A.2(b) is tested.

First, the incomplete tag which means the tag antenna broken by the absence of the

two terminals is tested, and found not to be readable no matter how close it was to the

reader antenna. Secondly, when the two terminals are attached on the antenna to fill

the gaps, the reading range is then found to be 580mm.

Page 233

Appendix B

Open Circuit Voltage of AHalf Wavelength Dipole

IN this appendix, a method for calculating the induced voltage of

a half wavelength dipole illuminated by a uniform plane wave

above an infinite ground plane is introduced as a complementary

material for Chapter 7.

Page 235

We will here make use of notation for terminal voltage and current as defined in Sub-

section 3.2.3 and confirmed in Figure B.1 below with a newly defined set of x, y and z

axes.

First, an incident uniform plane wave is defined. The electric field of this wave is

pointing along y axis and the field is propagated along the −z axis. The peak value

phasor Eiy repenting the electric field of this incident uniform plane wave is expressed

in (B.1):

Eiy = E0ejkz (B.1)

where E0 is the peak value of the phasor of the incident electric fields in the plane

z=0 and k = 2πλ is the propagation constant in free space. An infinite ground plane is

sitting on the xy plane (z = 0). According to the metallic boundary condition, there is

no tangential electric fields on the surface of the ground plane. Hence, the peak value

phasor of the reflected electric fields Ery is obtained as follows:

Ery = −E0e−jkz (B.2)

The total field Ety above the ground plane is then the sum of the incident wave and

reflected wave, as shown in (B.3)

Ety = Ei

y + Ery = 2jE0 sin(

2πzλ

) (B.3)

The magnitude of the r.m.s phasor of the total electric fields is derived by (B.4).

|Etyr.m.s

| = 2E0√2· | sin(

2πzλ

)| (B.4)

A half wavelength dipole in the corresponding rectangular coordinate system is shown

in Figure B.1.

The radiated electric fields of this half wavelength dipole in a far field zone is expressed

in spherical coordinates on page 182 of [37], and shown in (B.5).

Eθ ' −jηIine−jkr

2πr[cos(π

2 cos θ)sin θ

] (B.5)

where Iin is the input current at the open terminal of the dipole, η represents the char-

acteristic impedance of free space, and r denotes the distance between the observing

Page 236

Appendix B Open Circuit Voltage of A Half Wavelength Dipole

y

x

z P

qDipole

Figure B.1. A half wavelength dipole in the rectangular coordinate system.

point in the far field zone to the dipole. θ has been shown in Figure B.1. We note (B.5)

has a minuse sign not appearing in the corresponding equation in Balanis [37] and be-

lieve it comes about from our careful definition of the reference directions of terminal

voltage, current and the equivalent effective length.

According to the method introduced in Subsection 3.2.3, if the electric fields in the far

field zone, radiated by an antenna, is known, the effective length of this antenna can

be obtained by (B.6).

Ea = −jηkIin

4πrlee−jkr (B.6)

where Ea is the electric fields vector radiated by an arbitrary antenna in the far field

zone, le is the effective length vector which has a direction the same as that of electric

field. Comparing (B.5) and (B.6), the effective length vector is obtained and expressed

in B.7.

le = aθλ

π[cos(π

2 cos θ)sin θ

] (B.7)

The open circuit voltage or so called induced voltage of the half wavelength dipole can

be derived by (B.8).

Vin = Ei · le (B.8)

where Ei is the incident plane wave. The incident plane wave in this case is the total

electric field Ety in (B.3) which is the combination of incident uniform plane wave Ei

y

in (B.1) and the reflected uniform plane wave Ery in (B.2). The total electric field Et

y are

normally incident on the dipole, hence, θ = π2 . Then, the effective length becomes λ

π

according to (B.7). The r.m.s magnitude of the induced voltage of the half wavelength

Page 237

dipole above the infinite ground plane, incident by a uniform plane wave Eiy, can be

obtained by (B.9).

|Vinr.m.s| = λ

π|Et

yr.m.s| = 2E0λ√2π

· | sin(2πz

λ)| (B.9)

where |Etyr.m.s| is shown in (B.4).

Supposing that the peak value of the phasor of the incident electric field E0 is√

2V/m,

the resonant frequency of this dipole is 923MHz, the r.m.s magnitude of the induced

voltage of the half wavelength dipole |Vinr.m.s| as a function of z can be derived and it

is shown in Figure B.2, when the dipole is above the infinite ground plane and illumi-

nated by a uniform plane wave.

Similarly, the r.m.s magnitude of induced voltage of the half wavelength dipole, nor-

mally illuminated by a uniform plane wave Eiy (expressed by (B.1)) in free space can be

obtained by (B.10).

|V f reespacein r.m.s| =

λ

π|Ei

yr.m.s| = E0λ√

2π(B.10)

When E0 =√

2V/m and λ = 325mm (representing the resonant frequency 923MHz),

|V f reespacein r.m.s| is 0.1035V.

0 0.2 0.4 0.6 0.8 10

0.05

0.1

0.15

0.2

0.25

z/l

Induce

d v

olt

age

(V)

Figure B.2. |Vinr.m.s| as a function of the ratio z/λ at 923MHz.

Page 238

Appendix C

Original Testing DataCorresponding to the Work

in Section 8.4

THIS appendix contains the original testing data which are read-

ing range of the tag attached on various positions on a DVD case

with a disc inside.

Page 239

The term “ND” in the following tables presents the tag cannot be detected at all.

Table C.1. Original testing data corresponding to Table 8.1, unit: mm.

(a)↑ (a)→ (b)↑ (b)→Opening A 3580 60 770 ND

Spine 720 280 50 50


(a)↑ (a)→ (b)↑ (b)→Opening A 4530 110 680 420

Spine 90 460 2060 20


(a)↑ (a)→ (b)↑ (b)→Opening A ND 1420 60 1280

Spine 150 1150 ND 1180

Page 240

Appendix D

Evaluation of Reflections inthe Aperture Surrounded

by Absorbing Foam Used inChapter 8

THIS appendix contains additional material related to Chapter 8.

Since an aperture surrounded by absorbing foam is used in the

experiments described in Chapter 8, it is useful to investigate the

reflection inside the aperture caused by the foam. Reflections are unde-

sirable as they can cause field nulls and failure to read tags in particular

positions.

Page 241

D.1 Introduction

D.1 Introduction

This appendix aims to evaluate the reflectivity of the absorbing foam which is used in

the experiments of Chapter 8. As mentioned in that chapter, the foam is manufactured

by the Emerson & Cumming Company for use in the frequency range from 600MHz to

40GHz. The absorbing foam can achieve a minimum of 22dB return loss around 1GHz

if the incident wave is normal to the foam. The reflectivity performance can degrade

for off-normal incidence and at different rates for different polarisations [89].

In the experiment of Section 8.5, the absorbing foam is placed so as to build the small

aperture shown in Figure 8.16 in which many tagged DVDs will be stacked in the two

different arrangements shown in Figure 8.17. The circularly polarised reader antenna

has to be deployed just against the front side of the DVD stack to read all the tags

attached on the DVDs. Significant reflection from the foam will make this detection

unreliable, since the reflection can cause multiple path propagation and field nulls in-

side of the aperture. Hence, in this appendix, firstly the reflection coefficient of a wave

incident on a dielectric interface as a function of incident angle and polarisation is in-

vestigated theoretically in Section D.2. Then in Section D.3, the structure and method

of working of the absorbing foam are introduced. Based on this introduction and the

analysis on the reflection coefficient at the dielectric interface in Section D.2, the re-

flection from the absorbing foam is evaluated. Thirdly, the structure of the aperture

surrounded by the absorbing foams is described. By combining the discussion in the

previous two sections and the geometric analysis of the reader deployment in front of

the aperture, the reflection occurring inside of the aperture is investigated. Finally, a

conclusion is made of whether the reflection inside the aperture is significant or not,

and improvement by further minimising this reflection is proposed.

D.2 Reflection Coefficient of Waves Incident on a Loss-

less Dielectric Interface

In this section, the reflection coefficient of a wave incident on a material interface in

terms of incident angle and polarisation is given. The theory of reflection coefficient

Page 242

Appendix DEvaluation of Reflections in the Aperture Surrounded by Absorbing Foam Used in

Chapter 8

of waves incident on a material interface has been investigated by many excellent re-

searchers and described in some classical works on electromagnetics, such as “Time-

Harmonic Electromagnetic Fields” by Harrington [93] and “Foundations for Microwave En-

gineering” Collin [94] etc. This section summarises and restructures the work they have

done.

A plane wave incident on a dielectric interface is shown in Figure D.1. On the left side

is material 1 which is characterised by permittivity ε1 and permeability µ1. On the

right side is material 2 which is characterised by permittivity ε2 and permeability µ2.

θi, θr and θt are incident angle, reflected angle and transmitted angle respectively.

qr

qi

qt

y

z

Reflected

Incident

Transmitted

e1, m1 e2, m2

Figure D.1. Plane wave incident on a dielectric interface.

For the continuity of tangential E and H over the entire interface, the y variation of all

three partial fields must be the same. Hence, (D.1) is derived.

k1 sin θi = k1 sin θr = k2 sin θt (D.1)

where the k1 = ω√

µ1ε1 and k2 = ω√

µ2ε2 are the wave numbers of waves within

those two dielectric materials respectively. The two materials are assumed to be loss-

less dielectric materials which means the imaginary parts of the permittivity and per-

meability are zero. From the equality shown above, we have

θi = θr (D.2)

sin θt

sin θi=

k1

k2=

√µ1ε1

µ2ε2(D.3)

Page 243

D.2 Reflection Coefficient of Waves Incident on a Lossless Dielectric Interface

When the electric field of the incident wave is parallel to the z = 0 plane where the ma-

terial interface is, the reflection coefficient of the plane wave incident on the interface

with an angle θi can be expressed by (D.4) [93].

Γ1 =η2 sec θt − η1 sec θi

η2 sec θt + η1 sec θi(D.4)

where η1 =√

µ1ε1

and η2 =√

µ2ε2

are the characteristic impedances of the materials 1

and 2 respectively.

Supposing that material 1 is vacuum or air (ε1 = ε0, µ1 = µ0) and material 2 is a

nonmagnetic lossless dielectric material (µ2 = µ0) with relative permittivity is εr, then

(D.3) becomes (D.5).sin θt

sin θi=

1√εr

(D.5)

The relationship between η1 and η2 can be obtained by (D.6)

η2 =η1√

εr(D.6)

According to (D.5), cos θt can be expressed as a function of θi and εr as shown in (D.7)

cos θt =

√1− sin2 θi

εr(D.7)

By inserting (D.6) and (D.7) into (D.4), a new formula for the reflection coefficient con-

taining only the incident angle in material 1 and relative permittivity of material 2 is

derived as follows.

Γ1 =cos θi −

√εr − sin2 θi

cos θi +√

εr − sin2 θi(D.8)

For magnetic field parallel to the interface, which means that there is electric field com-

ponent perpendicular to the interface unless the incident angle is 90 degrees. The re-

flection coefficient can be expressed as (D.9) [93].

Γ2 =η2 cos θt − η1 cos θi

η2 cos θt + η1 cos θi(D.9)

Similarly, by inserting (D.6) and (D.7) into (D.9), a new formula for the reflection co-

efficient containing only the incident angle and relative permittivity of material 2 is

derived as follows.

Γ2 =

√εr − sin2 θi − εr cos θi√εr − sin2 θi + εr cos θi

(D.10)

Page 244


Chapter 8

According to (D.8) and (D.10), the reflection coefficients Γ1 and Γ2 can be expressed

as a function of the incident angle and the relative permittivity εr of material 2. For

εr = 1.5, the absolute value of reflection coefficients Γ1 and Γ2 are shown in Figure D.2.

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.60

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Incident angle

Ref

lect

ion c

oef

fici

ent

|G |1

|G |2

Figure D.2. The reflection coefficient at a dielectric interface as a function of incident angle,

for εr = 1.5. The blue cure represents |Γ1|, and the red curve represents |Γ2|.

In Figure D.2, two cases of special interest are 1) that of total transmission and 2) that of

total reflection. For |Γ1| defined when the electric field is parallel to the interface (this is

also called perpendicular polarisation), neither of these cases will happen. Moreover,

it is concluded that |Γ1| is positively related to the incident angle. For |Γ2|, defined

when the magnetic field is parallel to the interface (this is also called parallel polari-

sation), total reflection will not occur but there is an angle at which the incident wave

can transmit into the interface completely. This angle is named as polarising angle or

Brewster angle. In addition, when the incident angle is smaller than the polarising an-

gle, the reflection coefficient is negatively related to the incident angle. However, when

the incident angle exceeds this value, the reflection coefficient increases dramatically

as the incident angle increases.

Page 245

D.3 The Structure of the Absorbing Foam and Its Reflection Coefficient

The conclusions and observations made above can also be derived from the formulas

(D.8) and (D.10). For example, if it is desired that a plane wave achieve total transmis-

sion or total reflection on the dielectric interface on which it is incident, either |Γ1| or

|Γ2| has to satisfiy |Γ(1,2)| = 0 and |Γ(1,2)| = 1 respectively.

The case |Γ1| = 0 only happens when the numerator of the right side of (D.8) is

equal to 0 (and the denominator is not 0), i.e. cos θi −√

εr − sin2 θi = 0 (and cos θi +√

εr − sin2 θi 6= 0) which will lead to cos2 θi + sin2 θi = εr. The first of these relations

cannot be true unless the material 2 is, like the material 1, a vacuum or air.

The case |Γ1| = 1 only happens when either cos θi = 0 or√

εr − sin2 θi = 0. The

first situation will result in a 90 incident angle and the second condition cannot occur

unless material 2 is, like material 1, a vacuum or air. Hence, for the incident wave with

electric field parallel to the dielectric interface, there is neither total transmission nor

total reflection.

For |Γ2|, the reflection factor when the magnetic field is parallel to the dielectric inter-

face, the reason that |Γ2| 6= 1 is similar to that above for |Γ1| 6= 1. The case |Γ2| = 0 only

happens when the numerator of the right side of (D.10) is equal to be 0, (and the de-

nominator is not 0), i.e.√

εr − sin2 θi − εr cos θi = 0 (and√

εr − sin2 θi + εr cos θi 6= 0).

The solution of the incident angle to satisfy these two conditions is shown in (D.11),

which is named as polarising angle and Brewster angle as introduced before.

θi = cos−1

√1

εr + 1(D.11)

D.3 The Structure of the Absorbing Foam and Its Re-

flection Coefficient

This section can be divided into two parts, the first describing the structure of the ab-

sorbing foam and the second discussing the reflection from the surface of the absorbing

foam.

Page 246


Chapter 8

The structure and the basic principles of the absorbing foam we are using here can be

found on the website of the manufacture Emerson & Cuming Microwave Products [95].

The cross section of the foam is shown in Figure D.3.

Low loss

Medium loss

High loss

Incident wave

Reflected wave

Ground Plane

Figure D.3. Cross section of the absorbing foam.

If the incident wave meets a medium with very different electrical properties, the

medium will act as a reflector due to the impedance discontinuity. Hence, the first

layer of the absorbing foam is characterised by low dielectric constant (low in both real

and imaginary components). The dielectric constant is increased in the two deeper lay-

ers. There are waves reflected by all the three absorbing layers and the ground plane.

It is believed that the most significant reflection comes from the interface between the

first layer and air. The absorbing frequency band of the foam depends on the thickness

of the foam, particularly for the ECCOSORBr AN series products, and the relationship

between these two factors is the thicker the wider. The model of the foam we are using

is ECCOSORBr AN-79 which is the thickest one in the ECCOSORBr AN series and,

as mentioned before, it covers the widest frequency band from 600MHz to 40GHz .

Due to the structure of the absorbing foam, most of the incident wave is allowed to

propagate into the foam and is attenuated inside the foam. In addition, the prop-

erty of low loss of the first layer makes (D.8) and (D.10) valid to evaluate the reflec-

tion coefficient at the first layer. It is concluded that the absorbing foam can have a

Page 247

D.4 Reflection in the Aperture Surrounded by the Absorbing Foam

generally low reflection coefficient at a wide incident angle range of waves in differ-

ent polarisations, because the dielectric constant of the first layer is low according to

(D.8) and (D.10). The company also claims that “ECCOSORBr AN series is equally

effective against linear, elliptical, or circular polarisation and relatively insensitive to

incidence angles out to 70 degrees.” [96]. By searching the web site of the manufac-

turer, a document discussing the reflectivity as a function of the incident angle and

polarisation of ECCOSORBr AN series is found. Interestingly, the foam we used here,

ECCOSORBrAN-79, is absent from the document. You can find properties of some

members of the AN series in an E & C reference [97], but the foam we used is not

among the types for which properties are given in that reference. Since the manufac-

turer does not give the electrical parameters of the each layer of the foam, we cannot

examine the reflectivity in terms of incident angle and polarisation. However, our

studies suggest that it is quite possible for the manufacturer to achieve their claim of

low reflection for all polarisations out to incident angles of 70 degrees.

D.4 Reflection in the Aperture Surrounded by the Ab-

sorbing Foam

Based on the introduction in Sections D.2 and D.3, the reflection in the aperture shown

in Figure 8.16 and redrawn in Figure D.4 is analysed in this section.

Obviously, to form such aperture, five pieces of foam are needed. In order to distin-

guish the foam pieces surrounding the aperture, the pieces are marked with numbers

from 1 to 5 as shown in Figure D.4. In the experiment description in Section 8.5, we

know that the aperture will be filled with tagged packaged DVDs and the reader an-

tenna is just deployed in front of the aperture to detect the tags in the aperture. The

deployment of the reader antenna in front of the aperture is shown in Figure D.5.

As shown in Figure D.5, the yellow block represents the radiation element of the reader

antenna. The reader antenna used here is a circularly polarised patch antenna manu-

factured by FEIG Electronics company. The radiation element only denotes the top

patch of the patch antenna, but not the ground plane. Since the centre frequency of

Page 248


Chapter 8

300

49

6

610

61

0

114

11

411

4

1

2

3

4

(a) Schematic diagram of the aperture, unit: mm

1

2

3

45

(b) The front view of the real aperture

Figure D.4. Aperture structure illustration.

this antenna is 915MHz and it is a circularly polarised, the top patch is approximately

a 160mm metal square. The reader antenna is not placed in the middle of the front side

of the aperture, because of the DVD stack placement discussed in Section 8.5.

According to the geometric structure of this antenna deployment, the maximum in-

cident angle of the wave radiated from the antenna in relation to each foam piece

surrounding the aperture is given in Figure D.5. The angles are represented by the

symbols θni , n = 1, 2, 3, 4, 5, with the n representing the five absorbing foam pieces

respectively. In particular, the maximum incident angle in relation to the pieces 1, 3

and 5 can be calculated for the geometric structure shown in Figure D.5(b), in which

θ1i = 70.5, θ3

i = 78.9 and θ5i = 19.5; the maximum incident angle in relation to the

pieces 2 and 4 can be calculated by the geometric structure shown in Figure D.5(c),

in which θ2i = θ4

i = 83.5. As mentioned before, the company claims that the ab-

sorbing foam is equally effective against linear, elliptical, or circular polarisation and

relatively insensitive to incidence angles out to 70 degrees. According to the analysis

in Section D.2 this claim is reasonable.

In our case, the incident angle in relation to the pieces 1, 2, 3 and 4 can exceed 70

degrees and even goes up to 83.5 degrees, which are apparently not suitable to obtain

Page 249

D.5 Conclusions

qualified reflectivity in this aperture. However, if we look at Figures D.5(b) and D.5(c),

it is found that when the waves from the reader illuminate the pieces 1, 2, 3 and 4

at a maximum incident angle, the waves will be reflected by these pieces and incident

again on the piece 5 at the complementary angles of the reflected angles from the pieces

1, 2, 3 and 4 which angles are much smaller than 90 degrees. Hence, even though some

waves at large incident angles on the pieces 1 to 4 will cause significant reflection, the

reflection will be absorbed quite effectively by the piece 5. In total, it is believed that

the reflection inside the aperture resulting from the foam surrounding it will be small.

Additionally, these reflections can still be minimised further by enlarging the size of

the aperture and placing the reader antenna in the symmetrical centre of the front side

of the aperture.

D.5 Conclusions

According to the analysis in Section D.3, it is quite possible that the absorbing foam can

achieve relatively good reflectivity at a large incident angles ranging up to 70 degrees

as the manufacturer has claimed, even though we have not found any other related

data from the manufacturer. The investigation of the reader antenna deployment in

front of the aperture was conducted in Section D.4, in which, it is concluded that the

reflection caused by the absorbing foam inside of the aperture can be small.

Page 250


Chapter 8

1

2

3

4Radiation elementof reader antenna

12

0m

m

16

0m

m

21

6m

m

70mm

(a) Front view

21

6m

m1

20

mm

16

0m

m

1

3

5610mm

qi

1

qi

5

qi

3

(b) Cross view

2

4

5

70

mm

70

mm

610mm

qi

4

qi

2

(c) Top view

Figure D.5. The deployment of the reader antenna in front of the aperture.

Page 251

Bibliography

[1] M. Shell. How to Use the IEEEtran LATEXClass. [Online]. Available: http://www.ngi2009.eu/

IEEEtran HOWTO.pdf [29 July 2010].

[2] Bureau International des Poids et Mesures. The international system of units (SI). [Online].

Available: http://www.bipm.org/utils/common/pdf/si brochure 8 en.pdf [29 July 2010].

[3] S. Sarma, D. Brock, and D. Engels, “Radio frequency identification and the electronic product

code,” IEEE Micro, vol. 21, no. 6, pp. 50–54, 2001.

[4] W. Dallenbach and W. Kleinsteuber, “Reflection and absorption of decimeter-waves by plane di-

electric layers,” Hochfreq. u Elektroak, vol. 51, pp. 152–156, 1938.

[5] T. Tanaka, Y. Hashimoto, S. Iwagaki, and O. Hashimoto, “A study on λ/4 type MM-wave absorber

at 50 GHz range,” Transactions of the Institute of Electronics, Information and Communication Engineers,

vol. J76B-II, no. 4, pp. 301–303, Apr. 1993.

[6] Sung-Soo Kim, Yeo-Choon Yoon, and Ki-Ho Kim, “Electromagnetic wave absorbing properties

of high-permittivity ferroelectrics coated with ITO thin films of 377 Ω,” Journal of Electroceramics,

vol. 10, no. 2, pp. 95–101, May 2003.

[7] L. Ukkonen, L. Sydanheimo, and M. Kivikoski, “Effects of metallic plate size on the performance

of microstrip patch-type tag antennas for passive RFID,” IEEE Antennas and Wireless Propagation

Letters, vol. 4, pp. 410–413, 2005.

[8] M. L. NG, “Design of high performance RFID system for metallic item identification,” Ph.D. dis-

sertation, The University of Adelaide, Adelaide, Australia, 2008.

[9] L. Ukkonen, D. Engels, L. Sydanheimo, and M. Kivikoski, “Planar wire-type inverted-F RFID tag

antenna mountable on metallic objects,” in IEEE Antennas and Propagation Society International Sym-

posium, vol. 1, Jun. 2004, pp. 101–104.

[10] B. Yu, S.-J. Kim, B. Jung, F. Harackiewicz, M.-J. Park, and B. Lee, “Balanced RFID Tag Antenna

Mountable on Metallic Plates,” in IEEE Antennas and Propagation Society International Symposium,

Jul. 2006, pp. 3237–3240.

[11] M. L. NG, K. S. Leong, and P. H. Cole, “A small passive UHF RFID tag for

metallic item identification,” in 21st International technical conference on circuit systems,

Computers and Communications (ITC-CSCC), vol. 2, Chiang Mai, Thailand, 2006.

[12] D. Sievenpiper, Lijun Zhang, R. Broas, N. Alexopolous, and E. Yablonovitch, “High-impedance

electromagnetic surfaces with a forbidden frequency band,” IEEE Transactions on Microwave Theory

and Techniques, vol. 47, no. 11, pp. 2059–2074, Nov. 1999.

Page 253

Bibliography

[13] J. McVay, N. Engheta, and A. Hoorfar, “High impedance metamaterial surfaces using hilbert-curve

inclusions,” IEEE Microwave and Wireless Components Letters, vol. 14, no. 3, pp. 130–132, Mar. 2004.

[14] T. Endo, Y. Sunahara, S. Satoh, and T. Katagi, “Resonant frequency and radiation efficiency of

meander line antennas,” Electronics and Communications in Japan (Part II: Electronics), vol. 83, pp.

52–58, 2000.

[15] J. R. Brown, C. R. Lawremce, P. R. Clarke, and W. N. Damerell, “Electromagnetic radiation decou-

pler,” Great Britain Patent publication WO/2007/000 578, June 22, 2006.

[16] J. Landt, “The history of RFID,” IEEE Potentials, vol. 24, no. 4, pp. 8–11, Oct.-Nov. 2005.

[17] The history of RFID technology. RFID Journal. [Online]. Available: http://www.rfidjournal.com/

article/view/1338/1/129 [29 July 2010].

[18] H. Stockman, “Communication by means of reflected power,” Proceedings of the IRE, vol. 36, no. 10,

pp. 1196–1204, Oct. 1948.

[19] K. Finkenzeller, RFID handbook: fundamentals and applications in contactless smart cards and identifica-

tion, 2nd ed. John Wiley and Sons, 2003.

[20] P. H. Cole, “Electronic surveillance system,” Australia Patent No.444 838, 1972.

[21] P. H. Cole, “Electronic surveillance system,” U.S Patent No.3 706 094, 1972.

[22] EPCglobal. [Online]. Available: http://www.epcglobalinc.org/home [29 July 2010].

[23] International Organization for Standarization (ISO). [Online]. Available: http://www.iso.org/

iso/home.htm [29 July 2010].

[24] R. Want, “An introduction to RFID technology,” IEEE Pervasive Computing, vol. 5, no. 1, pp. 25–33,

Jan.-Mar. 2006.

[25] Consideration for RFID technology selection. Atmel corporation. [Online]. Available: http:

//www.atmel.com/dyn/resources/Prod documents/secrerf 3 04.pdf [29 July 2010].

[26] K. S. Leong, “Antenna position analysis and dual-frequency antenna design of high frequancy ratio

for advanced electronic code responding labels,” Ph.D. dissertation, The University of Adelaide,

Adelaide, Australia, 2008.

[27] ETSI EN 302 208-2 V1.2.1, European Telecommunications Standards Institute (ETSI) Std., Apr. 2008.

[28] CFR Title 47-Telecommunication, part15-Radio frequency devices. Federal Communications Com-

mission (FCC). [Online]. Available: http://www.gpo.gov/fdsys/pkg/CFR-2009-title47-vol1/

pdf/CFR-2009-title47-vol1-part15.pdf [29 July 2010].

[29] ETSI EN 302 208-2 V1.1.1, European Telecommunications Standards Institute (ETSI) Std., Sept.

2004.

Page 254

Bibliography

[30] Radiocommunications (Low Interference Potential Devices) Class Licence 2000,

Australia Communications and Media Authority Std. [Online]. Avail-

able: http://www.comlaw.gov.au/ComLaw/legislation/legislativeinstrumentcompilation1.nsf/

0/58DA61C0685CFB85CA2575ED0032C733/$file/RadcomLIPDClassLic2000.pdf

[31] Regulatory status for using RFID in the UHF spectrum. EPCglobal. [Online]. Available:

http://www.epcglobalinc.org/tech/freq reg/RFID at UHF Regulations 20100824.pdf [30 August

2010].

[32] Radio-Frequency Identity Protocols Class-1 Generation-2 UHF RFID Protocol for Communications

at 860MHz-960MHz Version 1.2.0. EPCglobal. [Online]. Available: http://www.epcglobalinc.org/

standards/uhfc1g2/uhfc1g2 1 2 0-standard-20080511.pdf [29 July 2010].

[33] ISO/IEC 18000-6:2004: Information technology–Radio frequency identification for item management–Part

6: Parameters for air interface communications at 860MHz to 960MHz, International Organization for

Standardization Std., 2004.

[34] U. Karthaus and M. Fischer, “Fully integrated passive UHF RFID transponder IC with 16.7µW

minimum RF input power,” IEEE Journal of Solid-State Circuits, vol. 38, no. 10, pp. 1602–1608, Oct.

2003.

[35] F. Fuschini, C. Piersanti, F. Paolazzi, and G. Falciasecca, “Analytical Approach to the Backscattering

from UHF RFID Transponder,” IEEE Antennas and Wireless Propagation Letters, vol. 7, pp. 33–35,

2008.

[36] F. Fuschini, C. Piersanti, F. Paolazzi, and G. Falciasecca, “Electromagnetic and System Level Co-

Simulation for RFID Radio Link Modeling in Real Environment,” in The Second European Conference

on Antennas and Propagation, Nov. 2007, pp. 1–8.

[37] C. A. Balanis, Antenna Theory, 3rd ed. Hoboken, New Jersey: John Wiley, 2005.

[38] Cushcraft Corporation, “Antenna properties and their impact on wireless system performance,”

White Paper, 2002.

[39] The Cost of RFID Equipment. RFID Journal. [Online]. Available: http://www.rfidjournal.com/

faq/20 [2 February 2010].

[40] P. H. Cole, L. Turner, Z. Hu, and D. Ranasinghe, “The Future of RFID,” in Unique Radio Innovation

for the 21st Century, D. Ranasinghe, M. Sheng, and S. Zeadally, Eds. Springer, 2010.

[41] G. De Vita and G. Iannaccone, “Design criteria for the RF section of UHF and microwave passive

RFID transponders,” IEEE Transactions on Microwave Theory and Techniques, vol. 53, no. 9, pp. 2978–

2990, Sept. 2005.

Page 255

Bibliography

[42] H. Nakamoto, D. Yamazaki, T. Yamamoto, H. Kurata, S. Yamada, K. Mukaida, T. Ninomiya,

T. Ohkawa, S. Masui, and K. Gotoh, “A passive UHF RF identification CMOS tag IC using ferro-

electric RAM in 0.35µm technology,” IEEE Journal of Solid-State Circuits, vol. 42, no. 1, pp. 101–110,

Jan. 2007.

[43] W. Che, N. Yan, Y. Yang, and M. Hao, “A low voltage, low power RF/analog front-end circuit for

passive UHF RFID tags,” Chinese Journal of Semiconductors, vol. 29, no. 3, pp. 433–437, Mar. 2008.

[44] H.-L. Lung, S. Lai, H. Lee, T.-B. Wu, R. Liu, and C.-Y. Lu, “Low-temperature capacitor-over-

interconnect (COI) modular FeRAM for SOC application,” IEEE Transactions on Electron Devices,

vol. 51, no. 6, pp. 920–926, Jun. 2004.

[45] No author stated, “Ferroelectric RAM,” Fujitsu Microelectronics Limited, Tech. Rep., Apr.

2006. [Online]. Available: http://www.fujitsu.com/downloads/EDG/binary/pdf/catalogs/

a05000249e.pdf [29 July 2010].

[46] P. H. Young, Electronic communication techniques, 3rd ed. New York: Merill, 1994.

[47] P. Nikitin and K. Rao, “Performance limitations of passive UHF RFID systems,” in IEEE Antennas

and Propagation Society International Symposium, Jul. 2006, pp. 1011–1014.

[48] J. Griffin, G. Durgin, A. Haldi, and B. Kippelen, “RF Tag Antenna Performance on Various Materials

Using Radio Link Budgets,” IEEE Antennas and Wireless Propagation Letters, vol. 5, no. 1, pp. 247–

250, Dec. 2006.

[49] D. Dobkin and S. Weigand, “Environmental effects on RFID tag antennas,” in IEEE International

Microwave Symposium Digest, MTT-S, Jun. 2005.

[50] J. Prothro, G. Durgin, and J. Griffin, “The effects of a metal ground plane on RFID tag antennas,”

in IEEE Antennas and Propagation Society International Symposium, Jul. 2006, pp. 3241–3244.

[51] P. Nikitin and K. Rao, “Theory and measurement of backscattering from RFID tags,” IEEE Antennas

and Propagation Magazine, vol. 48, no. 6, pp. 212–218, Dec. 2006.

[52] B. Jiang, K. Fishkin, S. Roy, and M. Philipose, “Unobtrusive long-range detection of passive RFID

tag motion,” IEEE Transactions on Instrumentation and Measurement, vol. 55, no. 1, pp. 187–196, Feb.

2006.

[53] S. Hodges, A. Thorne, H. Mallinson, and C. Floerkemeier, “Assessing and optimizing the range of

UHF RFID to enable real-world pervasive computing applications,” in PERVASIVE’07: Proceedings

of the 5th international conference on Pervasive computing.

[54] T. Rappaport, Wireless communications-principles and practie, 2nd ed. Prentice Hall, 2002.

[55] Product overview. HiggsTM-2. [Online]. Available: http://www.alientechnology.com/docs/

products/DS H2.pdf [29 July 2010].

Page 256

Bibliography

[56] 3M. Conductive Adhesive Transfer Tape 9703 (Data Sheet).

[Online]. Available: http://multimedia.3m.com/mws/mediawebserver?mwsId=

66666UuZjcFSLXTtnxfVMxz6EVuQEcuZgVs6EVs6E666666-- [29 July 2010].

[57] M. Santra and K. Limaye, “Estimation of complex permittivity of arbitrary shape and size dielec-

tric samples using cavity measurement technique at microwave frequencies,” IEEE Transactions on

Microwave Theory and Techniques, vol. 53, no. 2, pp. 718–722, Feb. 2005.

[58] R. Plumb and H. Ma, “Swept frequency reflectometer design for in-situ permittivity measure-

ments,” IEEE Transactions on Instrumentation and Measurement, vol. 42, no. 3, pp. 730–734, Jun. 1993.

[59] W. Choi, H. Son, C. Shin, J. H. Bae, and G. Choi, “RFID tag antenna with a meandered dipole and

inductively coupled feed,” in IEEE Antennas and Propagation Society International Symposium, Jul.

2006, pp. 619–622.

[60] G. Marrocco, “The art of UHF RFID antenna design: impedance-matching and size-reduction tech-

niques,” IEEE Antennas and Propagation Magazine, vol. 50, no. 1, pp. 66–79, Feb. 2008.

[61] H. Nakano, H. Tagami, A. Yoshizawa, and J. Yamauchi, “Shortening ratios of modified dipole

antennas,” IEEE Transactions on Antennas and Propagation, vol. 32, no. 4, pp. 385–386, Apr. 1984.

[62] J. Rashed and C. T. Tai, “A new class of wire antennas,” in Antennas and Propagation Society Interna-

tional Symposium, vol. 20, May 1982, pp. 564–567.

[63] G. Marrocco, “Gain-optimized self-resonant meander line antennas for RFID applications,” IEEE

Antennas and Wireless Propagation Letters, vol. 2, pp. 302–305, 2003.

[64] N. Michishita, Y. Yamada, and N. Nakakura, “Miniaturization of a small meander line antenna by

loading a high εr material,” in Communications, 2004 and the 5th International Symposium on Multi-

Dimensional Mobile Communications Proceedings. The 2004 Joint Conference of the 10th Asia-Pacific Con-

ference, vol. 2, 29 Aug.-01 Sept. 2004, pp. 651–654.

[65] Y. Yamada and N. Michishita, “Antenna efficiency improvement of a miniaturized meander line

antenna by loading a high εr material,” in IEEE International Workshop on Antenna Technology: Small

Antennas and Novel Metamaterials, IWAT, Mar. 2005, pp. 159–162.

[66] V. Fouad Hanna, “Finite boundary corrections to coplanar stripline analysis,” Electronics Letters,

vol. 16, no. 15, pp. 604 –606, Jul. 1980.

[67] ALN-9540 Squiggle Inlay. Alien Technology Corporation. [Online]. Available: http://www.

alientechnology.com/docs/products/DS ALN 9540 Squiggle.pdf [29 July 2010].

[68] ALN-9562 Squiggle SH-Inlay. Alien Technology Corporation. [Online]. Available: http:

//www.alientechnology.com/docs/products/ALN 9562 Squiggle-SH.web.pdf [29 July 2010].

Page 257

Bibliography

[69] P. Raumonen, L. Sydanheimo, L. Ukkonen, M. Keskilammi, and M. Kivikoski, “Folded dipole an-

tenna near metal plate,” in IEEE Antennas and Propagation Society International Symposium, vol. 1,

Jun. 2003, pp. 848–851.

[70] S. M. Pinho, L. M. Gregori, R. C. R. Nunes, and B. G. Sores, “Performance of radar absorbing

materials by waveguide measurements for X- and Ku-band frequencies,” European Polymer Journal,

vol. 38, pp. 2321–2327, 2002.

[71] B. A. Munk, Frequency Selective Surfaces: Theory and Design. John Wiley & Sons, 2000.

[72] J. D. Joannopoulos, R. D. Meade, and J. N. Winn, Photonic Crystals. Princeton University Press,

1995.

[73] F. Yang and Y. R. Samii, Electromagnetic Band Gap Structures in Antenna Engineering. Cambridge

University Press, 2008.

[74] Y. Eli, “Inhibited Spontaneous Emission in Solid-State Physics and Electronics,” Physical review

letters, vol. 58, no. 20, pp. 2059–2062, May 1987.

[75] H. Sagan, Space-Filling Curves. New York: Springer-Verlag, 1994.

[76] S. Best, “A comparison of the performance properties of the Hilbert curve fractal and meander line

monopole antennas,” Microwave and Optical Technology Letters, vol. 35, no. 4, pp. 258–262, Nov. 2002.

[77] K. Vinoy, K. Jose, V. Varadan, and V. Varadan, “Resonant frequency of Hilbert curve fractal anten-

nas,” in IEEE Antennas and Propagation Society International Symposium, vol. 3, Boston, USA, 2001,

pp. 648–651.

[78] J. McVay, A. Hoorfar, and N. Engheta, “Small dipole-antenna near Peano high-impedance sur-

faces,” in IEEE Antennas and Propagation Society Symposium, vol. 1, Monterey, California, 2004, pp.

305–308.

[79] I. J. Bahl and P. Bhartia, Microstrip Antennas. Dedham, Massachusetts: Artech House, 1980.

[80] D. Pozar, “Microstrip antennas,” Proceedings of the IEEE, vol. 80, no. 1, pp. 79–91, Jan. 1992.

[81] W. F. Richards, “Microstrip Antennas,” in Antenna Handbook: Theory, Applications and Design, Y. T.

Lo and S. W. Lee, Eds., New York, 1988, ch. 10.

[82] F. Lu, X. Chen, and T. T. Ye, “Performance analysis of stacked RFID tags,” in IEEE International

Conference on RFID, Apr. 2009, pp. 330–337.

[83] S. Weigand and D. Dobkin, “Multiple RFID tag plane array effects,” in IEEE Antennas and Propaga-

tion Society International Symposium, Jul. 2006, pp. 1027–1030.

[84] F. R. Byers, “Care and handling of CDs and DVDs,” Council on Library and Information Resources,

Tech. Rep., Oct. 2003. [Online]. Available: http://www.clir.org/pubs/reports/pub121/pub121.pdf

[29 July 2010].

Page 258

Appendix D Bibliography

[85] SPI Resin Identification Code - Guide to Correct Use. The Society of the Plastics Industry, Inc.

[Online]. Available: http://www.plasticsindustry.org/CareerCenter/content.cfm?ItemNumber=

823 [29 July 2010].

[86] B. Riddle, J. Baker-Jarvis, and J. Krupka, “Complex permittivity measurements of common plastics

over variable temperatures,” IEEE Transactions on Microwave Theory and Techniques, vol. 51, no. 3,

pp. 727–733, Mar. 2003.

[87] J. D. Bigley, “Secure optical information disc having a minimized metal layer,” U.S. Patent US

7,187,645 B2, Feb. 03, 2005.

[88] C. Grosvenor, R. Johnk, J. Baker-Jarvis, M. Janezic, and B. Riddle, “Time-domain free-field mea-

surements of the relative permittivity of building materials,” IEEE Transactions on Instrumentation

and Measurement, vol. 58, no. 7, pp. 2275–2282, Jul. 2009.

[89] ECCOSORBr AN. Emerson & Cuming Microwave Products. [Online]. Available: http:

//www.eccosorb.com/file/1079/an.pdf [29 July 2010].

[90] P. H. Cole and D. W. Engels, “21st Century Supply Chain Technology,” White Paper,

Auto-ID Lab, Apr. 2006. [Online]. Available: http://www.autoidlabs.org/uploads/media/

AUTOIDLABS-WP-SWNET-015.pdf

[91] G. Raballand and A. C. Enrique, “How Do Differing Standards Increase Trade Costs? The Case of

Pallets,” The World Economy, vol. 30, no. 4, pp. 685–702, Apr. 2007.

[92] ISO 6780 Flat pallets for intercontinental materials handling – Principal dimensions and tolerances, The

International Organization for Standardization (ISO) Std., 2003.

[93] R. F. Harrington, Time-Harmonic Electromagnetic Fields. McGraw-Hill, 1961.

[94] R. E. Collin, Foundations for Microwave Engineering, 2nd ed. McGraw-Hill, 1992.

[95] Basic Principle of a Graded Absorber. Emerson & Cuming Microwave Products. [Online].

Available: http://www.eccosorb.com/pages/119/Basic%20Principle%20of%20a%20Graded%

20Absorber [29 July 2010].

[96] ECCOSORB AN. Emerson & Cuming Microwave Products. [Online]. Available: http:

//www.eccosorb.com/products/2/ECCOSORB%20AN [29 July 2010].

[97] ECCOSORBr AN Polarizations. Emerson & Cuming Microwave Products. [Online]. Available:

http://www.eccosorb.com/file/665/anpolarizations.pdf [29 July 2010].

Page 259

solutions for hard-to-tag objects in uhf rfid...

Documents