solutions for hard-to-tag objects in uhf rfid...
TRANSCRIPT
Solutions for Hard-to-Tag Objects in
UHF RFID Systems
by
Zhonghao Hu
B.E. (Electrical & Electronic),Northwestern Polytechnical University, China, 2006.
Thesis submitted for the degree of
Doctor of Philosophy
in
School of Electrical and Electronic Engineering
The University of Adelaide, Australia
March 2011
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
6.4 A shielding tunnel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
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
8.8 A shielding tunnel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
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
8.2 Reading range test results of the tag shown in Figure 8.10 . . . . . . . . . 188
8.3 Reading range test results of the tag shown in Figure 8.11 . . . . . . . . . 189
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
C.2 Original testing data corresponding to Table 8.2, unit: mm . . . . . . . . 240
C.3 Original testing data corresponding to Table 8.3, 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
Chapter 1 Introduction and Motivation
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.
1.3 Original Contributions
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
Chapter 1 Introduction and Motivation
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
1.3 Original Contributions
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
Chapter 1 Introduction and Motivation
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
1.4 Thesis Structure
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 1 Introduction and Motivation
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
1.4 Thesis Structure
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
Chapter 2 RFID Background
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 Regulations and Standards
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
Chapter 2 RFID Background
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
2.4 Regulations and Standards
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
Chapter 2 RFID Background
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
2.4 Regulations and Standards
(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
Chapter 2 RFID Background
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
Page 24
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
Chapter 3 Operating Range Evaluation of RFID Systems
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 Fundamental Parameters of Antennas and the Friis Equation
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
Chapter 3 Operating Range Evaluation of RFID Systems
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
3.2 Fundamental Parameters of Antennas and the Friis Equation
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
Chapter 3 Operating Range Evaluation of RFID Systems
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
3.2 Fundamental Parameters of Antennas and the Friis Equation
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
Chapter 3 Operating Range Evaluation of RFID Systems
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.
3.4 Threshold Power of a Transponder
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
Chapter 3 Operating Range Evaluation of RFID Systems
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
3.4 Threshold Power of a Transponder
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
Chapter 3 Operating Range Evaluation of RFID Systems
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
3.4 Threshold Power of a Transponder
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
Chapter 3 Operating Range Evaluation of RFID Systems
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
Chapter 3 Operating Range Evaluation of RFID Systems
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
Chapter 3 Operating Range Evaluation of RFID Systems
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
3.7 Interpretation and limitations of the Friis Transmission Equation in an RFIDPerspective
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
Chapter 3 Operating Range Evaluation of RFID Systems
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
3.7 Interpretation and limitations of the Friis Transmission Equation in an RFIDPerspective
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
Chapter 3 Operating Range Evaluation of RFID Systems
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
Chapter 3 Operating Range Evaluation of RFID Systems
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
3.8 The Use of S-parameters in Analysing the Operating Range of RFID Systems
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
Chapter 3 Operating Range Evaluation of RFID Systems
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
3.8 The Use of S-parameters in Analysing the Operating Range of RFID Systems
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
Chapter 3 Operating Range Evaluation of RFID Systems
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
3.8 The Use of S-parameters in Analysing the Operating Range of RFID Systems
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
Chapter 3 Operating Range Evaluation of RFID Systems
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
3.8 The Use of S-parameters in Analysing the Operating Range of RFID Systems
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
Chapter 3 Operating Range Evaluation of RFID Systems
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
Chapter 4 Analysis and Design of Meander Line Dipole Antennas
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
Chapter 4 Analysis and Design of Meander Line Dipole Antennas
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=8, Equation (4.11)
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=2, Equation (4.11)
m=8, HFSS
m=8, Equation (4.11)
m=14, HFSS
m=14, Equation (4.11)
Meander line height 10mmh=
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=2, Equation (4.11)
m=8, HFSS
m=8, Equation (4.11)
m=14, HFSS
m=14, Equation (4.11)
The distance between the meanders and the end of MDA (mm)D
Meander line height 10mmh=
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
Chapter 4 Analysis and Design of Meander Line Dipole Antennas
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
4.3 Modifications on Equation (4.11) for RFID Tag Antenna Design
(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
Chapter 4 Analysis and Design of Meander Line Dipole Antennas
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
4.3 Modifications on Equation (4.11) for RFID Tag Antenna Design
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
Chapter 4 Analysis and Design of Meander Line Dipole Antennas
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
4.3 Modifications on Equation (4.11) for RFID Tag Antenna Design
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
Chapter 4 Analysis and Design of Meander Line Dipole Antennas
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
4.3 Modifications on Equation (4.11) for RFID Tag Antenna Design
(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
Chapter 4 Analysis and Design of Meander Line Dipole Antennas
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
Chapter 4 Analysis and Design of Meander Line Dipole Antennas
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
4.5 Radiation Pattern and Efficiency
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
Chapter 4 Analysis and Design of Meander Line Dipole Antennas
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
4.5 Radiation Pattern and Efficiency
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
Chapter 4 Analysis and Design of Meander Line Dipole Antennas
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
Chapter 5 A Security Tag Design
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
Chapter 5 A Security Tag Design
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
5.4 The Security Tag Antenna Design
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
Chapter 5 A Security Tag Design
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
5.4 The Security Tag Antenna Design
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
Chapter 5 A Security Tag Design
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
5.4 The Security Tag Antenna Design
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
Chapter 5 A Security Tag Design
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
5.4 The Security Tag Antenna Design
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
Chapter 5 A Security Tag Design
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
5.4 The Security Tag Antenna Design
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
Chapter 5 A Security Tag Design
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
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
6.2 The Antenna on Metal Problem
+ 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
Chapter 6 Solutions for the Antenna on Metal Problem
(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
6.2 The Antenna on Metal Problem
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
Chapter 6 Solutions for the Antenna on Metal Problem
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
6.2 The Antenna on Metal Problem
Figure 6.4. A shielding tunnel. The size of this tunnel inside is 1826mm×915mm×690mm.
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
Chapter 6 Solutions for the Antenna on Metal Problem
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 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
Chapter 6 Solutions for the Antenna on Metal Problem
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
6.3 Previous Solutions to the Problem
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
Chapter 6 Solutions for the Antenna on Metal Problem
(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
6.3 Previous Solutions to the Problem
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
Chapter 6 Solutions for the Antenna on Metal Problem
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
Chapter 6 Solutions for the Antenna on Metal Problem
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
Page 120
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
7.1 Introduction and Outline
7.1 Introduction and Outline
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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
-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.
7.5 Patch Antenna Resonant Property Analysis
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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
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
7.5 Patch Antenna Resonant Property Analysis
(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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
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
7.5 Patch Antenna Resonant Property Analysis
(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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
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
7.5 Patch Antenna Resonant Property Analysis
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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
• 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
7.5 Patch Antenna Resonant Property Analysis
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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
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
7.5 Patch Antenna Resonant Property Analysis
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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
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
7.5 Patch Antenna Resonant Property Analysis
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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
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.
7.6 Slitted Decoupler Parameter Settings
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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
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 Slitted Decoupler Parameter Settings
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
L=92mmL=90.5mmL=89.7mm
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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
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 Slitted Decoupler Parameter Settings
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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
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
L=92mmL=90.5mmL=89.7mm
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
7.6 Slitted Decoupler Parameter Settings
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
shown in Figure 7.22.
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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
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 Slitted Decoupler Parameter Settings
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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
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
7.7 A Dipole on the Slitted Decoupler
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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
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
7.7 A Dipole on the Slitted Decoupler
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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
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
7.7 A Dipole on the Slitted Decoupler
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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
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
7.7 A Dipole on the Slitted Decoupler
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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
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 2 (76.6mm×35mm) 3960mm
Decoupler 3 (74mm×80mm) 2810mm
Decoupler 4 (75mm×80mm) 3090mm
(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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
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.
Decoupler Name (size) Reading range
Decoupler 2 (76.6mm×35mm) 2460mm
Decoupler 4 (75mm×80mm) 2890mm
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).
Decoupler Name (size) Reading range
Decoupler 2 (76.6mm×35mm) 3140mm
Decoupler 4 (75mm×80mm) 3450mm
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
Chapter 7 The Slitted Decoupler Design for Metallic Item Detection
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
Page 170
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
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
8.1 Introduction and Motivation
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
Chapter 8 Detection of Massive Numbers of DVDs
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
8.1 Introduction and Motivation
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
Chapter 8 Detection of Massive Numbers of DVDs
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.
Chapter 8 Detection of Massive Numbers of DVDs
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
Chapter 8 Detection of Massive Numbers of DVDs
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
8.3 Theoretical Analysis and Simulation Verification of the Effect on a UniformPlane Wave from a Thin Metal Film
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
Chapter 8 Detection of Massive Numbers of DVDs
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
8.3 Theoretical Analysis and Simulation Verification of the Effect on a UniformPlane Wave from a Thin Metal Film
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
Chapter 8 Detection of Massive Numbers of DVDs
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.
Figure 8.8. A shielding tunnel. The size of this tunnel inside is 1826mm×915mm×690mm.
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
8.4 Investigation of Tag Labelling Method
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
Chapter 8 Detection of Massive Numbers of DVDs
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
8.4 Investigation of Tag Labelling Method
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.
Table 8.2. Reading range test results of the tag shown in Figure 8.10.
(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
Chapter 8 Detection of Massive Numbers of DVDs
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.
Table 8.3. Reading range test results of the tag shown in Figure 8.11.
(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
8.4 Investigation of Tag Labelling Method
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
Chapter 8 Detection of Massive Numbers of DVDs
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
8.5 DVD Detection in a Stack
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
Chapter 8 Detection of Massive Numbers of DVDs
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
8.5 DVD Detection in a Stack
Interrogator
(a)
Interrogator
(b)
Interrogator
(c)
Figure 8.15. Three forms of testing a DVD stack in terms of the three testing schemes
shown in Figure 8.13.
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
Chapter 8 Detection of Massive Numbers of DVDs
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
8.5 DVD Detection in a Stack
(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
Chapter 8 Detection of Massive Numbers of DVDs
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
8.5 DVD Detection in a Stack
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
Chapter 8 Detection of Massive Numbers of DVDs
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
8.5 DVD Detection in a Stack
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
Chapter 8 Detection of Massive Numbers of DVDs
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.
• Testing scheme 3
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.
• Testing scheme 2
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
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
Chapter 8 Detection of Massive Numbers of DVDs
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
8.6 Further Validation
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
Chapter 8 Detection of Massive Numbers of DVDs
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
8.6 Further Validation
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
Chapter 8 Detection of Massive Numbers of DVDs
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
8.6 Further Validation
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
Chapter 8 Detection of Massive Numbers of DVDs
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
8.6 Further Validation
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
Chapter 8 Detection of Massive Numbers of DVDs
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
8.6 Further Validation
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
Chapter 8 Detection of Massive Numbers of DVDs
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
8.6 Further Validation
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
Chapter 8 Detection of Massive Numbers of DVDs
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
Chapter 8 Detection of Massive Numbers of DVDs
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
Chapter 8 Detection of Massive Numbers of DVDs
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
Page 220
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.
9.2 Recommendations on Future Work
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
Chapter 9 Conclusions and Future Work
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
9.2 Recommendations on Future Work
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
Chapter 9 Conclusions and Future Work
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
Chapter 9 Conclusions and Future Work
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
Page 234
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
Table C.2. Original testing data corresponding to Table 8.2, unit: mm.
(a)↑ (a)→ (b)↑ (b)→Opening A 4530 110 680 420
Spine 90 460 2060 20
Table C.3. Original testing data corresponding to Table 8.3, unit: mm.
(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
Appendix DEvaluation of Reflections in the Aperture Surrounded by Absorbing Foam Used in
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
Appendix DEvaluation of Reflections in the Aperture Surrounded by Absorbing Foam Used in
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
Appendix DEvaluation of Reflections in the Aperture Surrounded by Absorbing Foam Used in
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
Appendix DEvaluation of Reflections in the Aperture Surrounded by Absorbing Foam Used in
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
Page 252
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