Antenna Positioning Analysis and Dual-Frequency Antenna Design of High Frequency Ratio for Advanced Electronic Code Responding Labels Rev. 0.

Transcription

1 Antenna Positioning Analysis and Dual-Frequency Antenna Design of High Frequency Ratio for Advanced Electronic Code Responding Labels Rev. 0.3 by Kin Seong Leong B.E. (Electrical & Electronic, First Class Honours), The University of Adelaide, Australia, Thesis submitted for the degree of Doctor of Philosophy in Electrical and Electronic Engineering University of Adelaide 2007

2 c 2007 Kin Seong Leong All Rights Reserved

3 Contents Contents iii Revision History xi Conventions xiii Publications xv Chapter 1. Introduction Title Background of Research Motivation Contribution Thesis Structure Chapter Structure Chapter 2. RFID System Introduction History of RFID A simple RFID System Variants of RFID Active and Passive RFID Tag Backscattering Common Frequency Bands for RFID Operations RFID and Supply Chain EPC Network Page iii

4 Contents Early stage EPC Network EPC Network at Current State RFID Standards Chapter 3. Path Loss and Position Analysis Introduction Path Loss Model Free Space Path Loss In-building Path Loss Experiment Path Loss Model for RFID Simple Two Reader Interference RFID Positioning Analysis Power Density Antenna Gain Pattern Frequency Spectrum Channeling Listen Before Talk Simulation Concept Results and Discussion One Antenna Simulation Two Antennas Simulation One antenna on the ground Antennas Operating at Different Channels A More Hostile Environment Real Life Application Reader Synchronisation Practical Position of Antenna Chapter Summary Page iv

5 Contents Chapter 4. Reader Synchronisation Introduction Background ETSI EPC Class 1 Generation 2 Protocol Problem in dense reader environment Reader synchronisation Actual implementation Connectivity Positioning of LBT Sensors Antenna Positioning Case Study Synchronisation Fine-tuning Reduction of output power Reduction of overall reader talking time Use of external sensors RF opaque or RF absorbing materials Frequent rearrangement of channels Variation of Synchronisation Separation of transmitting and receiving channels Separation of RFID and non-rfid signals Conclusion Chapter 5. Operational Considerations Introduction RFID EMC Background Frequency Hopping Spread Spectrum (FHSS) Page v

6 Contents Listen Before Talk (LBT) RFID Protocol Path Loss Measurement Experiment Main Objectives: Experimental Setup: Room Grid Experimental Results: Discussion: Sources of Simulation Error Path Loss Model Reflection, Refraction, and Diffraction Radiation Pattern of Antenna Simulation Result Interpretation and Analysis Second Carrier Sensing Effect of the Position of Tag Effect of Environment Factor Combining First and Second Carrier Sensing Conclusion Chapter 6. RFID Antenna Design Introduction Antenna Theory Antenna Parameters Resonant Circuit for Antenna Challenges in RFID Tag Antenna Design RFID Chips RFID Readers Page vi

7 Contents HF Reader UHF Reader HF RFID Antenna Design Simple Loop Antenna HF Planar Spiral Coil Antenna HF Antenna for a New Wine Closure HF Antenna for a Pig Ear Tag UHF RFID Antenna Design Common UHF RFID Tag Simple Planar Dipole UHF antenna for Sheep UHF antenna for Pig UHF antenna for Beer Keg UHF antenna for Wine Cork RFID Tag Fabrication Antenna Design and Simulation Ansoft HFSS Scripting in HFSS Plotting in HFSS ISO-Pro Conclusion Chapter 7. Dual Frequency RFID Antenna Introduction Current Dual-Frequency Antenna Design Design Aims Dual-Frequency Antenna Design Page vii

8 Contents Independent HF and UHF Antenna Design Quick Feasibility Test Tunability Test Redesign of the UHF dipole Compatibility of dipole in HF Merging of the new UHF dipole with the HF coil Antenna Fabrication and Testing Miniaturisation of Dual Frequency RFID Antenna with High Frequency Ratio Improved Novel Design Testing Final Design Conclusion Chapter 8. Measurement Introduction Background and Literature Review Settings and Connection Testings and Results A balanced bow tie antenna Half bow-tie antenna on ground plane Discussion and Improvement Conclusion Calculation of Coplanar Strips Calculation of S parameters and Power Transfer Efficiency Raw Experiment Data (To be removed in final version) Notes (To be removed in final version) Appendix A. Calculation of Coplanar Strips 195 Page viii

9 Contents Appendix B. MATLAB Code for Path Loss Calculation 197 Appendix C. MATLAB Code for HFSS VB Script Generation 203 Appendix D. MATLAB Code for Inductance Calculation 209 D.1 Main Code D.2 Inductance Calculation D.3 Mutual Inductance - Positive D.4 Mutual Inductance - Negative Appendix E. Path Loss Experiment 217 E.1 Preliminary Setting Up Procedure E.2 Experiment Procedure and Results E.2.1 Map E.2.2 Signal strength at different distance (within and between buildings) E.2.3 Reflection from typical wall and a conductive fence E.2.4 Propagation Loss Outdoors E.3 Conclusion Bibliography 227 Page ix

10 Page x

11 Revision History This page will be removed in the final version. Version Date Changes Dec Inclusion of Chapter 1 to Chapter 5, and Appendix A. Some sections in Chapter 2 need updates Feb Updated using feedback from Alf and Bevan. Added Conventions page Mar Added Chapter 7 and first part of Chapter May 2007 Chapter 6 completed. Draft of chapter 8 added. Page xi

12 Page xii

13 Conventions Typesetting This thesis is typeset in Times New Roman and Sans-Serif using LATEX2e. Referencing and citation style are based on the Institute of Electrical and Electronics Engineers (IEEE) Transaction style [1]. For electronics sources, the last access date is enclosed within parentheses and is placed immediate behind the author(s) name [2]. Units The International System of Units (abbreviated SI units) [3] is used in this thesis. Prefixes nano, micro, and mili is preferred but prefix cm is avoided. Spelling English spelling in this thesis is based on Australian English. The only exception is in some special cases the proper noun is used. For example: Auto-ID Center, not Auto-ID Centre. Page xiii

14 Page xiv

15 Publications Journal [1] K. S. Leong, M. L. Ng, A. Grasso, and P. H. Cole, Dense RFID reader deployment in Europe using synchronization, Journal of Communications, vol. 1, no. 7, pp. 9 16, Conference [1] K. S. Leong, M. L. Ng, and P. H. Cole, The reader collision problem in RFID systems, in IEEE 2005 International Symposium on Microwave, Antenna, Propagation and EMC Technologies for Wireless Communications, Beijing, China, [2] K. S. Leong, M. L. Ng, and P. H. Cole, Positioning analysis of multiple antennas in a dense RFID reader environment, in 2006 International Symposium on Applications and the Internet (SAINT) Workshop, RFID and Extended Network: Deployment of Technologies and Applications, Phoenix, Arizona, USA, [3] K. S. Leong, M. L. Ng, A. Grasso, and P. H. Cole, Synchronization of RFID readers for dense RFID reader environments, in 2006 International Symposium on Applications and the Internet (SAINT) Workshop, RFID and Extended Network: Deployment of Technologies and Applications, Phoenix, Arizona, USA, [4] K. S. Leong, M. L. Ng, and P. H. Cole, Operational considerations in simulation and deployment of RFID systems, in 17th International Zurich Symposium on Electromagnetic Compatibility, Singapore, [5] K. S. Leong, M. L. Ng, and P. H. Cole, Dual-frequency antenna design for RFID application, in 21st International Technical Conference on Circuits/Systems, Computers and Communications (ITC-CSCC 2006), Chiang Mai, Thailand, [6] K. S. Leong, M. L. Ng, and P. H. Cole, Investigation of RF cable effect on RFID tag antenna impedance measurement, in IEEE AP-S International Symposium, Honolulu, Hawaii, USA, [7] K. S. Leong, M. L. Ng, and P. H. Cole, Miniaturization of dual frequency RFID antenna with high frequency ratio, in IEEE AP-S International Symposium, Honolulu, Hawaii, USA, [8] K. S. Leong, M. L. Ng, and P. H. Cole, Investigation on the deployment of HF and UHF RFID tag in livestock identification, in IEEE AP-S International Symposium, Honolulu, Hawaii, USA, Page xv

16 Publications [9] M. L. Ng, K. S. Leong, and P. H. Cole, Design and miniaturization of an RFID tag using a simple rectangular patch antenna for metallic object identification, in IEEE AP-S International Symposium, Honolulu, Hawaii, USA, [10] M. L. Ng, K. S. Leong, and P. H. Cole, Analysis of constraints in small UHF RFID tag design, in IEEE 2005 International Symposium on Microwave, Antenna, Propagation and EMC Technologies for Wireless Communications, Beijing, China, [11] M. L. Ng, K. S. Leong, D. M. Hall, and P. H. Cole, A small passive UHF RFID tag for livestock identification, in IEEE 2005 International Symposium on Microwave, Antenna, Propagation and EMC Technologies for Wireless Communications, Beijing, China, [12] 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 Circuits/Systems, Computers and Communications (ITC-CSCC 2006), Chiang Mai, Thailand, [13] D. C. Ranasinghe, K. S. Leong, M. L. Ng, D. W. Engels, and P. H. Cole, A distributed architecture for a ubiquitous item identification network, in Seventh International Conference on Ubiquitous computing, Tokyo, Japan, [14] D. C. Ranasinghe, K. S. Leong, M. L. Ng, D. W. Engels, and P. H. Cole, A distributed architecture for a ubiquitous RFID sensing network, in 2nd International Conference on Intelligent Sensors, Sensor Networks and Information Processing (ISSNIP), Melbourne, Australia, Non-refereed [1] K. S. Leong, and M. L. Ng, A simple EPC enterprise model, in Auto-ID Labs Workshop, Zurich, [2] K. S. Leong, M. L. Ng, and D. W. Engels, EPC network architecture, in Auto-ID Labs Workshop, Zurich, [3] M. L. Ng, K. S. Leong, and D. W. Engels, Prospects for ubiquitous item identification, in Auto-ID Labs Workshop, Zurich, Page xvi

17 Chapter 1 Introduction T HIS chapter explains the title of this thesis, which is Antenna Positioning Analysis and Dual-Frequency Antenna Design of High Frequency Ratio for Advanced Electronic Code Responding Labels. The background of the research topic, Radio Frequency Identification (RFID), is presented. The motivation and contribution to this research is discussed, followed by the elaboration of the structure of this thesis. Page 1

18 1.1 Title 1.1 Title The title of this thesis is Antenna Positioning Analysis and Dual-Frequency Antenna Design of High Frequency Ratio for Advanced Electronic Code Responding Labels. There are two related areas of research presented in this thesis. The first area is Antenna Positioning Analysis, while the second area is Dual-Frequency Antenna Design of High Frequency Ratio for Advanced Electronic Code Responding Labels. In Antenna positioning analysis, the term antenna refers to the Radio Frequency Identification (RFID) reader s antenna. The positioning analysis done is on the placement of RFID readers antenna to minimise readers interference, to offer more coverage while conforming to local regulations. Dual-Frequency Antenna Design of High Frequency Ratio for Advanced Electronic Code Responding Labels involves the design of an RFID tag antenna and is not related to RFID reader antenna. Frequency ratio is computed through the division of the higher frequency by the lower frequency and hence is always greater than 1. A dual frequency antenna with low frequency ratio (<5) is common, but not for a dual frequency antenna with high frequency ratio. The two frequencies of interest in this thesis are the HF band (13.56 MHz) and the UHF band ( MHz); the intended dual frequency antenna has a frequency ratio in excess of 70. The work presented in this thesis includes the design of HF tag antenna and UHF tag antenna as will be shown in details in Section Background of Research Radio Frequency Identification (RFID) forms the background of this research. An RFID system, is in fact a very complex system, which involves various fields of studies, such as antenna design, signal processing, RF hardware design and information networking. The focus will be on an RFID system for supply chain, which focuses on either HF or UHF RFID system using passive RFID tags. A simple RFID system is as shown in Fig A detailed review of RFID system is presented in Chapter 2. Page 2

19 Chapter 1 Introduction RFID Network (EPC Network) 1 2 Host RFID Computer Reader Antenna 3 RFID Tag Figure 1.1. A simple RFID system. (1) Host computer to link RFID readers to RFID Network (also known as Electronic Product Code (EPC) Network). (2) An RFID reader with one or more reader antennas connected. (3) Passive RFID tags powered up by reader antennas so that these tags can be read by the readers. The research areas involved, according to the societies of the Institute of Electrical and Electronics Engineers (IEEE), are [4]: Antennas and Propagation Electromagnetic Compatibility Microwave Theory and Techniques 1.3 Motivation RFID is not a new technology. As will be discussed in Chapter 2, the history of RFID can be traced back to the early 20th century [5]. However, it is only in the late 1990 s that RFID gained widespread attention when its potential in automating the supply chains was investigated. It is believed that RFID can revolutionise the supply chain using real time monitoring, which can offers instant restocking capabilities, elimination of overstocking problem, reduction in counterfeiting and many others. To fully embrace the potential of an RFID system in a supply chain, we need to attach an RFID tag to every box, carton or pallet in the supply chain. Also, every RFID tag contains a chip which holds a unique identification code of the object the tag is attached to. Hence only passive RFID tags are feasible to be deployed in this manner, or else the Page 3

20 1.4 Contribution total cost of any product with an RFID tag attached will increase to an unacceptable level. As mentioned before, the focus of this research is on the deployment of RFID systems in the supply chain. A supply chain poses a challenging environment for the deployment of passive RFID tags. Firstly, we are dealing with thousands to millions of RFID tags in a relatively small area, such as a warehouse. In other words, we have a dense RFID tag environment. Intensive research needs to be carried out to ensure good RFID coverage, reliable RFID systems, minimising interference between tags and readers, while optimising the frequency spectrum in accordance to local regulations. Secondly, in a conventional passive RFID system, the reader antenna provides power to the passive RFID tags before any communication link between a reader and a tag can be established. The quality of this communication link depends heavily on the environment. In the worst case scenario, a hostile environment can render an otherwise functioning RFID system useless. One very interesting point is that ionised liquid products are very common along supply chain. Ionised liquid, such as water, can absorb microwave power easily and hence a UHF RFID system will not perform well in a ionised liquid filled environment. This prompts for an investigating to design a dual frequency RFID tag antenna, which can operates in UHF in normal operation and also in HF when ionised liquid is present. 1.4 Contribution There are two main areas of contribution offered by this thesis with respect to the two motivations as discussed in Section 1.3. The first area of contribution is on the dense reader deployment. This thesis presents a solid and practical way for reader antenna positioning analysis. The research started off with the study of simple path loss model. This simple model was then modified to take into consideration of environmental factors to cater for different deployment situations. With a comprehensive model, simulation code was written in MATLAB Page 4

21 Chapter 1 Introduction to predict the effect of RFID readers on each other. This offers an easy way to examine the interference between readers. Though approximate, as the actual environment is dynamic in nature, this allows examination of potential interference before actual implementation is carried out. Also, this thesis investigates some novel techniques in synchronising the operation of RFID readers in a certain vicinity to optimise the performance of an RFID system when operating under strict local regulations. The local regulation of interest is the European Listen Before Talk provision. The author believes that Listen Before Talk is not beneficial for the wide spread adoption of RFID system. Intensive studies carried out proved that the threshold values as specified in the Listen Before Talk provision are too low to allow wide spread adoption of RFID to be feasible. It is hoped that the results presented here will strengthen the case for the relaxation of the regulation imposed upon RFID systems in Europe. Though not entirely related to those issues discussed above, this thesis extends its reader positioning study into the implementation of a new technique, named Second Carrier Sensing. This technique is to prevent the RFID tag from being confused by multiple interrogation signals from different reader antennas which exist in a same area. The second area of contribution started off with the art of RFID tag antenna design. This thesis presents the crafting of a novel RFID tag antenna from the very basic HF and UHF tag antenna. The design specifications of the tag antennas in this thesis are mostly industry driven, where various actual implementations are possible (some were already implemented in the course of this research). Design decisions are clearly presented to offer a reference for those interested in tag design for other applications. The research on tag antenna design is then furthered into the design of a dual frequency RFID tag antenna. This dual frequency antenna supports both the HF band and UHF band operation and has a frequency ratio of more than 70. This is the first dual frequency antenna that supports such a high frequency ratio while maintaining a single feed structure. With this dual frequency antenna, an RFID tag can support two very different frequency bands with just a single chip, and is especially important as Page 5

22 1.5 Thesis Structure HF will offer a better read range in a environment with lots of liquid products. Indirectly, this will contribute to the effort in spreading and encouraging the usage of RFID in the supply chain. Also documented in this thesis are the techniques developed over the the entire research period for measuring the performance of newly design RFID tag antennas. This will provide a compact guide for anyone who is interested in the area of RFID tag design. 1.5 Thesis Structure This thesis is structured into 9 main chapters, excluding conclusion, appendix and bibliography, and is as shown in Fig Chapter 2 RFID System Chapter 3 RFID Positioning Analysis Chapter 6 RFID Antenna Design Chapter 4 Reader Synchronization Chapter 5 Operational Considerations Chapter 7 Dual Frequency RFID Antenna Chapter 9 Dual Frequency Protocol Chapter 8 Antenna Measurement Figure 1.2. The complete structure of thesis These 9 chapters can be categorised into 3 major area: 1. Introduction Page 6

23 Chapter 1 Introduction Chapter 1 of this thesis (which is this chapter) explains the title of this thesis and outlines the background of this research. Also highlighted is the motivation of this research and its contribution to the engineering community. Chapter 2 presents a comprehensive study on RFID system, which includes a brief history of RFID, presentation of simple RFID system and variants of RFID systems. Also discussed are the deployment of RFID in the supply chain, RFID standards and the future of RFID. 2. Antenna Positioning Analysis Chapter 3 is on the RFID reader antenna positioning analysis. Simple path loss models are discussed and applied in the prediction of the strength of RFID reader interrogating signals. Furthermore, a simulation coded in MAT- LAB was developed. Chapter 4 presents the idea of RFID reader synchronisation, where RFID readers are synchronised to share the limited bandwidth allowed under certain local regulations. A variation of the synchronisation method is introduced to cater for different needs in real life implementation. Chapter 5 investigates the operational consideration for RFID deployment and simulation. It extends the research presented in chapter 3 and 4 and offers guidelines to avoid pitfalls in setting up simulation software for RFID reader deployment. Also, the idea of Second Carrier Sensing is presented with its limits discussed in relation to the positions of reader antennas. 3. Dual Frequency RFID Tag Antenna Design and Analysis Chapter 6 is a large chapter, and mainly deals with the design of RFID tag antennas. In the beginning of the chapter, the design constraints of small RFID tag antennas are discussed. The latter parts can be separated into two major parts, the design of HF RFID tag antennas and the design of UHF RFID tag antennas. Page 7

24 1.6 Chapter Structure In the design of HF RFID tag antennas, the theory of simple HF loop antenna (magnetic dipole) is reviewed, followed the practical design, simulation, fabrication and testing of HF RFID tags for various applications which include wine cork tagging and pig ear tagging. Similar with above, in the design of UHF RFID tag antenna, the theory of the simple electric dipole and the simple planar dipole is reviewed, followed by the practical design, simulation, fabrication and testing of UHF RFID tags for various applications which include sheep ear tagging and beer keg tagging. Chapter 7 focuses on the design and analysis of dual frequency RFID tag antenna using different methodologies. Firstly, a dual frequency RFID tag antenna is designed by merging a HF antenna and a UHF antenna. The second method is to have a HF antenna acting as a UHF antenna at the UHF band. In either method, theories for antenna design were reviewed before actual design, fabrication, and actual tag testing was carried out. Chapter 8 discusses the measurement technique used in carrying out all the measurements to obtain the characteristics of the RFID tag antenna. Focus was on the technique of impedance and radiation pattern measurement. Chapter 9 looks into the necessity of dual frequency protocol to enable the operation of a dual frequency RFID tag. 1.6 Chapter Structure A chapter begins with a brief introduction, followed by background study on materials related to the research and discussion of that chapter. Chapter 2 will be referred to if any basic RFID background is involved. Most of the important results in any chapter have been published as conference papers. Any publication used in a chapter will be mentioned at the end of the introduction section of that chapter or the beginning of a section. Page 8

25 Chapter 2 RFID System T HIS chapter presents a detailed background study of an RFID system. This detailed study provides a strong foundation on the research carried out by the author and will be referred to throughout this thesis. Page 9

26 2.1 Introduction 2.1 Introduction Radio Frequency Identification (RFID) is a technology which involves several distinct areas of expertise and study. As explained in Chapter 1, the focus of research presented in this thesis is on the RFID antenna positioning analysis in a dense reader deployment zone, and the antenna design for a dual frequency antenna. Nonetheless, for completeness, a detailed background of RFID is presented in this chapter, together with the operational principles of RFID, and will be referred throughout this thesis. Papers on the networking aspect of RFID have been co-authored [6, 7], and the key points of that paper are duplicated in this chapter. 2.2 History of RFID RFID, in fact, is not a new technology and has a long history. It is believed that the birth of RFID can be traced back to the early 20th century, when Radio Detection And Range (Radar) was invented, where the reflected wave is used to detect an object. However, it was then discovered that detecting an object is not sufficient in some situation, especially during the war time, when an identification is also required. Hence, the first RFID deployment was carried out as a long-range transponder systems of identification, friend, or foe (IFF) for aircraft, so that a friendly aircraft would not be misidentified as the enemy [8, 5]. The first patent on RFID is by Harris in 1960, titled Radio Transmission Systems With Modulatable Passive Responder. In this patent, Harris presented several ways of communication between a reader and a passive responder (tag). [9]. Early journal publications on RFID include: Small resonant scatters in field measurement by Harrington [10] in 1962, where a general formulation for a backscattered field from loaded objects is given. Electromagnetic scattering by antennas by Harrington [11] in 1963, where he relates the theory of loaded scatterer to antenna theory. Page 10

27 Chapter 2 RFID System Identification using modulated RF backscatter by Koelle, Depp and Freyman [12] in 1975, where a successful demonstration of identification using RFID backscattering is shown. However, because of cost and size issues, RFID had limited usage and coverage in 70 s and 80 s. It was deployed in certain toll collection and object tracking applications. With the advancement in semiconductor and miniaturisation in 90 s, the cost and size of the passive transponders reduced to a very acceptable level and at the end of 20th century, it was started to be used in the supply chain. 2.3 A simple RFID System RFID is a technique used to identify objects by means of electromagnetic waves. An object can be tagged with an electronic code responding label. An electronic tag consists of an antenna and an integrated circuit. Upon receiving any valid interrogating signal from any interrogating source, such as a reader, the tag will respond according to its designed protocol. The relationship between a tag and a reader is illustrated as Figure 2.1. Antenna Reader Tag Tag Tag Tag Forward Link Return Link Figure 2.1. RFID system: Reader and tag relationship. One or more antennae can be connected to a reader, and are used to communicate with tags within field of interrogation. Signals sent from reader through an antenna to a tag are called forward link signals while the signals from a tag back to reader are called return link signals. A detailed description of RFID system can also be found in RFID Handbook by Finkenzeller [13]. Page 11

28 2.4 Variants of RFID 2.4 Variants of RFID Active and Passive RFID Tag There are two types of tags, active and passive. Both active and passive RFID tags are defined in [14]. An active tag is defined as an RFID device having the ability of producing a radio signal. Normally an active tag has its own battery source; it has a greater read range compared to a passive tag but limited by the life time of its battery. A passive tag is defined as an RFID device which reflects and modulates a carrier signal received from an interrogator. A passive tag is normally energised by an interrogating signal from a reader antenna and does not have any other internal power source. It typically has shorter read range as compared to an active tag. There are several good RFID passive tag chip designs available in the literature. For example, a low power high performance UHF passive tag chip was presented by Karthaus [15], which requires only 16.7 µw to power up the tag chip and has a read range of around 10 m. There is also a low cost UHF passive tag design offered by Glidden [16]. Tag chips are mentioned here because the tag chip input impedance is essential in antenna design Backscattering Concept of backscattering (to be added) Common Frequency Bands for RFID Operations There are 4 popular bands in RFID, which are the LF band (less than 135 khz), the HF band (13.56 MHz), the UHF band ( MHz) and the microwave (2.45 GHz) [17]. Each band has positive and negative characteristics. The Auto-ID Center has published protocols for HF and UHF RFID operations. LF has been left out, as under the vision of Auto-ID Center, each tag has to bear an EPC Page 12

29 Chapter 2 RFID System number, and must be at least 64-bit in length [18], and for low frequency operation, such as in the LF band, very slow reading in a heavily populated tag environment would result. Also, LF RFID tags require large antenna components and hence difficult to implement and are susceptible to electrical noise, which HF can handle [19]. Microwave tags can offer comparatively very fast reading, but the performance of microwave will suffer, worse than any other bands described above, in the presence of liquid or metal [17]. It is very difficult to conclude whether HF or UHF is better for RFID application in supply chains. This is because in any supply chain, there are lots of different scenarios, and in some HF is better and in the others, UHF outperforms HF operations. The strength of UHF lies in the fact that it offers 4-5 m average read range (up to 10 m in state of art RFID systems) and is cheaper to produce in large volume [17]. UHF tags run at a higher frequency and hence has a higher read rate [20]. In other words, in an UHF system, more data can be transferred within a fixed period of time, and hence UHF is often deployed in a situation where more on board data is required [21]. However UHF RFID operation has difficulties in environments that are packed with liquid or contain lots of metal, or even will not work properly in livestock industries such as in the area of feedlots or slaughterhouse [22]. The is caused by the nature of the radio frequency spectrum absorption by water as shown by [23], in which UHF is highly absorbable. On the other hand, HF has shorter read range (around 1 m) and has a poorer tag discrimination as compared to UHF [21], though it performance is still acceptable with a read rate of more than 200 tags per second. However, HF can penetrate water better than UHF. According to the radio frequency spectrum absorption by water shown by [23], the absorption of HF by water is insignificant as compared to UHF. Ref. [20] supports the claim of HF can penetrate water better because of its longer wave length as compared to UHF, though HF also suffers in the presence of metal just like UHF. Schurmann also reported that lower frequency HF systems are primarily used in the case where better penetration of objects is needed as quoted by [13]. Page 13

30 2.5 RFID and Supply Chain As a general consensus, both HF and UHF RFID systems are applicable in supply chains, depending on the RFID application. 2.5 RFID and Supply Chain As mentioned before, the deployment of RFID systems in supply chains in large numbers, with the potential to revolutionise supply chains, has sparked public interest in RFID technology. A lot of case studies have been carried out by market analysts, which includes [24, 25, 26, 27, 17, 28, 29, 21, 30]. Most of these articles discuss the impact of RFID on the current supply chain, and how a company can maximise the benefits of an RFID system. The formation of the Auto-ID Center in 1999 aimed to develop an open standard architecture for creating a seamless global network of physical objects [31]. The ultimate vision was to create an Internet of things, and the concepts were well explained by Cole [24]. With the standardisation of HF protocol (ISO standardisation for MHz: ISO/IEC [32] and ISO/IEC [33]) and UHF protocol (EPCglobal standardisation: UHF Class 1 Generation 2 [34] and EPC tag data standard [18]), major parties in supply chains are pushing for RFID adoption. For example, one of the biggest retail stores in the USA, Walmart, mandated the use of RFID in its supply chains by its top 100 suppliers by Jan 2005 [35]. Also, the USA Department of Defence mandated the use of passive RFID in its supply chains by Jan 2005 [36]. According to a Georgia Institute of Technology business research report, supply chain problems cost companies between 9 and 20 percent of their value over a six-month period [26]. This shows that the operation of the supply chains cannot afford to be compromised, and any operational mistake would be costly. Currently, inventory checking and tracing in a supply chain is done by reading barcodes. A barcode, explained in [13], is a binary code comprising a field of bars and gaps arranged in a parallel configuration. It is read by optical laser scanning, i.e. by the different reflection of a laser beam from the black bars and white gaps. In other words, Page 14

31 Chapter 2 RFID System barcodes scanning requires manual handling and orientation of each object and also requires line-of-sight. A thorough comparison between barcodes and RFID was offered by [13] and supported by various other independent sources [17, 20, 37, 28]. They claimed that, since RFID does not require line-of-sight, scanning is faster, more effective and hence saves cost. Furthermore, RFID identifies a unique product item, and enhances Warehouse Management System (WMS) in dealing with stock replenishment, over-stocking problems and other supply chains related problems such as counterfeiting. For example, in an international pharmaceutical supply chain, seven percent of drugs may be counterfeit. Not only RFID can assure a safer supply of drugs to the consumers, RFID can also save up to 8 billion dollars by 2006 [20]. This is something that cannot be achieved by barcodes. The USA Food and Drug Administration (FDA) trusts RFID as a tool to help authenticate drugs, and has taken initiative to set up guidelines in using a RFID system in their pharmaceutical supply chain [38]. Also, business cases were presented to support RFID deployment and it is believed that RFID will offer near perfect inventory information visibility in supply chains [25]. An RFID tag costs as low as 5 cents have been projected [39]. This is also the required cost of a tag, so that every object in supply chains can be tagged with RFID tag. Although this is not the case currently, it is believed that the cost of a RFID tag can be reduced to that level when the RFID systems are deployed in large numbers and across multiple applications so that RFID tag production volume is increased to billions of tags. One of the biggest challenges of RFID large scale deployment is the consumer privacy issue. Though this is not included in this research, it is mentioned here to show the author s belief in RFID policies to avoid any individual privacy invasion and is the subject of research by others in the group. There are many reports on consumers fears of RFID deployment in supply chains [21]. Since RFID scanning is a type of automatic scanning [28], or in other words, a person holding an item will not know when that item is being scanned, consumers may fear that they are being tracked unlawfully. Page 15

32 2.6 EPC Network The current RFID technology strives to address this issue. The measures taken include public education on RFID based on the facts that RFID does not have long range scanning capability, and the option that a tag can be killed permanently whenever an item is purchased under endorsed RFID protocols [34]. 2.6 EPC Network The Electronic Product Code (EPC) Network, originally developed by the Auto-ID Center with its standards now managed by EPCglobal Inc., was designed and implemented to enable all objects in the world to be linked via the Internet. EPC Network has undergone tremendous changes since its first been introduced. At the beginning of this section, the early stage EPC Network will be introduced, followed by the current state of an EPC Network. A joint research paper on the EPC Network when it was first introduced had been published [6, 7], which outlines components of EPC Network and their functionality Early stage EPC Network Figure 2.2. Early stage EPC Network. Fig. 2.2 shows the structure of a typical EPC Network in the early stage. The EPC Network consists of three major components, which are the Savant, the EPC Information Service (EPCIS), and the Object Name Service (ONS). Strictly speaking, the reader Page 16

33 Chapter 2 RFID System should be considered as part of the EPC Network. However, the reader is considered to be a pure RFID tag interrogator under the control of an Savant, though implementations of some readers will integrate at least the base functionality of a Savant into the reader itself. Savant The Savant is a middleware system located between a reader (or multiple readers) and the applications in RFID systems. Applications operate on top of, or within, the Savant operating environment. The Savant passes requests from the application(s) to the reader(s) and receives unique tag identifiers and possibly other data from sensors, and passes that information to the application(s). The Savant has several fundamental functions integrated into its design, some of which are data filtering, aggregation and counting of tag data. These fundamental functions are required in order to handle the extreme large quantity of data that RFID systems can generate through the continuous interrogation of tags. Object Name Service The function of the Object Name Service (ONS) in an EPC Network is to identify the location of the server hosting the appropriate information needed by an application. In other words, the ONS acts like a reverse phone directory as the ONS uses a number (EPC number) to retrieve the location (of data) from its database. To encourage rapid development of the ONS, the ONS is purely based on existing Internet technology and infrastructure. The first generation of ONS system designs were based upon DNS systems with the first implementations utilising existing DNS implementations with customised configurations. Physical Markup Language The Physical Markup Language (PML) defines the way information is transferred in the EPC Network system. PML Core is based on the existing standard, XML Schema Language. It uses tags to format the data before the data is sent. PML does not use short tags for data formatting. Although more bandwidth will be required for data transfer, compared to using short tags, the use of longer Page 17

34 2.6 EPC Network descriptive tags increases human readability and hopefully will avoid mistakes in interpreting and understanding the data, and how that data is to be handled. EPC Information Service EPC Information Service (EPCIS) is the gateway between any requester of information and the database. It receives and sends messages in PML with any requester of information, although it s communication with the database can be in any format or standard EPC Network at Current State The new EPC Network is relatively different from the original version; it is more defined and more well structured. The explanation of the EPC Network at the current state in this section is based on the EPCglobal Architecture Framework published in 2005[]. A whole picture of an EPC Network is as shown in Fig Basically, there are two distinct types of components in the new EPC Network: software/hardware role, and interface. They are self explanatory. Software/hardware role components are either a piece of software or hardware which handle tasks allocated to it. Interface governs how software/hardware role components communicate to each other. All the role components and interface components shown in Fig. 2.3 are explained in detail in [40]. Some important parts are briefly highlighted to compare with the original version of an EPC Network: Application Level Events Ref. [41] is the complete specification of Application Level Events (ALE). It is an interface component which handle the data flow of the filtering and collection on the inventory of RFID tags. EPC Information Services Page 18

35 Chapter 2 RFID System Figure 2.3. Current EPC Network [40] c 2006 EPCglobal. Page 19

36 2.7 RFID Standards Ref. [40] is the complete specification of EPC Information Services (EPCIS). There are several interface and role components bearing the name of EPCIS in Fig Basically any component related to the EPCIS contribute to the data accessing, sharing and storing between all the authorised RFID network. By linking RFID networks together, a global size network with real-time data sharing, also known as EPC Network, is made possible. Object Name Service Ref. [42] is the complete specification of Object Name Service (ONS) as an interface component. An ONS interface is used to locate the reference to EPCIS service or any other related other services. In Fig. 2.3, there are Local ONS and ONS Root. They are related to but not the same as an ONS interface. Reader Management Ref. [43] is the complete specification of Reader Management. As suggested by its name, Reader Management is a role component which manages the operation of RFID readers, including reader configuration and firmware update. Reader Protocol Ref. [44] is the complete specification of Reader Protocol. Reader protocol is an interface component. Basically, reader protocol allows data from RFID readers to be passed up the EPCglobal Network stack, for example the ALE. The basic idea of both the EPC Networks (early stage and current stage) is the same to enable real-time global data sharing for product tracking. The current EPC Network is an updated version of the early stage EPC Network, retaining all the operating principles while offering better clarity. The EPC Network is still evolving, depending on the needs of RFID adopters. 2.7 RFID Standards To enable and encourage wide spread adoption of RFID systems, one of the most important factors is to have RFID systems standardisation. Not only does this simplify Page 20

37 Chapter 2 RFID System the hardware and software design for RFID systems, it also increases user confidence in RFID. Basically, there are two different areas of standardisation. The first one is to regulate the spectrum usage and is always governed by local regulatory authorities. The common standards are: Title 47, telecommunication, chapter 1, Part 15, radio frequency devices published by Federal Communications Commission, in 2001 [45]. This regulations is mainly adopted by the United States of America. EN v1.1.1: electromagnetic compatibility and radio spectrum matters (ERM); radio frequency identification equipment operating in the band 865 MHz to 868 MHz with power levels up to 2W published by European Telecommunications Standards Institute (ETSI) for European Standard (Telecommunications Series) in 2004 [46]. This regulations is mainly adopted by the countries in the European Union. AS/NZS 4771:2000 technical characteristics and test conditions for data transmission equipment operating in the 900 MHz, 2.4 GHz and 5.8 GHz bands and using spread spectrum modulation techniques, incorporating Amendment No. 1 published by Standards Australia. The second type of standardisation is to have a uniform air interface and command sets between an RFID reader and an RFID tag so that an RFID reader produced by a company can be integrated into an RFID network setup by another company. There are two bodies responsible in providing standards, ISO (International Organization of Standardization) [47] and EPCglobal [48]. The following standards are some of the standards available from ISO: For LF RFID operation: ISO 11784:1996: Radio-frequency identification of animals - code structure [49] and ISO 11785:1996: Radio-frequency identification of animals - technical concept [50]. Page 21

38 2.7 RFID Standards For HF RFID operation: ISO/IEC :2004: information technology - radio frequency identification for item management - Part 3: parameters for air interface communications at MHz [32]. For UHF RFID operation: ISO/IEC :2004: information technology - radio frequency identification for item management - Part 6: parameters for air interface communications at 860 MHz to 960 MHz [51]. For RFID terminolgy: ISO/IEC , information technology - automatic identification and data capture (AIDC) techniques - harmonized vocabulary - part 3: radio frequency identification (RFID) [14]. The following standards are some of the standards available from EPCglobal: For the RFID operation in HF band, the standard is documented in MHz ISM band class 1 radio frequency identification tag interface specification: candidate recommendation, version [52]. For the RFID operation in UHF band, the standard is documented in EPC radiofrequency identity protocols class-1 generation-2 UHF RFID protocol for communications at 860 MHz MHz version [34]. As every tag contains a unique number and this unique number should be retrievable by any reader which conforms the the above stated standards, the tag data contained by any tag must be standardized too. The standards is EPC tag data standards version 1.1 rev [18]. Page 22

39 Chapter 3 Path Loss and Position Analysis A simple path loss model for RFID systems is used to demonstrate the reader collision problem in a dense reader environment. Based on the path loss model, a simulation script is written in MATLAB to visualise the power level in various positions when multiple readers are deployed in a same vicinity. With this simulation software, antenna positioning analysis can be carried out efficiently. Page 23

40 3.1 Introduction 3.1 Introduction In a dense reader environment, there will be multiple RFID readers and cross-reader interference will reduce the reliability and efficiency of the RFID system. RFID readers can use different channels to minimise collision. However, with limited channels, channel collision will occur in dense reader environments, where several readers try to interrogate tags at the same time in the same vicinity. The results can be unsatisfactory with respect to read times and an unacceptable level of misreads may result. In the worst case, the interference could paralyse an entire RFID system. Furthermore, the RFID signal interference problem is not only limited to RFID devices disrupting the operation of other RFID devices operating in close proximity, but also includes RFID devices disrupting the operation of other RF devices in nearby frequency bands. Regulatory bodies have set some strict restrictions on RFID radiation to minimise the RFID interference problem. The concept of Listen Before Talk (LBT) has been included in the European regulation, ETSI EN [46], which places severe restrictions on RFID deployment and causes some uncertainties over the feasibility of RFID global deployment. Hence, the objectives of this research are to report on the analysis of the reader collision problem and provide a solution to RFID deployment regarding the reader collision and the LBT problems for the benefit of those eager to set up RFID systems. Also, this research investigates the art of antenna positioning to optimise RFID field performance, while adhering to strict regulations. Intensive simulations were carried out to visualise RFID reader signal strength at various distances with respect to the transmitting antenna, either operating in a same channel or in any neighbouring channels. It is hoped that this research can provide sufficient and useful guidelines on safe distance between antennas in a dense reader environment or even in a Listen Before Talk regulated area. It is also the belief of the author that by careful planning of antenna positioning, a RFID system can be optimised and its interference to other RF systems in the surrounding area can be minimised. Section 3.2 explains the theory of the path loss model, and shows how an in-building path loss model is related to a free space model. Page 24

41 Chapter 3 Path Loss and Position Analysis Section 3.3 explains the path loss experiments carried out around the laboratory, along with results and discussions. Section 3.4 forms a path loss model for RFID based on the results in Section 3.3, and this model is used to investigate a two-reader interference in Section 3.5. Also, the safe distance between two RFID reader antennas, before collisions occur in a same channel, was computed and discussed. Section 3.6 discusses the theory of power density, antenna gain pattern of RFID antenna, frequency channelling and the concept of Listen Before Talk. This section is mostly informative but provides a strong background study for the RFID positioning analysis. Section 3.7 explains the concepts behind the simulation software developed in MAT- LAB. Section 3.8 includes both the simulation results and discussions on various antenna positioning. Some real life applications including the potential of reader synchronisation in handling problems that may arise in a dense reader environment are presented in Section Path Loss Model Free Space Path Loss For a pair of lossless antennas in free space with optimum orientation we may write the power transfer ratio in the form ( ) P r λ 2 = g t g r (3.1) P t 4πd where λ = wavelength; P t = transmitted power; P r = available received power; g t = transmitter antenna gain; g r = receiver antenna gain; d = separation distance between antennas. Page 25

42 3.2 Path Loss Model For some purposes it is desirable to separate the effects of antenna gain and distance between antennas, and give the name free space path loss to the factor in the above equation. By expressing this factor in db we have the free space path loss expression: ( ) 4πd PL(dB) = 20 log 10 λ (3.2) Using the frequency of f = 915 MHz, λ = c f = 0.33 m, for a separation d = 1 m, PL(dB) = db In-building Path Loss An in-building path loss is a path loss that occurs in a physical building, and it is not the same as a free space path loss, as an in-building path loss will normally take in consideration of path obstruction, reflection, absorption and other attenuation effects introduced by the presence of objects inside a building. Free space path loss has a simple d 2 variation but in-building path loss has a more complex structure. The inbuilding path loss model chosen for the purpose of simulation is given in (3.3), and is explained by Rappaport [53]: PL(dB) = PL(d 0 ) + 10 n log 10 ( d d 0 ) (3.3) where d 0 = arbitrary reference distance; n = a value that depends on the surroundings and building types; and d = the separation distance between two antennas. d 0 will be chosen as 1 m and PL(d 0 ), which is the in-building path loss at 1 m away, will be approximately the same as the free space path loss at 1 m away, as room reflections are not huge at this small distance. Hence, from (3.2), PL(1 m) is approximately 32 db, and (3.3) becomes: PL(dB) = n log 10 ( d d 0 ) (3.4) As explained in (3.3), n is a empirical value that depends on the surroundings and building types, and is only obtainable through experiment. An environment with high Page 26

43 Chapter 3 Path Loss and Position Analysis n is a hostile environment for radiation, and its in-building path loss will be higher when compared to the case in low n environment. 3.3 Experiment An interrogating RFID antenna was set to transmit a query signal while a measuring spectrum analyser was moved away from the transmitting antenna. The strength of the received signal was recorded versus the distance away from the transmitting antenna. Removing antenna gain from the measured values gives us the values of path loss. The transmitting and receiving antenna used in this experiment both have a gain of 6 dbi. Fig. 3.1 is plotted using logarithmic scale, and we have a straight line approximation of: PL(dB) = log 10 ( d 1 ) (3.5) for d > 8 m. The results resemble the model based on (3.6) and also strengthen the belief that reader collision must be solved for large scale RFID deployment. Figure 3.1. Results from experiment Page 27

44 3.4 Path Loss Model for RFID Path Loss (db) n=6 n=5 n=4 n=3 n=2 n= Transmitter-Receiver Separation (m) Figure 3.2. Plot of In-building Path Loss Model Against Distance 3.4 Path Loss Model for RFID For the theoretical calculation in this chapter, we chose n = n 2 = 3.5, when the separation is more than or equal to 8 m, and we chose n = n 1 = 2.5 when the distance is less than 8 m. i.e. we are using approximation equations: PL(dB) = PL(d 0 ) + 10 n 1 log 10 d d 0 0 < d 8 m (3.6) PL(d 0 ) + 10 n 2 log 10 d d 0 d 8 m For the purpose of simulation, the in-building path loss model from (3.4) was modified slightly and is given in (3.6) and is plotted as shown in Fig Simple Two Reader Interference In this section, a scaled down version of reader collision is discussed, where the collision involves only two readers, and the readers are assumed to transmit and receive in the same channel of a multi-channel frequency band. Consider the case where there are two readers, A and B, using a same channel, channel C. It is assumed that Reader Page 28

45 Chapter 3 Path Loss and Position Analysis A and Reader B are identical, the antennas for both of them are the same and have the same gain. Also, both of the antennas are facing each other. Reader A uses channel C to interrogate a tag and the tag will have in-band backscattering to response to Reader A. If we have a transmitted power of 0 dbw and an antenna gain of 6 dbi, we will have a total of 6 dbw EIRP. Fig. 3.3 shows diagrammatically how the interrogation between Reader A and the tag occurs. The paths 1 and 2 are the signal paths. At 1 m away, the path loss as obtained using (3.6) is 32 db. Hence the total path loss is approximately 64 db (2 x 32 db). Antenna Reader A Tag Antenna Reader B Figure 3.3. A simple illustration of reader collision where reply from a tag will be interfered by signal sent from another nearby reader. In the example illustrated, tag reply 2 is interfered by interrogation signal 3 from Reader B. The tag antenna has a gain of approximately 1.5. However, we have to take in consideration a probable tag misalignment, i.e. the tag antenna is not positioned in the most optimal way. In our case, we assume the tag has a unity antenna gain. If we further assume that the efficiency of the tag is 10%, the signal will suffer another 10 db loss. All the losses (path loss + tag efficiency-tag antenna gain) summed up to be 74 db. Since Reader B is also using the same channel, channel C, the interrogation signal sent by Reader B will interfere with the in-channel backscattered signal from the tag. The question is how near Reader B needs to be to interfere with the backscattered signal. The comparison is made between (a) path loss of path 1 and path 2, tag antenna gain, and tag efficiency loss, and (b)path loss of path 3. This comparison is only applicable in the situation as described in this section. As a common term used throughout this Page 29

46 3.6 RFID Positioning Analysis Table 3.1. The effect of tag distance on multi-reader interference Distance of Tag Total Loss (db) Minimum Dis- Distance in Free (m) tance for B to Space to give interfere (m) the same loss (m) , , ,460.5 section, the losses in both (a) and (b) are called Total Loss. If we look at path 3 as shown in Fig. 3.3 and using (3.6) to calculate path loss, a distance of around 28.7 m is needed to have path loss, or total loss in this case, of 74 db. Table 3.1 shows some result on the minimum distance for Reader B to interfere with the tag reply. Again, all these results are computed using (3.6), and (3.6) takes into consideration in-building propagation loss. It is very natural to also raise the question of what will be the case, if we only consider free space propagation loss. The results obtained using free space propagation loss model are attached as column 4 in Table 3.1. We continue to make the assumption that free space path loss is not applicable in our case, and that at the distances emerging from these calculations, at least some obstacles will be present, and a within-building propagation loss model is appropriate. It is discovered that, if a tag is located 10 m away from an interrogating reader, the antenna of the other readers must be around 2,865 m away. Since a state of art reader in the market can have a read range of around 10 m when reading a passive tag, to put the next reader more than 1 km away is not sensible. Section 3.9 in the latter part of this chapter provides ideas to solve this problem. 3.6 RFID Positioning Analysis Page 30

47 Chapter 3 Path Loss and Position Analysis Power Density Power density, S, which at a distance d for a transmitting antenna in the direction of maximum gain, has the value: S = ( ) EIRP 4πr 2 (3.7) where EIRP (Equivalent Isotropic Radiated Power) is the product of transmitted power and antenna gain. Thus, EIRP = g t P t (3.8) ERP (Effective Radiated Power) is a more popular term in Europe and is the product of transmitted power and antenna gain with respect to the gain of a dipole. A gain of a dipole is approximately 1.64 times of the gain of an isotropic radiator and hence, Thus, ERP = g t 1.64 P t (3.9) EIRP = 1.64 ERP (3.10) Antenna Gain Pattern As shown by (3.1), antenna gain must be taken into consideration, apart from path loss, to relate transmitted power to received power. A typical RFID antenna is a directional circularly polarised antenna with a gain of 6 dbi. The polar plot of the gain of a typical RFID antenna is shown in Fig. 3.4, and has been used to simulate and analyse the antenna positioning in dense reader environment presented in this paper. Page 31

48 3.6 RFID Positioning Analysis dbi Figure 3.4. Polar Plot of the Antenna Gain of a Directional Circularly Polarised RFID Antenna with a Gain of 6 dbi Frequency Spectrum Channeling Frequency spectrum channelling is a popular technique used in reducing reader collision problem. One of the biggest challenges in frequency spectrum channelling is the sideband interference, where the sideband of a particular channel interferes with the transmission in a neighbouring channel. Hence, under the latest UHF Class 1 Generation 2 protocol [34], the sidebands of a transmission must adhere to the limits as shown in Fig The dbch is defined as decibels referenced to the integrated power in the reference channel. The simulation software developed also uses this limit to investigate interactions of readers operating using different frequency channels Listen Before Talk In the European Regulation as outlined in ETSI EN V1.1.1 ( ) [46], a reader must listen and confirm that a particular channel is not occupied before it can use that particular channel to interrogate any tag. The transmit power and the corresponding threshold values are extracted from the above-mentioned ETSI document and integrated into Table 3.2. Page 32

49 Chapter 3 Path Loss and Position Analysis Figure 3.5. Transmit Mask for Dense-Interrogator Environments [46]. Table 3.2. Transmit Power and Corresponding Threshold Values ERP (W) ERP (dbw) Threshold (dbw) Path Loss (db) Distance (m) Up to 0.1 Up to to to to to Similar to the calculation done in Table 3.1 in Section 3.5, the Distance column in Table 3.2 is computed using (3.5), where we consider an in-building propagation model with n value set to 3.5. As mentioned before, the value of n changes from building to building. However, the distance for different values of n still can be obtained from Fig The main point here is if we are going to deploy readers on a large scale, most likely the system will not work in optimal operation mode. This is due to the fact that LBT will effectively shut down many of the channels, though those channels might have been, in the absence of the LBT provision, freely available for interrogation between readers and tags. This problem is not exactly a reader collision problem but it is also covered in this paper because the LBT problem will be very serious in a dense reader environment, or a place where reader collision is a serious issue. Page 33

50 3.7 Simulation Concept Around 52 dbw Figure 3.6. Results From Simulation (Vertical and Horizontal Axes in m, Received Power in dbw) for Simulation Concepts Example. 3.7 Simulation Concept The simulation script used is written in MATLAB code. The resolution setting in the simulation script defines the distance between each simulated point. The resolution of the simulation results on antenna signal strength can be adjusted, but by default the resolution is set to be 1 m 1 m. Another common resolution used is 10 m 10 m. All the antennas used in the simulation are directional circularly polarised antenna, with antenna gain pattern shown in Fig. 3.4, and are located 1 m above ground level (Refer Fig. 3.9 for visualisation of default antenna configuration). The software uses the path loss model as shown in (3.6). By specifying a transmit power, (3.1) is used to compute the received power with the gain specified in Fig. 3.4, and path loss from (3.6). The simulation result is in graphical form and is able to show the received power at any location, bounded by a limit set by the user, when a receiving antenna of 0 db gain is used, and is always from the top view at 1 m above the ground. For example, in the default configuration of the simulation software, 0 dbw of transmit power is fed through a 6 dbi gain antenna, resulting a 4 W EIRP radiation. The 6 dbi gain is only valid in the front direction as shown in Fig Hence, at a distance of 10 m away directly in front of the antenna, the path loss will be 58 db, as computed from Page 34

51 Chapter 3 Path Loss and Position Analysis (3.6). The received power at that location as detected by a 0 db gain antenna is hence 0 dbw (transmit power) + 6 dbi (transmitter antenna gain) -58 db (path loss) + 0 dbi (receiver antenna gain), giving a measured received power of -52 dbw. The simulation result is shown in Fig A horizontal guideline is added to show the point located 10 m away from the antenna. At that point, the received power is simulated to be approximately -52 dbw, which agrees with the calculated result. 3.8 Results and Discussion The following simulations are completed following the concepts outlined in Section 3.7. The path loss model used is shown in (3.6), with n = 2.5 for near distance and n = 3.5 for far distance One Antenna Simulation The one antenna simulation described here is similar to the example given in Section 3.7. The only difference is the simulation here covers a larger area to identify available zones to deploy a second antenna working in a same channel. From Table 3.2, the maximum transmit power allowed by current European regulations is 2 W ERP, or equivalently 3.2 W EIRP. From Table 3.2, the minimum threshold is -126 dbw. The simulation result is shown in Fig. 3.7, and shows that for the next antenna to be able to operate in the same channel with the initial antenna under consideration, both antennas must be separated by a distance of around 350 m away in the horizontal direction. If the second antenna is to be placed in a back to back position with the first antenna, the minimum distance between them would be shorter, which is approximate slightly further than 200 m Two Antennas Simulation From the one antenna simulation, it has been shown that another antenna operating in a same channel can be deployed around 350 m horizontally away from the first Page 35

52 3.8 Results and Discussion Figure 3.7. Results From Simulation (Vertical and Horizontal Axes in m, Received Power in dbw) antenna. In this two antennas simulation, a second antenna is deployed 350 m horizontally away from the first one, and the result is as shown in Fig The bold line is the approximate safe boundary of the deployment of the third reader operating in a same channel. The distance is approaching 1.4 km in front of both antennas and around 400 m horizontally away from the second antenna. This shows that for antennas transmitting in a same frequency, they have to be at a great distance apart and hence is not practical in real life applications One antenna on the ground Another antenna placement configuration is to place an antenna on the ground with the interrogation field projected upwards, as shown in Fig It is assumed that an antenna located 2 m above the ground with the interrogation field projected downwards (in Fig. 3.9) will have a similar radiation power at far field as compared with the antenna on the ground. By replacing the rightmost antenna on Fig. 3.8 with an antenna on the ground, the simulation result, shown in Fig. 3.10, shows that the boundary to deploy another reader operating in the same frequency channel moves closer to the first and second antennas, which improves the situation slightly. Page 36

53 Chapter 3 Path Loss and Position Analysis x x Figure 3.8. Results From Simulation (Vertical and Horizontal Axes in 10 x m, Received Power in dbw) Figure 3.9. The Configuration of Antenna on Ground with respect to Normal Configuration of Antenna Page 37

54 3.8 Results and Discussion Reduced x Reduced x 10 Figure Results From Simulation (Vertical and Horizontal Axes in 10 x m, Received Power in dbw) Antennas Operating at Different Channels Previous discussions are on multiple readers operating in a same channel. In this section, the effect of readers operating on neighbouring channels will be investigated. The limit of sideband interference applied is as shown in Fig For two readers operating in direct neighbouring channels, the simulation results on the frequency band of interest are as shown in Fig This result shows that for a transmission operation in the band of interest, the antenna must be placed 50 m away in all direction from an antenna on the ground. For an antenna in the default configuration (Refer Fig. 3.9) operating on the direct neighbouring channel, the distance is around 180 m from the front, 45 m from both sides and 30 m from the back. Table 3.3 shows all the simulation results corresponding to different antenna configurations. From the table, it is very difficult to conclude which configuration is better than the other. It all depends on the type of application an RFID system is deployed for. However, this table is sufficient to serve as a guideline on the safe distance between antennas in a deployment zone. The only restriction is that all the readers must be able to be pre-programmed to operate only in a pre-selected frequency channel. Page 38

55 Chapter 3 Path Loss and Position Analysis Table 3.3. Safe Distance for Different Antenna Configuration Channel Difference Antenna (Default Configuration) Front (m) Side (m) Back (m) Antenna on Ground Figure Results From Simulation (Vertical and Horizontal Axes in m, Received Power in dbw) A More Hostile Environment All the previous simulations are performed based on the path loss model as shown in (3.6), using n = 2.5 for near distance and n = 3.5 for far distance. Hence, the simulation results are approximate indications of various real life situations. A more hostile environment will have a higher n and the safe distance in Table 3.3 will be reduced. On the other hand, a less hostile environment will have an increased distance. However, it does not mean that RFID will perform better in a hostile environment. It is just that in a dense reader and hostile environment, reader placement would be more restrictive under the Listen Before Talk provision when compared to a less hostile environment. Page 39

56 3.9 Real Life Application 3.9 Real Life Application Throughout this paper, it is assumed that the antennas of two different readers are facing each other directly. In other words, the interference caused by other readers will be maximised resulting in the greatest LBT impact. However, by manipulating the placement of the antennas, the interference between two interrogating readers can be minimised. Figure Configuration of antennas Type A configuration as shown in Fig is the worst orientation possible and has been discussed extensively throughout this chapter, with the minimum distance between the two antennas shown in Table 3.1. For configuration Type B, the minimum distance will be reduced. Configuration C will not help much because of the side lobe problem. Configuration D will not solve the problem either, it only minimises the interference between readers. A more complete treatment of readers arrangement is included in Chap. 5. The important point is by just arranging the antenna position and orientation, interference and LBT effect can be reduced but not to a very low level. Hence, readers synchronisation as described below, is required. Page 40

57 Chapter 3 Path Loss and Position Analysis This method is to synchronise all the readers in a particular area. For example, synchronisation of all the readers in a warehouse. All the readers have an absolute sense of time and may be linked using local area network system. They are set to Listen at the same time and since all readers are Listening, all the channels will be unoccupied. Following the ETSI EN , if any reader finds that a channel it begins to examine is unoccupied, its Listening period is fixed. Hence, in a synchronised system, all the readers may start to Listen in a fresh channel at a same time, finish Listening at a same time, and also start to interrogate tags at the same time as shown in the following Fig. 3.13: Using this way, readers can avoid the LBT problem completely. The collision problem can be minimised using proper separation of readers and use of alternate channels for transmitting and receiving. Figure Readers synchronisation Reader Synchronisation Table 3.3 can be used as a guideline for placing readers. However, all the readers have to be assigned channels through a centralised system or through tedious initial set up. Furthermore, addition of new readers into an existing system may require all the readers to be repositioned again. One simple method to solve this problem and adhere to the European regulation at the same time is to synchronise all the readers. All the readers will start to Listen at the same time, assuming no other short range device is operating, and hence will detect no signal at all as no reader is transmitting signals. Following the European regulation, if Page 41

58 3.10 Chapter Summary Figure Reader Synchronisation no signal is detected in the Listen period, the Listen period will end at the same time for all the readers [46], and the readers can start interrogating tags, as shown in Fig Nonetheless, the readers must still be placed carefully to avoid severe reader interference Practical Position of Antenna Whenever the term antenna on ground is mentioned throughout this paper, it is referred to the configuration as illustrated in Fig However, it is usually not feasible to place antenna directly on the ground, except in some very specific applications. For example, in checkout points of retail outlets, it is more sensible to place the antenna on the checkout counter level rather than ground level (Fig. 3.15). The simulation results presented on the antenna on ground are still valid at far distance. There will be a slight difference in the results at near distance but these can be ignored as they will be compensated by the approximation of environment factor, n Chapter Summary We have introduced two methods that can be used to minimise the problem of reader collision and LBT effect. We also have carried a detailed analysis on RFID indoor propagation model. There is room for extension of this research field. In the future, we will identify blind zones in a multi-reader environment, in which a tag will not be able to Page 42

59 Chapter 3 Path Loss and Position Analysis Figure Lifting a Reader on the Ground to Counter Level be detected by any reader. This effort aims to create a complete guide to allow fast and successful deployment of large scale RFID systems and to maximise its potential and benefits. This chapter has presented a path loss model that can be applied to predict the signal strength of RFID interrogation signal at a certain distance away from a transmitting antenna. The model has been implemented using a simulation script written in MATLAB. Essential simulations results are presented, together with some insights and discussions. Safe distances between antennas are suggested and these results can be used in RFID large-scale deployment area to avoid readers interference and to adhere to strict regulations, such as the European Listen Before Talk provision. It has been shown that strict regulations will in fact limit large-scale deployment of RFID system. By using frequency channelling, the situation can be improved. Actual field testing will be carried out in the future, especially in warehouses, where dense RFID reader environments may most likely exist. Page 43

60 Page 44

61 Chapter 4 Reader Synchronisation F OR a dense RFID reader deployment, such as in a warehouse, where hundreds of readers will be positioned in a building, the interference between all these readers must be studied carefully to avoid the disruption of logistics operations. Strict RFID regulations and standards have been imposed, trying to address the problem of reader collision and also the problem of RFID devices interfering with other devices operating in the same and nearby frequency bands. However, these guidelines and regulations are not entirely friendly for dense RFID reader deployment; in some cases it is not possible to have a feasible RFID system while adhering to these regulations. Hence, this paper proposes the synchronisation of RFID readers to enable successful dense RFID reader deployment. A case study targeted at European operations is presented in this paper to illustrate the actual synchronisation of RFID readers in real applications. Some fine-tuning methods are also suggested to further improve the performance of readers in a high reader density population area. Page 45

62 4.1 Introduction 4.1 Introduction Radio Frequency Identification (RFID) has received much attention recently as it is widely believed that RFID can revolutionise supply chain management, complementing barcodes as the main object tracking system. Several major supply chain operators and retailers, such as Wal-Mart in the USA, have deployed RFID systems in some of their supply chains [35]. Initial test runs of RFID deployment show encouraging results [54], and hence large scale RFID deployment is planned. However, before any successful deployment can be achieved, some RFID issues have to be resolved. One of them is the RFID reader collision problem, which is the focus of this chapter. The term reader collision(s) is discussed extensively in [55] and [56]. In this paper, reader collision is simply defined as the phenomenon where an interrogation signal from a certain reader disrupts the communication between a tag and another reader, and this reader collision problem is potentially magnified in a dense reader environment, such as in a warehouse. Various regulatory and standardisation bodies have tried to regulate the operations of RFID readers. In this research, the ETSI as introduced by the European regulatory body and the EPC Class 1 Generation 2 as recommended by EPCglobal are used as the basis of RFID reader operations. However, as will be discussed in more detail in the latter part of this chapter, the restrictions that are put on the operation of RFID readers are very strict, making it quite impossible to have an uncoordinated large scale deployment of RFID readers. Hence, this chapter introduces the idea of RFID reader synchronisation, to enable good RFID performance in a dense reader environment, while adhering to strict regulations. The next section introduces the ETSI and EPC Class 1 Generation 2 Protocol and their impact on RFID reader deployment. Section 4.3 explains the concept of RFID reader synchronisation and how it adheres to strict regulations. Section 4.4 suggests possible ways in implementing a RFID synchronisation system. A case study on RFID reader synchronisation is presented in Section 4.5. Ways of fine-tuning RFID reader positioning are discussed in Section 4.6. Variations of possible reader synchronisation schemes are presented in Section 4.7, followed by conclusion in Section 4.8. Page 46

63 Chapter 4 Reader Synchronisation Table 4.1. Transmit Power and Corresponding Values Channel ERP (dbw) Threshold (dbw) Up to 0.1 Up to to to to to Background ETSI ETSI is a European regulation governing the operation of RFID readers [46]. It allocates the frequency band of 865 to 868 MHz for RFID deployment. This frequency band is then divided into fifteen sub-bands or channels, each spanning a total of 200 khz. However, when a reader is operating at the maximum radiated power, which is 2 W ERP (Effective Radiated Power), only ten sub-bands are available, while the remaining five are utilized as guard bands or for lower power readers. ETSI also introduces the concept of Listen Before Talk. An extract from the ETSI best describes the essence of Listen Before Talk. It states Prior to each transmission, the receiver in the interrogator shall first monitor in accordance with the defined listen time for the presence of another signal within its intended sub-band of transmission. The listen time shall comprise a fixed period of 5 ms plus a random time of 0 ms to 5 ms in 11 steps. If the sub-band is free the random time shall be set to 0 ms [46]. The threshold to determine the presence of another signal within the intended sub-band is shown in Table 4.1. The measurement method is defined in the standard. Furthermore, once a sub-band has been selected, the RFID reader is permitted to use that sub-band for up to 4 s. After use, it must free the sub-band for at least 100 ms. A reader can however, listen to another sub-band for 5 ms and if free use that new sub-band immediately. Page 47

64 4.2 Background EPC Class 1 Generation 2 Protocol EPC Radio-frequency Identification Protocols Class 1 Generation 2 UHF RFID Protocol for Communication at 860 MHz MHz [34], in short EPC C1G2, is the standard protocol developed by EPCglobal for RFID devices for use within the supply chain. This protocol outlines the air interfaces and commands between an RFID reader and an RFID tag. It also includes the spectrum management for RFID operation. Frequency hopping or frequency agile systems are the suggested techniques. An allocated frequency band, as allowed by local regulatory body, is divided into sub-bands or channels. A reader will only use a certain channel for communication, not the entire allocated frequency band. EPC C1G2 covers both dense reader mode and multiple reader mode; multiple reader mode is for an environment where the number of simultaneously active readers is modest relative to the number of available channels while dense reader mode is for an environment where the number of simultaneously active readers is comparable to or more than the number of available channels. This document only focuses on dense reader mode. In dense reader mode, for narrow bandwidth (European 200 khz) channels, it is suggested in this protocol that odd-numbered channels should be used for tag backscatter while even-numbered channels will be used for reader interrogation. For a wide bandwidth channel (USA FCC 500 khz channel [45]), all available channels can be used for reader interrogation, as tag backscatters will be located at the boundaries of these channels Problem in dense reader environment With the implementation of ETSI and EPC C1G2, it is clear that when a reader is operating at a certain sub-band or channel, this reader will effectively prevent other readers from using that channel within an unacceptably large area. Ref. [57] and [58] have presented detailed discussions and analysis on this matter and Table 4.2, as extracted from [57], summarizes the minimum distance (calculated using a piece-wise path loss model with variable environmental factor) between two antennas connected to readers before one antenna operating at a certain channel will prevent the other antenna from using that channel. It should be noted that these results are obtained using Page 48

65 Chapter 4 Reader Synchronisation Table 4.2. Minimum Distance Between 2 Antennas Antenna by Default Channel Difference Antenna on Ground Front (m) Side (m) Back (m) a 0 db isotropic receiving antenna, and do not represent any real life situation, as a typical RFID antenna will be a directional antenna. Nonetheless, the data presented in Table gives sufficient evidence that a low threshold value for the LBT as specified in ETSI is severe enough to impede the reader deployment in a dense RFID reader system. 4.3 Reader synchronisation Under the concept of reader synchronisation, all the RFID readers in a certain area, for example all the readers in a warehouse, are networked together through a central control unit. The connection method can be the common Ethernet connection, or equivalent, and will be discussed in the next section. Since all the readers are linked together, physically or wirelessly, they can be directed to execute commands at a same time. Also, they can be assigned channels dynamically, so that the spectrum management is optimised while the reader collision is minimised. European regulation allows ten channels when maximum radiated power, 2 W ERP, is used. Following the recommendation of EPC C1G2, under dense reader mode, five of them, the even-numbered channels, are used for reader interrogation. All the readers are Listen Before Talk compatible. They are configured to start to Listen at the same time, and then at the end of the listen period, they can all synchronously start to Talk as shown in Fig This is due to the fact that according to ETSI , Page 49

66 4.4 Actual implementation if there is no signal detected in the intended channel of interest, the Listen time is fixed. Hence, all the readers, which start Listening at the same time, will start Talking at the same time. If a reader is turned on at a different time, or if a reader loses synchronisation that reader can be made to start again in synchronism with the rest of the readers, after the last reader has finished its Talk session. Figure 4.1. Synchronisation of All Readers. All the readers start to Listen at a same time and finish Listen at a same time too. 4.4 Actual implementation Connectivity In actual implementation, the RFID readers must be able to communicate with each other to enable synchronisation of the RFID readers. There are basically two ways to connect all the readers; either using wired (physical) connection or using wireless connection. A physically connected system or wired system cannot support mobile readers. Also, a wired connection may suffer from data latency in the network. Ref. [59] shows that time synchronisation in a wired network is possible, but will require additional hardware and system reconfiguration. In the best case, the time difference achievable can be better than 1 ms. A wired system is often considered as a more reliable and a more secure communication method than a wireless communication. A wireless system signals through an RF link. This link can use one of the five guard bands, mentioned in Section 4.2, for sending a synchronising signal. A synchronising Page 50

67 Chapter 4 Reader Synchronisation signal can be a signal with a special pattern. A wireless system can also use any existing wireless protocol such as Bluetooth technology. Wireless synchronisation supports mobile readers but is inevitably vulnerable to interference (signal integrity problem), and unauthorised signal sniffing (security problem). Both connectivity methods have their own advantages and disadvantages. The decision in choosing either of these two methods is largely dependent on the positioning of the LBT sensor, which is discussed below Positioning of LBT Sensors LBT Sensor Reader Reader... Reader Antenna Figure 4.2. Centralized LBT System. Readers are connected to one LBT sensor in a nearby surroundings. Reader LBT Sensor Reader... Reader Antenna Figure 4.3. Localized LBT System. Each reader has its own LBT sensor. An LBT sensor of an RFID reader is responsible for detecting signals in the channel of interest prior to transmission in that channel. This LBT sensor must have a power sensitivity level better than -126 dbw as specified in [46]. If not, this LBT sensor will not be Page 51

68 4.4 Actual implementation able to function efficiently in determining whether there exists a signal with a power level higher than the power level specified in regulations in the channel of interest. An LBT sensor can be the RFID antenna used for transmitting and receiving signals in the communication with RFID tags. A LBT sensor can also be a separate antenna connected to an RFID reader. Also, several RFID readers could share an LBT sensor within a close vicinity. This is also known as a centralised system. A localised system is where each and every RFID reader has its own LBT sensor. A centralised LBT system is as shown in Fig The LBT sensor will constantly monitor all the channels allocated for RFID operation, and dynamically assign available channels to all the readers connected to it. The central control system has to be configured during the initial setup of the system. A fine-tuned centralised LBT system offers high reliability. However, it requires additional network hardware, to connect all the readers to the LBT sensor. Also, a centralised LBT system will not be able to be implemented effectively when mobile readers are dominant in the surroundings. This is because although the readers can communicate with the centralised LBT sensor through a wireless link, it is very difficult for the centralised LBT sensor to estimate the position of mobile readers, and hence is not possible to allocate the best channels for mobile readers. For example, if two mobile readers operate simultaneously in an enclosed area, there is a probability that the two readers move near to each other at some time. The centralised LBT sensor may at that time allocate very nearby channels to those two readers and serious interference between those two readers may occur. Also, if two nearby areas are running on different RFID wireless networks and they are un-coordinated, interference with each other can occur, and in the worst case, cause a complete system shut down. The coordination of wireless networks in different premises will be time and cost consuming. A localised LBT system is as shown in Fig Each reader has its own LBT sensor. The LBT sensor can either be a separate antenna (Fig. 4.3), or be the same antenna a reader uses to establish communication with an RFID tag within its interrogation zone. As compared to centralised LBT system, a localised LBT system with wireless connectivity enables new readers to be easily integrated into an existing system, with no additional cabling or setup needed. However, a localised LBT system has the problem of management of channel sharing, signal interference and possibly creation of Page 52

69 Chapter 4 Reader Synchronisation unwanted shielding. In actual fact, the connectivity of readers and the positioning of LBT sensor are closely related. In [60], a wired system and a centralised LBT are linked together as one configuration, while a wireless system and a localised LBT are linked together as another configuration Antenna Positioning Truck Docks Moderate Read Zone Antenna A Best Read Zone Antenna B h Figure 4.4. Antenna Positioning. A typical antenna setting at dock door, with h being the height of the antenna from the base of a dock door. The positioning of RFID interrogation antennas depends primarily on the application. Detailed operational considerations for the deployment of an RFID system are presented in [61]. In this chapter, only one example will be given, which is the dock door situation, as it will be used in the case study in the next section. A dock door is usually 2 to 3 m in width and approximately 3 m in height. The most effective way to create an RFID interrogation zone is to position two antennas at the sides of the dock door, face-to-face and with an height elevation, h, as shown in Fig The height elevation, h, mainly depends on the average height of objects being shipped through the dock door. A normal choice of h is between 0.5 to 1 m. Also, antenna A and antenna B normally will be using different channels for tag interrogation. However, if antenna A Page 53

70 4.5 Case Study and antenna B are operating at the same time, and a tag is located in the middle of the dock door, the tag may be confused by the interrogation signals from both of the antennas with the result that the tag is misread. This effect is known as the tag confusion problem. The discussion of this issue is outside the scope of this chapter but a simple solution to this is to alternate the operation of antenna A and antenna B every query cycle. 4.5 Case Study A case study on dense RFID reader deployment at the dock doors of a warehouse is presented here. As shown in Fig. 4.5, the dark colour rectangles represent trucks loading or unloading goods at the dock doors of a warehouse. Each door is around 3 m in width, and has two RFID antennas facing each other for tag interrogation. Since all the readers are synchronised in a way described in Section 4.3, they will start Listening at a same time and will be assigned a channel for interrogation at the end of Listen period. The assignment of channels will be geographically influenced. Two readers assigned to be operating in the same channel will be as far apart as possible. Also, the neighbouring antennas will be using channels as far apart as possible. As illustrated in Fig. 4.6, the spectrum is split into ten channels, all five of the odd-numbered channels are reserved for tag backscattering while all five of the even-numbered channels are assigned for reader interrogation. Fig. 4.5 shows how the channel assignment is done. The antenna on the furthest left is using channel 2 for interrogation. The next antenna on its immediate right is using channel 8, which is six channels away. Channel 10, though it is the furthest channel away, is not chosen. This is because the arrangement of 2, 8, 4, 10, 6 gives best channel separation between every channel. Page 54

71 Truck Truck Truck Truck Truck Truck Time Truck Truck Truck Truck Truck Truck Chapter 4 Reader Synchronisation "Listening" Channel Figure 4.5. Alternating of Listening and Talking Mode Channel frequency Figure 4.6. Channelling of The Allocated Frequency Spectrum. Odd-numbered channels will be used for tag backscattered while even- numbered channel will be used for readers interrogation. 4.6 Synchronisation Fine-tuning Fine-tuning of a synchronised RFID system, as presented in this section, can be carried out to further reduce the tendency of reader collision. The fine-tuning methods discussed below include the reduction of output power, the reduction of overall reader talking time, the use of external sensors, the use of RF opaque or absorbing materials, and the frequent rearrangement of channels allocations. Page 55

72 4.6 Synchronisation Fine-tuning Figure 4.7. Reduction of output power. Estimation of required radiated power given that maximum read range corresponds to maximum radiated power Reduction of output power Although up to 2 W ERP can be used in single or small population reader environment, in dense reader populations this higher power may not be necessary. Currently, a state of the art reader can read up to 10 m. However, normal reading operations do not require such a read range. In the case study presented in Section 4.5, the dock doors of the warehouse are around 3 m in width. Since two antennas are positioned facing each other in every dock door, the read range required is also around 1.5 to 2 m. By reducing the radiated power of readers, the minimum distance between two antennas using the same channel can also be reduced, which is beneficial in a dense reader environment. Fig. 4.7 gives an approximation on the reduction of output power. In the far field region, using the Friis equation, the power received is the inverse function of the square of distance (r 2 ). If the maximum read range corresponding to maximum radiated power (2 W ERP) of a RFID reader is known, we can compute the required radiated power for a shorter read range. For example, if the maximum read range of a reader is 5 m using 2 W ERP (shown in Fig. 4.7), and if only a read range of only 2 m is required, the required radiated power can be lowered to 0.32 W ERP. This estimation may not be accurate in real life due to complex electromagnetic propagation phenomena, such as reflection caused by the surroundings objects, but it demonstrates that power reduction is a viable option. Page 56

73 Truck Truck Truck Truck Truck Chapter 4 Reader Synchronisation Reduction of overall reader talking time While it is possible to talk for 4 s, reader applications should be configured to talk for only the time necessary to capture tag data. There is no optimum talking time. It depends on the application and also the surroundings of the deployment zone. On-site fine-tuning and measurements are needed before the reduction of talking time can be carried out Use of external sensors Sensors can be used to turn RFID readers on only when tags are approaching to further reduce reader interference in that area. This will free up the channels allocated for those antennas, and also to avoid unnecessary interference to other surrounding reader antennas. For example, external sensors can be attached to the dock door in the case study in Section 4.5. When the dock door is not in use, the designated RFID readers would be switched off, as shown in Fig Optionally, the central control unit can (as shown) dynamically shift the channels assigned for the antennas at door 3 to door Channel Figure 4.8. Using Sensors in RFID system. Both the antennas at dock door 3 are switched off when the absence of truck 3 is detected. This will free up the channels allocated for those antennas, and also to avoid unnecessary interference to other surroundings reader antennas. Page 57

74 Truck Truck Truck Truck Truck Truck 4.6 Synchronisation Fine-tuning RF opaque or RF absorbing materials Another effective, but more expensive, way to reduce reader interference and collision, is to utilise RF opaque or RF absorbing materials to contain the interrogating signal within the designated zone of interrogation. For the case study presented in Section 4.5, the use of such materials is shown in Fig Although there will still be some signal leakage through the door openings, it would not have caused much interference. This is due to the fact that the signal strengths at the sides of the antenna are relatively weak as compared to the front of the antenna. According to [57], the gain at the side of a typical RFID antenna is approximately 20 db less than the gain at the front of the antenna RF Absorbing Materials Figure 4.9. Use of RF Absorbing Materials. All the antennas are separated using RF absorbing materials. The antennas facing each other at the same door, is at least 4 channels away Frequent rearrangement of channels Interrogating channels can be switched around every cycle of Listen Before Talk. This is to prevent the jamming of the interrogation signal by any external noise. Fig shows a simple example on how the switching is done. There are other more complex switching methods involving higher artificial intelligence in the central control unit, depending on the noise received from the surrounding environment, but these await further development. Page 58

75 Chapter 4 Reader Synchronisation Channel Channel Figure Channel Switching Within Antennas. 4.7 Variation of Synchronisation In the previous sections, suggestions on the implementation of a real life RFID reader synchronisation system are presented, together with some deployment options, such as the connectivity of all the readers. Also, fine tuning methods are presented. In this section, some of the interesting variations of RFID reader synchronisation schemes are presented. These variations may not be readily incorporated into the suggested methods mentioned in previous sections, but are presented here for future reference and for completeness Separation of transmitting and receiving channels 100 mw 500 mw Fig Complete W Figure The complete frequency band allocated for RFID operation as compared to Fig Page 59

76 4.7 Variation of Synchronisation Tx Tx Tx Tx Rx Rx Rx Rx Rx Figure Variation in the separation of Transmitting (Tx) and Receiving (Rx) channel of a RFID reader. Figure Transmit mask for dense-interrogator environments [34]. For the RFID, full power operation (2 W ERP) as governed by ETSI , only ten channels are available, as shown in Fig However, as discussed in Section 4.2, there are actually fifteen channels available for RFID in total. Five of the fifteen channels, though used as guard bands, can be used for RFID operation with reduced maximum allowable radiated power. There are three channels located lower in frequency than the normal ten channels, which can only be operated below 100 mw ERP, while there are two channels higher in frequency than the normal ten channels, which can be operated below 500 mw ERP. The complete frequency range for RFID operation, with respective regulated power level is as shown in Fig The channel numbering system shown in Fig. 4.6 is included in Fig. 4.11, along with a new channel numbering system to simplify the discussion hereon. Channel 4, 7, 10 and 13 are assigned to be the reader transmitting channels while the tag reply channels are the four channels beside the transmitting channels [62]. For example, transmitting channel 4 uses channel 2, 3, 5, and 6 for tag reply. Although the transmitting channels Page 60

77 Chapter 4 Reader Synchronisation are reduced from a total of five down to four, the transmitting channels are placed two channels away rather than one channel away (Fig. 4.12). From the transmit mask shown in Fig. 4.13, an improvement of 5 db can be obtained. Hence with the reduction of interference between transmitting channels, readers can be placed nearer to each other Separation of RFID and non-rfid signals Another variation of synchronisation is to differentiate an RFID signal from a non- RFID signal. A method using signal recognition is presented in [63]. The idea is that all the RFID readers in a certain region can be treated as a single entity in the regulation as outlined in the ETSI Hence, it is only required to avoid the signal interference between all the RFID readers and the rest of the short-range devices. If this concept is valid, the interrogation signals of RFID readers are not treated as a signal in a channel when a LBT test is carried out. The main advantage of this method is that a lot of readers can be deployed in a small confinement area. However, reader antenna positioning can become more challenging, as all the readers can choose any channel for transmission as long as there is no other type of short-range device around. 4.8 Conclusion This chapter has identified synchronisation of RFID readers as a mechanism to assist in RFID reader deployment in dense reader environments. Some implementation methods, and several fine-tuning methods are also presented in optimising the performance of a synchronised RFID system. As compared to conventional unsynchronised RFID systems, a synchronised RFID system can offer more coverage, less reader collision or interference, while strictly following the European regulations and the EPC C1G2 recommendation, and can, with variation of the normal operating procedure, deal also with the effects of tag confusion. However, these benefits require the use of more complex hardware and hence can marginally increase deployment costs. Reader Page 61

78 4.8 Conclusion synchronisation has not been tested in a real situation, and hence will require future study in this area. Page 62

79 Chapter 5 Operational Considerations L ARGE-scale radio frequency identification (RFID) deployment is needed for efficient item identification in supply chains. To reduce cost and save time, simulations are often carried out before actual implementation, especially when RFID is used in regions in which strict regulations and standards must be adhered to. However, due to the unpredictable environmental effects on radio propagation, simple simulation results can be misleading and questions have been raised over the validity of many wireless simulations. This research study reviews, from the point of RFID antenna deployment, the sources of error in wireless simulations reported in some publications. Also, the idea of second carrier sensing is investigated. Page 63

80 5.1 Introduction 5.1 Introduction A common supply chain RFID system runs within the frequency range of MHz in the UHF band, depending on which country the RFID system is being deployed. In most cases, the band RFID is using is also known as the unlicensed Industrial, Scientific and Medical (ISM) band. Sometime there is other electronic equipment, intentional or non-intentional radiators, that are operating in this band. Unlike some of the other signaling equipment permitted to use the band, RFID antennas use comparatively intense RF power, to energise passive tags within their interrogating zones and hence can interfere with nearby electronic equipment operating in the band of interest. Due to this reason, RFID is subjected to very strict regulation around the world. Designers of RFID systems must be aware of the strict regulations to avoid system incompatibility with local limits in European jurisdictions. The provision of Listen Before Talk (LBT) imposes great challenges for antenna positioning. Hence, a good simulation of RFID deployment is essential. Such simulation enables visualisation of RFID deployment prior to actual implementation. However, a simulation result depends on the complexity of the model used for the simulation. A simple simulation can often lead to incorrect conclusion while a complex simulation may consume too much computational time to be feasible. Several papers in the literature have investigated the validity and credibility of wireless network simulation [64, 65, 66]. Of particular interest is [64], where several common simulation assumptions, which contribute the most to simulation inaccuracy, are raised. In this chapter, common simulation errors, such as inappropriate simulation model, neglect of antenna gain, and misinterpretation of simulation results, will be discussed in the context of RFID, along with the exploration of the challenges in RFID simulations. A simulation program has been written to compare with measurement results, and several suggestions have been offered to minimise the possible sources of error. RFID EMC related background information is discussed in the next section, followed by the brief introduction of EPCglobal C1G2 RFID protocol and its proposed transmit mask in a dense reader environment. Section 5.5 looks into several common sources of error in wireless simulation, and provides suggestions and ways of minimising them. Page 64

81 Chapter 5 Operational Considerations Section 5.6 offers some insight into the idea of second carrier sensing, its threshold limit selection and its dependency on antenna positioning, and Section 5.7 provides the conclusions of this research. 5.2 RFID EMC Background The regulatory status for using RFID in the UHF spectrum around the world can be summarised into two categories: governed under Frequency Hopping Spread Spectrum (FHSS), and governed under Listen Before Talk (LBT). By the end of 2005, it is expected that 50 countries, representing 83 % of the global GNI (Gross National Income) will have RFID regulations [67]. It is essential to understand the differences between each regulation before an RFID simulation on deployment is carried out, especially on the allocated bandwidth and the maximum allowable radiated power Frequency Hopping Spread Spectrum (FHSS) An example under this category is the USA FCC Title 47 Part , with operation within the band MHz [45]. This bandwidth is then subdivided into 50 channels of 500 khz. Each channel has a minimum transmission time of 0.4 s. Maximum transmit power is 1 W with a maximum of antenna gain of 6 dbi, giving a maximum total radiated power of 4 W EIRP (Effective Isotropic Radiated Power). It is adopted mainly in North and South America Listen Before Talk (LBT) An example under this category is the European ETSI [46] that has been adopted by some European countries. It allocates the frequency band of MHz for RFID deployment. This frequency band is then divided into 15 sub-bands or channels; each spans a total of 200 khz. When a reader is operating at the maximum total radiated power, which is 2 W ERP (Effective Radiated Power) or equivalently to 3.2 W EIRP, Page 65

82 5.3 RFID Protocol Table 5.1. Transmit Power and Corresponding Threshold Values ERP (W) ERP (dbw) Threshold (dbw) Up to 0.1 Up to to to to to only 10 sub-bands are available, while the remaining 5 are utilized as guard bands in which a low ERP is allowed. A small extract from the ETSI best describes the essence of Listen Before Talk. It says, Prior to Transmission, the interrogator must listen for the presence of another signal within its intended sub-band of transmission. The listen time shall comprise a fixed period of 5 ms plus a random time of 0 ms to 5 ms in 11 steps. If the sub-band is free the random time shall be set to 0 ms [46]. The threshold to determine the presence of another signal within the intended sub-band is shown in Table RFID Protocol Figure 5.1. Transmit Mask for Dense-Interrogator Environments [34]: dbch is defined as decibels referenced to the integrated power in the reference channel. The operation of an RFID system is also standardised to encourage wide spread deployment. EPCglobal has produced EPC Radio-frequency Identity Protocols Class-1 Page 66

83 Chapter 5 Operational Considerations Generation-2 UHF RFID Protocol for Communication at MHz [34], in short EPC C1G2, as the standard operational protocol. This comprehensive protocol includes the detailed air interface between readers and tags and also standard commands for tags interrogation. The section in the protocol which is closely related to this research is on the transmit mask for dense-interrogator environments as shown as Fig A more relaxed transmit mask is used for low reader density surroundings [34] but is not presented here, as the focus of this research will be in a dense reader environment. For a narrow bandwidth channel (European 200 khz channel), it is suggested in this protocol that odd-numbered channels should be used for tag backscatter while even-numbered channel will be used for reader interrogation. For a wide bandwidth channel (USA FCC 500 KHz channel), all available channels can be used for reader interrogation, as tag backscatters will be located at the boundaries of these channels. 5.4 Path Loss Measurement Experiment Main Objectives: 1. To demonstrate the multi-path effect on in-building path loss. 2. To provide experimental data for the investigation of in-building path loss model, and to suggest the best model for a particular room Experimental Setup: The equipment used during the experiment is shown in Fig The antenna used has a gain of 8 dbi, which is higher than the allowable antenna gain for RFID reader antenna of 6 dbi. However, in this experiment, we are using a signal with much lower transmit power which is generated using the HP ESG-3000A signal generator, and hence is with compliance with regulations. Page 67

84 5.4 Path Loss Measurement Experiment (a) Cushcraft 8 dbi S9028PC directional antenna. (b) Cushcraft antenna label. (c) HP 8594E spectrum analyzer 9 k GHz (d) HP ESG-3000A signal generator 250 k MHz Figure 5.2. Equipment used in experiment Room Grid Measurements were carried on a specified grid as shown in Fig The grid was created in Room N204, Engineering North, the University of Adelaide, North Terrace Campus, South Australia, with spacing 1 m between each grid points, in the vertical and horizontal directions Experimental Results: The raw data obtained from the experiment is shown in Table 5.2. The explanation corresponding to each column is as follow: r: The distance between the transmitting and receiving antennas. Page 68

85 Chapter 5 Operational Considerations Door Antennas A B C D Figure 5.3. Room Grid for Measurement Purposes Tx-Pt: The location of the transmitting antenna with reference to the room grid as shown in Fig Rx-Pt: The location of the receiving antenna with reference to the room grid as shown in Fig Tx-Pw: The transmitted power (in dbm) as delivered by Equipment 3. The power level is set, prior to the experiment, to -20 dbm and is used for the entire measuring session. Rx-Pw: The received power (in dbm) as measured by Equipment 2. Gain: The total antenna gain (in dbi) of both the transmitting and receiving antenna (Equipment 1) used in the experiment. Loss: The total losses caused by coaxial cables used in connecting Equipment 1 to Equipment 2 and Equipment 1 to Equipment 3. The loss value presented in Table 5.2 is an assumed value (in db). PL: The path loss (PL), in db, calculated using the measured power, taking consideration of antenna gain and cable loss: PL = (Rx-Pw) - (Tx-Pw) + (Gain) - (Loss), where Rx-Pw, Tx-Pw, Gain and Loss are defined above. FSPL: Theoretical free space path loss (FSPL), in db, of the distance r, calculated using ( ) λ 2 FSPL = (5.1) 4πr Diff: The difference between the PL and FSPL, calculated using (PL) - (FSPL). Ave: The average of the Diff for all the same r. Page 69

86 5.4 Path Loss Measurement Experiment Table 5.2. Experimental Results: Explanation of column titles can be found in Section 5.4.3, with Tx-PW = -20dBm; Gain = 16dBi; and Loss = 3dB. r Tx-Pt Rx-Pt Rx-Pw PL FSPL Diff Ave 1 B8 C A8 B B9 B C9 C A6 A D7 D A8 C B9 B C9 C A6 A D7 D A9 A D7 A A6 A D7 D C7 C D9 D C9 C C7 C A9 A D9 D B9 B A9 A C7 C Continued on next page Page 70

87 Chapter 5 Operational Considerations Table 5.2 Continued from previous page r Tx-Pt Rx-Pt Rx-Pw PL FSPL Diff Ave 6 D9 D A9 A C8 C B9 B A9 A N/A 7 C9 C N/A 7 B9 B N/A 8 A9 A N/A Discussion: 1. It is suspected that PL at all near distances would be slightly lower than the FSPL. This maybe due to the multi-path effect where the propagating waves transmitted by the antenna are reflected by the surrounding objects. At certain locations, constructive interference will take place and will result in a detection of a stronger received signal as compared to a received signal in a reflection free environment (such as an anechoic chamber). However, it has to be noted that destructive interference may also occur. To illustrate this effect, we can consider the addition of two waves, with different phase. The power expression is as shown below: P received = (acosβ 1 z + bcosβ 2 z) 2 = a 2 cos 2 β 1 z + b 2 cos 2 β 2 z + 2abcosβ 1 zcosβ 2 z If a = 1, b can have the value of 1, j, 1, j. The P received is then 4, 2, 0, 2, giving an average P received of 2. A different approach is to compute the expected value of P received as shown below: Page 71

88 5.4 Path Loss Measurement Experiment E [P received ] = E [(acosβ 1 z + bcosβ 2 z) 2] [ ] = E a 2 cos 2 β 1 z + b 2 cos 2 β 2 z + 2abcosβ 1 zcosβ 2 z [ ] [ ] = E a 2 cos 2 β 1 z + E b 2 cos 2 β 2 z +E [2abcosβ 1 zcosβ 2 z] = a2 2 + b ab = a2 2 + b2 2 { 1 2 [cos(β 1z β 2 z) + cos(β 1 z + β 2 z)] } The first approach predicts that although there appears to be nulls due to reflections, on average a gain is expected. The second approach predicts on average there will be a gain due to reflection. (Note: More Discussion required to further the investigation on this matter to be included in this thesis.) 2. From the Diff column in Table 5.2, it can be seen that at some distance away the PL computed using the measured signal strength is constantly lower than the FSPL. For example, when r is 3 m. Figure 5.4 shows the measuring configuration from side view. For simplicity, only one reflection wave is considered, which is the reflection from the ground. The reflection from the ceiling is weaker than the reflection from the ground because the measurements were carried out at an elevation of 1 m as shown in Fig The room where the measurement took place is approximately 3 m in height and the reflection path from ceiling is much longer than the reflection path from the ground. However, reflections from the surrounding objects, such as metal cupboards, walls or metal shelves, can have large impact on the receiving signal, depending on the material of the objects and the path of reflection. Although these reflections from the surrounding objects are neglected in the discussion below, inconsistency in measurements result (as seen in Table 5.2) is most likely contributed by them. Page 72

89 Chapter 5 Operational Considerations r è 1m è 1 2 D Figure 5.4. Measuring Configuration from Side View: and receiving antenna, D is the total length of the reflection path. r is the distance between the transmitting Table 5.3. Difference in Wavelength Between r and D: r is the distance between the transmitting and receiving antenna, D is the length of the reflection path, D-r is the difference between D and r, and wl is D-r in wavelength of 915 MHz (Approximately m),ave is as defined in Section for Table 5.2. r D D-r wl Ave Another assumption is that the angles θ 1 and θ 2 in Fig. 5.4 are the same. This will not be the case if the ground surface is not flat or uneven. Again, for simplicity and to have a predictable reflection path length, a flat and even ground is assumed and hence those angles are the same. The results are shown in Table 5.3. It is very possible that the difference between PL and FSPL is dependent on the difference in wavelength between the incident path length and reflected path length. A difference of half a wavelength after subtraction of full wavelengths will most probably result in higher PL, while a difference of 0 wavelength will result in lower PL. For example, for r = 3 or r = 5, the difference in wavelength is between 0.1 and 0.2, and hence the computed PL values based on experiment measurements are Page 73

90 5.5 Sources of Simulation Error on average lower than the FSPL. On the other hand, for r = 2 or r = 4, the difference in wavelength is between 0.4 and 0.6 and the computed PL values based on experiment measurements are on average higher then the FSPL. However, this argument is not strongly supported by measurements for r = 6. From Table 5.3 and based on the pattern shown by r = 2, 3, 4 and 5, as explained above, the PL values should be lower than FSPL. This inconsistency is most probably caused by reflection from nearby objects. For r greater than 6 m, too few measurements are taken to form any conclusion. For a measurement more than 8 m, the transmitting and receiving antennas are separated by walls and the effect of walls must be taken into account. It is assumed that wall will decrease the measured signal strength and hence increase the computed PL value. This experiment has achieved objective 1 but has yet to produce a model, which can best describe the observation of multi-path effects, mainly caused by reflection. More measurements in the future in different locations would be helpful in gaining more understanding in this study. 5.5 Sources of Simulation Error A good simulation program should be able take into account the EMC regulations and protocol standards discussed in Section 3 and Section 4. Also, the common sources of simulation error of a wireless system as outlined in [64] and [65] must be minimised Path Loss Model The term path loss is widely used in the literature and is carefully explained by the authors in [58]. Often the path loss model in a simulation is considered as a simple function of distance. This is only true for free space propagation, and is used for satellite communication simulations [53]. A typical RFID deployment zone is a warehouse filled with commercial products and logically a more complex model is required. Hashemi has categorised all the path loss models found in literature, apart from the Page 74

91 Chapter 5 Operational Considerations simple free space path loss, into four groups [68]. The authors have studied these four categories carefully and have supplemented those studies with their own work. It is found that a path loss model with variable environmental factor, n, is suitable for the case of RFID as shown in (5.2) below: PL(dB) = PL(d 0 ) + 10 n log 10 ( d d 0 ) (5.2) where d 0 is an arbitrary reference distance, n is the environment factor, d is the separation distance between two antennas and PL(d 0 ) is the free space path loss for a distance d 0. The n value takes into consideration, path obstruction, reflection, absorption and other attenuation effects introduced by the presence of objects inside a building. However, Equation (5.2) does not consider the fact that the n value will increase as the distance increases, and hence is modified as shown as (5.3) below and is explained in detail in [58]: PL(dB) = PL(d 0 ) + 10 n 1 log 10 d d 0 0 < d 8 m (5.3) PL(d 0 ) + 10 n 2 log 10 d d0 d 8 m The comparison with various models and experimental results are plotted in Fig. 5.6 that shows a path loss model in free space, three models based on (5.2) with n = 2, 3 or 4, one model based on (5.3) and also the practical measurements from experiment. It can be seen that (5.2), with n = 2 and PL(d 0 ) set to the value for free space, is actually the same as free space path loss. As the n value increases, the path loss increases for a fixed distance. The practical measurement is set up as shown in Fig A room with metallic cupboards, drawers, and wooden tables is chosen as an experimental site to represent a typical storage area. The room is divided into a grid system, with markers positioned 1 m apart. Two antennas of known gain are used as the transmitter and receiver. These two antennas are directly facing each other (with each at exactly the same height) in order to obtain an orientation with maximum directivity. Several measurements are taken at various positions for a chosen separation and averaged. Page 75

92 5.5 Sources of Simulation Error Figure 5.5. Experiment Grid for Signal Strength Measurement: For measurement more than 8 m, the receiver antenna is moved out of the room through the door shown. The antennas are facing each other with maximum gain attained for all measurements. Figure 5.6. Plot of Path Loss using Different Models Together with Measured Results. Page 76

93 Chapter 5 Operational Considerations The measurement shows several things. The first is that within the room, the path loss is less than the free space path loss generally by a few db. This we attribute to reflections within the room, which would be expected to produce this effect. Secondly, the practical measurement results follow an approximately free space path loss pattern until the first wall is encountered along the measuring path. As the distance increases, more major obstacles appear along the path of measurement, and the practical measurement results have shown path losses higher than a free space path loss. Equation (5.3) best mimics the measured results. Although the error margin can be up to 10 db, it can be minimised, if more appropriate values of n 1 and n 2 are used in (5.3). However, the n value changes from location to location. On-site measurement must be carried out to determine the best n for that certain area before the values of n are used in simulations. Walls have been shown to have a great impact on the value of n. Within a room, one fixed n value can be used, but n must be increased when a wall is encountered as the distance increases. Equation (5.3) offers quick path loss estimation. More complex surroundings with more obstacles in nearby vicinity requires a better model as discussed in next section Reflection, Refraction, and Diffraction Path loss model with environment n (such as (5.3)) has high margin of inaccuracy if the environment of interest is full of obstacles, such as walls, metallic cupboards, and narrow corridors. For example, in Fig. 5.7, the path loss in (a) and (b) will not be the same, but will not be predicted by (5.3). This is caused by wave propagation effects such as reflection, refraction and diffraction. Also, the appearance of obstacles, such as the wall, must be accounted for the calculation of path loss. The ray tracing method is well explained in the literature, such as [69, 70]. It is based on geometrical optics and complemented by geometrical theory of diffraction to take into consideration reflection, refraction and diffraction. On the other hand, [71, 72] modifies (5.2) empirically to cater for reflection, refraction and diffraction. We combined both of the above ideas to produce a hybrid path loss model. The argument is that ray tracing normally requires many traces. If we limit ourselves to not Page 77

94 5.5 Sources of Simulation Error Ceiling P1 Tx Rx Tx Rx P2 Metal Box ground (a) ground (b) Figure 5.7. Signal Received by Receiver Antenna (Rx) in (b) is Stronger than the Signal Received in (a) more than 5 rays, with each ray using an empirical model [71, 72], it is hoped that a simple and accurate model can be obtained. Using the settings presented in Fig. 5.7 (b), we constructed two additional rays (P 1 and P 2 ) apart from the direct ray from transmitting antenna (Tx) to the receiver antenna (Rx). The distance of P 1 and P 2 is measured and with (5.3), the signals strength arrived at Rx through direct path, P 1 and P 2 were added together to obtain the signal strength as detected at Rx. P 1 and P 2 are traced above and below the direct path. Using the same method, additional two rays are constructed at the two sides of the direct path of the transmitted signal to the receiver. The addition of these five signals strength results in the received signal strength as detected by the receiver antenna as shown in Table 5.4. Also, predicted signal strengths using this hybrid method and using (5.3), are compared with mean value of measured signal strengths from experiment. It is assumed that 50% of the power is reflected in the calculation whenever a reflection occurs. The results shows that this hybrid method reduces the error margin. Though this hybrid method is still in its infancy, the main point here is that wave propagation characteristic (especially reflection) will have great impact on signal transmission and must be taken account in simulation planning. Page 78

95 Chapter 5 Operational Considerations Table 5.4. Comparison of Measured and Predicted Signal Strength using Path Loss Model (5.3) and using Hybrid Path Loss Model. Units in dbm unless Otherwise Stated. Distance Signal Eq. (5.3) Eq. (5.3) Hybrid Hybrid (m) (Measured) Results Error Results Error As a side note, fluctuations of signal strength along the direct path between the transmission and the receiver antennas is observed. This is caused by constructive and destructive interference by reflected waves. Simulation models presented in this research are not able to predict and is left as future research Radiation Pattern of Antenna The power transfer between any two antennas is dependent on the angle of orientation between them [64], unless they are isotropic radiators. A typical RFID antenna is a 6dBi gain directional antenna and an example of the radiation pattern of a typical RFID antenna is shown in Fig Despite the fact that in actual RFID deployment, it Page 79

96 5.5 Sources of Simulation Error dbi Figure 5.8. Polar Plot of the Antenna Gain of a Directional Circularly Polarised RFID Antenna with a Gain of 6 dbi is very difficult to ensure all stationary RFID antennas are mounted in the fashion intended, with exact orientation, elevation and angles, the radiation pattern of any type of antenna to be deployed must be acquired and fed into RFID simulation program as accurately as possible. Experience shows that even two antennas of a same model manufactured by a same company will have a slightly different antenna gain patterns. Although this is the case, the difference is small and hence, the simulation program can assume the same gain pattern Simulation Result Interpretation and Analysis A simulation program will use a model to compute the strength of signal at a certain distance away from any transmitting antenna. There are many ways of presenting a simulated result, either numerically or graphically. Fig. 5.9 shows an example of simulated results in a graphical form. However, in real life applications, as shown in [64], the reception of a signal does not exhibit a sharp cliff. For example, the line in the top of Fig. 5 shows signal strength of -110 dbw. This line does not represent a Page 80

97 Chapter 5 Operational Considerations Figure 5.9. Results From Simulation (Vertical and Horizontal Axes in m, Received Power in dbw) clear threshold line between a region with signal weaker than -110 dbw and region with signal stronger than -110 dbw. This is due to uncontrollable factors that present themselves in a real life scenario, such as fluctuation of path loss and antenna gain. However, a boundary zone (Fig. 5.9), rather than a boundary line, can be specified in the simulation result to give good estimate of the overall system performance before an actual deployment is carried out. The size of the boundary zone should reflect the previously measured uncertainties in the propagation loss which have already stated to be a few db. 5.6 Second Carrier Sensing To maintain a low production cost, an RFID tag normally does not have the ability to filter a valid interrogation signal in a channel from the other valid interrogation signals in adjacent channels. Hence the RFID tag can be confused when more than one reader attempts to interrogate it. This situation is common in a dense reader environment, such as a distribution warehouse. Page 81

98 5.6 Second Carrier Sensing B d A d 1 B d 1 (b) B A d d A d 1 (a) (c) Figure Common Antenna Positioning. (a) Face-to-face (b) Side-by-side (c) Corner Mitsugi first classified the Listen Before Talk as first carrier sensing and has suggested a second carrier sensing to avoid the RFID tag being confused by multiple interrogation signals [73]. After a reader has found a free channel using first carrier sensing to send signal, this reader will scan the entire frequency band allocated for RFID systems under the local regulation. If this reader detects any signal in the frequency band higher than a pre-programmed threshold value, it will decide not to transmit any further. Results on random antenna placement have been presented. However, we believe that antenna positioning is not random in nature. In practice, when antennas are mounted, they follow a certain common pattern, such as side-by-side or corner mounting, as shown in Fig We also stressed that second carrier sensing is not required in an environment where only fixed position antennas are deployed, as in this case, they will either fail or pass the second threshold sensing most of the time. However, in an environment where hand held or mobile readers are used, second carrier sensing can prevent tag confusion. Referring to Fig. 5.10, we put a tag at location X, d distance directly in front of antenna A. First, the signal from A (2 W ERP) received by the tag at X is calculated. Then to have a BER of 10 4, a SNR (or CIR) of approximately 15 db is needed. Hence, the signal from B received by tag at X must be at least 15 db lower than the signal from A. Page 82

99 Chapter 5 Operational Considerations From there, the minimum distance of antenna B from the tag and also the minimum threshold for second carrier sensing are obtained. For case (b) and (c) as shown in Fig. 5.10, the antenna gain pattern must be used to find the optimum angle, θ, to have a low d 1 and a low threshold value for second carrier sensing Effect of the Position of Tag To observe the effect of the position of a tag, the environment factor, n, is set to 3.0 for all results shown in this subsection. As the distance of the tag from antenna A increases, the minimum separation distance between antenna A and antenna B, d 1, increases. Once d 1 is obtained, the minimum threshold value corresponds to d 1 is computed as shown in Table 5.5. From the results, we concluded that face-to-face orientation (Fig (a)) should be avoided, as a very high threshold value is required for a very high minimum distance. A side-by-side orientation offers the best results among these three orientations, with relatively low minimum separation distance. A low minimum threshold value for second carrier sensing is desirable as it provides flexibility. A high minimum threshold value (> -30 dbm) will reduce the effectiveness of second carrier sensing. However, too low a threshold value will prevent many readers from sending their interrogation signals at a same time Effect of Environment Factor To observe the effect of environment factor, the tag distance from antenna A, d, is set to 3 m for all results shown in this subsection. For n from 2.5 to 4 with increment of 0.5, the minimum d 1 is calculated for each case together with the minimum threshold value. The result is as expected (shown in Table 5.6), as incrementing n results in higher path loss value. Hence a lower distance is required to attenuate the transmit signal to an acceptable level at the receiver side. Again, similar to the analysis of the effect of position of tag in the previous subsection, side-by-side orientation offers better flexibility for antenna positioning. Page 83

100 5.6 Second Carrier Sensing Table 5.5. Minimum Threshold Values for Second Carrier Sensing with respect to Tag Read Range for Different Antenna Orientation. D (m) Min. d 1 (m) Min. Threshold (dbm) Face-to-face Side-by-side Corner Combining First and Second Carrier Sensing If second carrier sensing is to be applied in a real life situation, the threshold value of the second threshold sensing should be used as the positioning guideline before the threshold value of the first threshold sensing. To strengthen this view, we can imagine a situation where two antennas are positioned near to each other. For the first carrier sensing, most of the time, both of the antennas will be able to find a channel for transmission. (If the two antennas are very near to each other, the two chosen channels will be far from each other.) However, second carrier sensing involves the whole band. If the threshold value is too low and the antennas are too near, they will not be able to fulfill the second carrier sensing. Hence, it is sensible to set a value for the threshold Page 84

101 Chapter 5 Operational Considerations Table 5.6. Minimum Threshold Values for Second Carrier Sensing with respect to Environment Factor, n, for Different Antenna Orientation. n Min. d 1 (m) Min. Threshold (dbm) Face-to-face Side-by-side Corner of second carrier sensing, and position the antenna based on second carrier sensing before the first carrier sensing is used for positioning analysis. In a dense reader environment where antennas placement is very restrictive for both first and second carrier sensing, fine-tuning methods for RFID systems are suggested by the authors in [74]. 5.7 Conclusion This research study has suggested ways of minimising common errors viz. ignoring multi-path propagation, antenna orientation and pattern, and variation of path loss as obstacles are traversed, in the simulation of RFID antenna deployments. Some errors Page 85

102 5.7 Conclusion are unavoidable, but the extent estimated from empirical measurements can be shown to be not severe. Hence it is essential not to interpret any simulation results without bearing all the possible errors and underlying assumptions in mind. This study hopes to provide a clear view of the pitfalls for researchers interested in RFID simulator development. Also, this study has highlighted EMC regulations which are essential in understanding the actual implementation of an RFID system, in order to produce a sensible simulator, which will definitely contribute to the vision of automating supply chains using RFID technology. Page 86

103 Chapter 6 RFID Antenna Design T HIS chapter focuses on the design of RFID tag antennas, beginning with the introduction of relevant antenna theory. Also, the challenges in designing small RFID tag antenna is discussed, before the the design and analysis of various types of HF and UHF RFID tag antennas are presented. The tools used in the process of simulation and fabrication of antenna prototype are also examined in this chapter. Page 87

104 6.1 Introduction 6.1 Introduction The design of an RFID tag antenna for the RFID tag depends heavily on the application of an RFID tag. In the current RFID worldwide market there is a wide variety of RFID tags with very different RFID tag antennas, different sizes, shapes, materials, and operating frequencies. This is to cater for different needs in various applications. Often, an RFID tag antenna designer will be given the specifications of which frequency band the tag is to operating in, the location on the product where the tag is going to be attached to, and the read range performance requirement of for the tag. There are standard design procedures for an RFID tag antenna. They are very useful as a reference when the RFID tag antenna designed is to be attached to an object that does not have a space constraint and which is to be deployed in a RFID friendly environment. However, there are cases where the design of an RFID tag antenna is challenging. Throughout the author s research candidature, some very challenging RFID tag antennas design problems have been encountered, mostly presented by industry partners. This chapter will start off with a brief discussion on the antenna theory involved in RFID tag designing. With a firm grasp of the fundamental electromagnetic properties of antennas, the discussion is extended to the limitations and challenges of designing an RFID tag antenna. Section 6.8 begins with the design of simple HF RFID tag antennas, followed by more complex HF RFID tag antennas which cater for specific needs and specifications. Section 6.9 is very similar to Section 6.8, except the focus is on UHF RFID tag antenna rather than HF RFID tag antenna. As a completion of this chapter, the simulation and prototyping software is included as Section The fabrication process and testing methods of a prototype antenna are also discussed. A detailed investigation on antenna performance measurements are presented in Chap. 8. Page 88

105 Chapter 6 RFID Antenna Design 6.2 Antenna Theory Part of the main focus of this thesis is on the design of HF and UHF RFID tag antennas. Intensive background studies based on [75] and [76] were carried out to establish a strong foundation in antenna theory and to distinguish the difference between the design of a HF RFID tag antenna from the design of a UHF RFID tag antenna. This section lists several important points of the related antenna theory. Common terms in electromagnetism The common terms in the study of electromagnetism are defined by The International System of Units [3] and are as shown in Table 6.1. Table 6.1. SI units for common terms used in antenna theory. Electric field strength E Vm 1 Electric flux density D Cm 1 Magnetic field strength H Am 1 Magnetic flux density B Wbm 1 Permittivity ε Fm 1 Permeability µ Hm 1 Maxwell s Equation The fundamental antenna theory is based on the Maxwell s Equations, first introduced by James Clerk Maxwell in unifying the theories of electromagnetism. The four Maxwell s Equations are the combination of Faraday s Law, Ampere s Law as modified by Maxwell, Gauss Law for electric flux and Gauss Law for magnetic flux. In differential form: Page 89

106 6.2 Antenna Theory E = B t H = J + D t (6.1) (6.2) D = ρ (6.3) B = 0 (6.4) Source and vortex To visualise the electric, magnetic or electromagnetic field, the concepts of source and vortex are introduced. A source like field is as shown in Fig. 6.1(a) and vortex like field is as shown in Fig. 6.1(b). Normally, an electric field is considered as a source like field while a magnetic field is considered as a vortex like field. (a) Source (b) Vortex Figure 6.1. The difference of source and vortex. Boundary condition A brief treatment of electromagnetic boundary conditions is presented here, with its important impact on RFID tag antenna design highlighted. A complete treatment of boundary conditions can be found in [77]. Electromagnetic waves will undergo changes when they transverse from a certain medium into another medium. If one of the mediums is a conductor, such as a metallic surface, there can only exist tangential magnetic field and normal electric field at the boundary. Fig. 6.2 shows the changes on electric and magnetic field near a metallic surface. Near field and far field The analysis in this section is from [75] and [78]. The analysis applies to the case where r > 0. Page 90

107 Chapter 6 RFID Antenna Design Wire Charge Metallic Surface Figure 6.2. Boundary condition affecting electric and magnetic field. If the metallic surface is a perfect conductor, there are only tangential magnetic field and normal electric field at the boundary. The E and the H fields for a infinitesimal dipole can be expressed as: H φ = β2 I 0 l 4π E r = η β2 I 0 l 4π E θ = η β2 I 0 l 4π H r = H θ = 0 (6.5) [ j (βr) + 1 ] (βr) 2 e jβr sinθ (6.6) [ 2 (βr) 2 2j ] (βr) 3 e jβr cosθ (6.7) [ j (βr) + 1 (βr) 2 j (βr) 3 ] e jβr sinθ (6.8) E φ = 0 (6.9) When r = 2π λ, the first and the third terms in the bracket of (6.8) are the same in magnitude but opposite in sign, and hence will cancel out each other. However, when r < 2π λ, the magnitude of the third term in the bracket of (6.8) will be greater than the second term in the same bracket. In the case of r 2π λ, the third term in the bracket of (6.8) will be completely dominant over the other two terms. It can be easily proven from (6.5) - (6.9), and by using the formula for time-average power density: W av = 1 2 Re[E H ] (6.10) that since the E and H fields are in time-phase quadrature, there is no no timeaverage power in the region r 2π λ. This zone (where r λ 2π ) is known as Page 91

108 6.2 Antenna Theory the energy storing zone where the energy is imaginary. On the other hand, when r > 2π λ, the first term will be dominant. In this zone, the energy is radiating and real. A similar inspection can be carried for a magnetic dipole. The E and the H fields of a small loop or infinitesimal magnetic dipole can be expressed as: E φ = β2 jωµ 0 IA 4π H r = 1 β 2 jωµ 0 IA η 4π H θ = 1 η β 2 jωµ 0 IA 4π E r = E θ = H φ = 0 (6.11) [ j βr + 1 ] (βr) 2 e jβr sinθ (6.12) [ 2 (βr) 2 2j (βr) 3 [ j βr + 1 (βr) 2 ] e jβr cosθ (6.13) j (βr) 3 ] e jβr θ (6.14) When r 2π λ, the third term in the bracket of (6.14) will be completely dominant over the other two terms. The E and H fields are in time-phase quadrature, and hence there is no no time-average power in the region of r λ 2π. Hence, r = 2π λ is defined as the boundary of the near field and far field. The near field is the energy storing field while the far field is the energy radiating field. For the operation of HF and UHF, this boundary is at a distance of 3.5 m and m respectively. Coupling and radiation Coupling and radiation are two different means of power transfer from a point to another point without using any physical connection between those two points. In other words, power is transferred using a wireless connection, Its possibility was predicted by the Maxwell equations. Coupling normally takes place in the near field, which has been defined previously, in the form of either inductive or capacitive coupling. In coupling, energy flows out of the source in the form of electric or magnetic field and flows back the to source. Ideally, there will be no energy loss if there is no load near the field. Hence, a near field is also known as the energy storage field. If there is a load Page 92

109 Chapter 6 RFID Antenna Design in the near field, energy will be transferred from the source to the load through coupling. Radiation normally takes place in the far field, which has been defined previously. In the case of radiation, energy propagates or radiates away from the source and never returns to the source. When a load adsorbs radiated energy, the source will not be affected. However, the term couple or coupling is also commonly used in literature with a different meaning. For example, the phrase antenna A is coupled to the magnetic field does not mean that the antenna A is coupled in the near field. It means that antenna A is sensitive to the magnetic field, whether it is in the near or far field. Power Transfer In the near field operation, to estimate the power received by an RFID tag chip, coupling volume theory as introduced in [79] is applied. Coupling volume is a figure of merit. For a magnetic dipole, the equations involved are given by: V c = Reactive power flowing in the untuned label coil when it is short circuited Volume density of reactive power created by the interrogator at the label position (6.15) V d = Recative power flowing in the inductor of the interrgator field creation coil Volume densty of reactive power created by the interrogator at the label position (6.16) P 2 P 1 = V c V d Q 1 Q 2 (6.17) Page 93

110 6.2 Antenna Theory Coupling volume for a electric dipole is slightly different in definition: V c = Reactive power flowing in the untuned label capacitor when it is open circuited Volume density of reactive power created by the interrogator at the label position (6.18) V d = Recative power flowing in the capacitance of the interrgator field creation electrodes Volume densty of reactive power created by the interrogator at the label position (6.19) Radiation theory is often applied when working in the far field zone. The power transfer from the transmitting antenna to the receiving antenna can be calculated using: ( ) P r λ 2 = g t g r (6.20) P t 4π where λ = wavelength; P t = transmitted power; P r = available received power; g t = transmitter antenna gain; g r = receiver antenna gain; d = separation distance between antennas. However, (6.20) assumes perfect matching. An RFID reader antenna often has a good match with an RFID reader RF front end. However, owing to size constraint, small RFID tags may not have a perfect match between the RFID tag antenna and the RFID tag chip. This mismatch loss must be taken into account using (6.21), which will be discussed next. Maximum power transfer A passive RFID tag normally does not have its own power source and it is powered up by an interrogating antenna using CW transmission. Page 94

111 Chapter 6 RFID Antenna Design Hence to enable maximum read range, power transmission between the tag antenna and the tag chip must be maximised. In theory and it can be proven easily, the maximum power transfer occurs when the impedance of the antenna is the conjugate of the impedance of the tag chip. If there is no conjugate match, the power loss can be computed through [75]: P lost P available = Z ant Z cct Z ant + Z cct 2 (6.21) where Z ant is the input impedance of the antenna (same in our case) and Z cct is the input impedance of the circuit which is connected to the antenna at its input terminals (in our case, this is the input impedance of an RFID tag chip used in prototyping). Polarisation Polarisation is the curve traced by the end point of a vector representing the instantaneous electric field [75]. There are three types of polarisation, linearly, circularly and elliptical. An RFID tag antenna is usually linearly polarised. However, the orientation of an RFID tag antenna may change rapidly as the object it is attached too is being transported along the supply chain. If a linearly polarised RFID reader antenna is deployed and if the RFID tag antenna is aligned orthogonally to the reader antenna, the reader will not be able to communicate with the RFID tag. Based on this, RFID reader antennas are often circularly polarised. Polarisation efficiency is defined as the ratio of the power received by an antenna from a given plane wave of arbitrary polarisation to the power that would be received by the same antenna from a plane wave of the same power flux density and direction of propagation, whose state of polarisation has been adjusted for a maximum received power [75]. It can be observed easily, without mathematical proof, from a Poincáre sphere [76], that the polarisation efficiency between a circularly polarised RFID reader antenna and a linearly polarised RFID tag antenna is 0.5 (or -3 db). Page 95

112 6.3 Antenna Parameters Scattering Parameters Scattering parameter of a network can be defined by: S ij = V i V + j (6.22) where V + j is a known signal entering through port j; V i is the signal detected in port i when port i is terminated with a matched detector. It should be noted that all other ports must be terminated with matched load. A normal antenna can be considered as a single port network. Hence, the most common S parameter is the S 11. S 11 is the ratio of the reflected signal and the incident signal, also known as the input reflection factor. S 21 is sometimes used when two antennas are involved, one as the transmitter and the other one as the receiver. S 21 measures the transmission quality between those two antennas. In the case of two passive antenna systems, S 21 = S 12, due to reciprocity. In measuring an antenna using a network analyser, the reflection measurement is defined as the ratio of reflected power and incident power, which is S Antenna Parameters Common terms to describe the characteristics of an antenna, also known as the antenna parameters, are referred to [80, 75, 76, 81, 82]. Proper definitions are obtained from the IEEE Standards Definitions of Terms for Antennas [80] to avoid any ambiguity: 1. Directivity Directivity of an antenna is the ratio of the radiation intensity in a given direction from the antenna to the radiation intensity averaged over all directions. The average radiation intensity is equal to the total power radiated by the antenna divided by 4π. If the direction is not specified, the direction of maximum radiation intensity is implied. Page 96

113 Chapter 6 RFID Antenna Design 2. Gain In simple form, gain is the directivity of an antenna, taking into consideration the dissipative loss of an antenna. If an antenna does not have dissipative loss, the gain of this antenna is equivalent to the directivity of this antenna. Gain is defined as the ratio of the radiation intensity, in a given direction, to the radiation intensity that would be obtained if the power accepted by the antenna were radiated isotropically. Gain does not include losses arising from impedance and polarisation mismatch. 3. Radiation Pattern The radiation pattern is the spatial distribution of a quantity that characterises the electromagnetic field generated by an antenna. The quantities that are most often used to characterise the radiation from an antenna are proportional to, or equal to, power flux density, radiation intensity, directivity, phase, polarisation, and field strength. When the quantity is not specified, an amplitude or power pattern is implied. In this thesis, the radiation pattern plots are showing either the gain or directivity of the antenna under test, unless otherwise stated. 4. Polarisation See Section Input Impedance The impedance presented by an antenna at its terminals. Impedance of an antenna, Z, is normally presented in the form of R + jx, where R is the resistance of the antenna and X is the reactance of the antenna. 6. Quality Factor Quality factor, Q, of a resonant antenna is the ratio of 2π times the energy stored in the fields excited by the antenna to the energy radiated and dissipated per cycle. For an electrically small antenna, it is numerically equal to one-half the magnitude of the ratio of the incremental change in impedance to the corresponding Page 97

114 6.4 Resonant Circuit for Antenna incremental change in frequency at resonance, divided by the ratio of the antenna resistent to the resonant frequency. 7. Bandwidth The range of frequencies within which the performance of the antenna, with respect to some characteristic, conforms to a specified standard. 8. Radiation Resistance Radiation resistance is the ratio of the power radiated by an antenna to the square of the RMS antenna current referred to a specified point. The total power radiated is equal to the power accepted by the antenna minus the power dissipated in the antenna. 6.4 Resonant Circuit for Antenna To estimate the performance of antenna, it is easier to visualise an antenna as its resonant circuit, which consists of just resistors, inductors and capacitors. From a resonant circuit, it is convenient to compute the quality factor of an antenna. However, it must be stressed that this is purely an estimation as at high frequency, a distributed circuit is more accurate than a discrete component resonant circuit. Nonetheless, a discrete circuit is sufficient to provide a good initial estimation of the performance of a new design. There are two types of simple resonant circuits; series and parallel resonant circuits: Series Resonant Circuit A series resonant circuit consists of a resistor, R s, an inductor, L s, and a capacitor, C s, linked together in series as shown in Fig Resonant frequency of a series resonant circuit is given by: = 1 2π LC (6.23) Page 98

115 Chapter 6 RFID Antenna Design R s L s V C s Figure 6.3. Series Resonant Circuit Quality factor of a series resonant circuit is given by: Q = X L r (6.24) where X L the reactance of the inductor. Bandwidth of of a series resonant circuit is given by: BW = f 0 Q (6.25) where f 0 is the resonance frequency of the circuit. Parallel Resonant Circuit A parallel resonant circuit consists of a resistor, R p, an inductor, L p, and a capacitor, C p, aligned in parallel as shown in Fig I Rp Lp Cp Figure 6.4. Parallel Resonant Circuit Resonant frequency of a series resonant circuit is given by: = 1 2π LC (6.26) Quality factor of a parallel resonant circuit is given by: Page 99

116 6.4 Resonant Circuit for Antenna Q = Bandwidth of of a parallel resonant circuit is given by: r X L (6.27) BW = f 0 Q (6.28) Practical Parallel Resonant Circuit The parallel resonant circuit shown in Fig. 6.4 is often known as the ideal parallel resonant circuit. In the case of antenna design, no actual resistor will be added in the antenna. The resistor in a resonant circuit of an antenna normally represents the losses introduced by the inductor. Hence, as an alternative, Fig. 6.4 can be rearranged as Fig. 6.5, also known a practical parallel resonant circuit. r I L p C p Figure 6.5. Practical Parallel Resonant Circuit Though Fig. 6.5 better characterises an antenna, it is easier to analyse mathematically when an antenna is represented in the form of ideal parallel resonant circuit. It can be done be transforming the losses in inductor, r, to the parallel resistor, R p using: R p = (ω 0L p ) 2 r (6.29) where ω 0 is the angular resonant frequency of the circuit. Different types of RFID tag chips are discussed in Section For the case of HF chip, the input impedance is just the input capacitance. Hence, the capacitor (C s in Fig. 6.3 and C p in Fig. 6.4) Page 100

117 Chapter 6 RFID Antenna Design is the input capacitance of the HF chip. The inductor and the resistor represent the inductance of the HF coil antenna and the losses within the coil respectively. The case of UHF RFID tag chips is more complex, in the sense that the input impedance of a UHF RFID tag chip is represent by a resistor in parallel with a capacitor. For example, a Texas Instruments RI-UHF-STRAP-08 UHF Gen2 Strap [83] (More details in Section ) has an input impedance specified by 380 Ω 2.8 pf, or j in series. A UHF RFID tag antenna normally has a resultant inductive input impedance to minimise the power transfer loss dictated by (6.21) between the antenna and its UHF RFID tag chip. Hence, it can be represented by an inductor and resistor. Note that, this does not mean an antenna does not have self-capacitance. It is just that at the frequency of interest, the antenna is inductive in impedance. A resonant circuit can then be formed using the inductor and resistor from the antenna, together with the capacitor and resistor from the chip. Note also that the bandwidth of a UHF RFID tag does not equal to the bandwidth of a UHF antenna. This is caused by the additional resistance from the input impedance of the UHF chip. 6.5 Challenges in RFID Tag Antenna Design (More to be added in. Suggestions welcome.) The design procedures of an HF RFID tag antenna is mature and well-documented, such as [84]. Basically, an HF RFID tag antenna provides enough inductance to resonate with the HF RFID tag chip input capacitance, to operate at the frequency of MHz. The HF tag is designed to have a reasonably high quality factor, to have a satisfactory read range (Some HF tags are purposely designed to have a low Q as specified in Part 3 Mode 2). The read range of HF system is often limited by the reader antenna size, not the HF tag antenna, though in some cases the limiting factor is the size of the tag antenna itself. Page 101

118 6.5 Challenges in RFID Tag Antenna Design However it is not the same case for the design of UHF RFID tag antennas. The UHF RFID tag limitations is one of the main performance limitation of a passive UHF RFID systems, which are chip sensitivity, antenna gain, polarisation and impedance matching as according to [85]. All of the limitations mentioned above are related to an RFID tag antenna, apart from the limitation of chip sensitivity. The following investigation in this section on the challenges in designing UHF RFID tag antenna is based on two co-authored papers: Analysis of constraints in small UHF RFID tag design [86] and Small UHF RFID label antenna design and limitations [87]. Size Limitation An electrically small antenna is often defined as an antenna with length lesser than 10 1 of wavelength at its operating frequency. 1 For HF (13.56 MHz), 10 of wavelength is approximately 2.2 m while for UHF ( MHz), 10 of wavelength is approximately mm. Since the size of an RFID tag is always depends on the size of its antenna, and for most application sit is not feasible to have a large RFID tag, an RFID tag antenna is often small in size and considered as an electrically small antenna. A small antenna often has small radiation resistance. From (6.24), a small r will result in a high quality factor, Q. A high Q means small bandwidth. Ref. [88] shows that many electrically small antenna can be matched to a 50 Ω line. However, the quality factor of the antenna is not significantly changed. Bandwidth Limitation According to Bode and Fano, the fundamental limitation on impedance matching takes the form [89]: 0 ln 1 Γ dω π RC (6.30) Page 102

119 Chapter 6 RFID Antenna Design where Γ is the reflection coefficient of the load and its assumed lossless matching network with respect to the source impedance R S, and R and C is the resistance and capacitance, respectively, that comes from the parallel RC load. To have a wide matching frequency band, a maximum mismatch outside ω (the band of interest) is required.. π The best utilisation of RC is to keep Γ constant (say at Γ inband ) over the band ω, and unity outside this band as shown in Fig Figure 6.6. Reflection coefficient for best utilisation of λ RC. Based on the case shown in Fig. 6.6, (6.30) becomes Γ inband e 1 2 f RC (6.31) Gain Limitation An RFID tag antenna is categorised as small antenna. An efficient small RFID tag antenna is normally assumed to have a gain of 1.5. For an operating RFID reader antenna the maximum transmitted power that is allowed under FCC regulations is 1 W into an antenna with a maximum gain of 6 dbi. Polarisation Loss Page 103

120 6.6 RFID Chips From Section 6.2, the polarisation efficiency between a circularly polarised RFID reader antenna and a linearly polarised RFID tag antenna is 0.5 (or -3 db). However, it is almost impossible to have a perfect circularly polarised antenna. An antenna is more likely to be elliptically polarised and depending on the axial ratio, the maximum polarisation loss will increase with respect to the -3 db of a circularly polarised antenna. Ref. [90] shows that with a 3 db axial ratio (2:1 ratio), the maximum polarisation loss will increase to 4.77 db. Impedance Matching A RFID chip is always capacitive in nature due to its reservoir capacitor in its rectifying circuit. A matching antenna to a capacitive RFID chip must be inductive in nature. In the case of HF, the inductance of the HF antenna is tuned to resonate with the capacitance of the HF chip at the frequency of operation. The inductance needed can be provided by a multi-turn loop antenna. More discussion can be found in Section 6.8. In the case of UHF, Eq. (6.21) is used to approximate the power transfer from the antenna to the chip. The best case is when the impedance of the UHF antenna is the conjugate match of impedance of the UHF chip. However, a simple dipole that is usually less than half-wavelength in length normally has a capacitive input impedance [75]. A matching network is required to match the dipole to the RFID chip. Commercial UHF RFID tag antennas normally have a matching network very similar to a conventional T-match network, which can be found in [75]. 6.6 RFID Chips This section shows the common RFID chips used in experiments to test the performance of an RFID tag antenna in terms of readability and its maximum read range. Page 104

121 Chapter 6 RFID Antenna Design In the early stage, HF ICs were obtained from Texas Instruments (RF-HDT-SJME-G1 Tag-it HF-I transponder IC) [91] and NXP-Phillips (I Code SLI SL2ICS20 Smart Label) [92]. Both of these tag ICs conform to ISO Part 3 Mode 1 [32] and ISO [33]. The problem is that there is no way to attach a IC directly onto an antenna. Hence a discussion was held with a technical group from Defence Science and Technology Organisation (DSTO), Australia. It was discovered that a solder bump can be used to connect electrically an IC to a thin metal film as show in Fig Solder HF Chip 1.5 mm Bond Solder 3 mm Metallic Pad Ceramic Few micron 0.38 mm Figure 6.7. The proposed chip bonding onto a thin metal strip. However, due to cost factor, the idea of bonding of ICs onto a conductive strips was abandoned. As an alternative, ICs from existing tags were reused. The tag of choice is Texas Instruments Tag-It HF-I C04, as shown in Fig. 6.8(a). A HF chip is cut out carefully from a Tag-It HF-I C04 with a small portion of the metal film remains attached. The chip is then attached onto a prototyped antenna as shown in Fig. 6.8(b). There is an electrical connection from the chip to the antenna through the metal film. A special tape is used to stick the metal film and the antenna together. A thorough discussion can be found in Section The chip reused from a Tag-It HF-I C04 will be known as the TI HF chip in this thesis. It has an input capacitance of 23.5 pf ±10% [91] at the operating frequency of MHz. Another tag chip which can be reused is the ISD72128 or C220, originally designed by Integrated Silicon Design Pty Ltd (an Australian Company) and manufactured by Chartered Semiconductor Manufacturing (Singapore)as shown in Fig However this tag chip uses a proprietary (Tag Talk First) protocol and can only be interrogated by proprietary RFID readers, also available originally from Integrated Silicon Design Pty Ltd. The functionality of this TTF chip is as followed with respect to the induced Page 105

122 6.6 RFID Chips Tag Chip (a) A Tag-It HF-I C04 (b) The chip of a Tag-It HF-I C04 reused on prototyped antennas Figure 6.8. voltage at its input terminals: (1) At low voltage, the tag chip will generate a 100 µs burst of 100 khz squarewave. This is an EAS (Electronic Article Surveillance) burst. (2) At higher voltages the tag will generate a reply burst, 4 cycles of either 250 or 400 khz, depending upon the code written to memory. The sequence repeats at a duty cycle of between 10 to 20 percent, depending upon excitation levels, 10 percent for high excitations and 20 percent for low excitations Figure 6.9. A ISD72128 or C220 HF RFID tag. There is a variant of HF tag chips with much higher input capacitance. For example the NXP-Phillips SL1 ICS31 01 ICODE1 Label IC [92] has an input capacitance of 97 pf. The advantage of having a higher input capacitance is that the antenna for the chip can have fewer turns to reduce the overall dimension, as the antenna does not need to provide a high inductance to resonate with the input capacitance of the chip. Also used was the ISD72128 packaged in a smart card module. The tag chip is as shown in Fig The tag chip which is already mounted on the smart card module, makes it easy to solder onto an antenna. Page 106

123 Chapter 6 RFID Antenna Design Figure TTF chip on smart card module. Unlike the difficulties found in mounting the HF chip onto an antenna substrate, UHF chips are readily available attached onto a thin metal straps. A special tape (Section 6.10) is used to attached a UHF strap onto a UHF antenna. An example of UHF straps used in the fabrication of UHF tag is as shown in Fig. 6.11, which is the RI-UHF-STRAP-08 UHF Gen2 Strap from Texas Instruments [83]. It has a input impedance of pF which is j at 920 MHz. Figure A Texas Instruments UHF straps. Another type of UHF straps used in experiments is the C1G2 UHF straps from Alien Technology (Fig. 6.12). Figure A Alien C1GX UHF straps. Page 107

124 6.7 RFID Readers 6.7 RFID Readers This section lists the common RFID readers used for the all the actual testings presented in this thesis. HF readers operates at MHz. UHF readers can to controlled to operate within 860 to 960 MHz. Unless a testing was carried in a controlled environment, such as an anechoic chamber, the testings presented in this thesis were carried out in adherence to Australia regulation, which is in the band of 920 to 926 MHz, at 4 W EIRP (Effective Isotropic Radiated Power) HF Reader ID ISC.LR2000 HF reader and ID ISC.ANT300 HF reader antenna, both from FEIG Electronics. It supports ISO 15693, ISO Part 3 Mode A (a) Reader (b) Antenna Figure (a) 1: Connected to reader antenna; 2: Optional connection to a computer through RS232 interface; 3: Connection to a computer through a LAN connection using TCP/IP protocol. (Connection to power supply not shown) (b) 1: Connected to a reader. L120 HF reader and HF AC201 reader antenna, both from Gemplus, previously known as Integrated Silicon Design Pty Ltd. It supports the proprietary TTF protocol. The suitable RFID tags are the ISD72128 (C220) UHF Reader ID ISC.LRU2000 UHF RFID reader from FEIG Electronics. It supports Part 6, EPC C1G1 and EPC C1G2 tags. (Awaiting the antennas from FEIG). Page 108

125 Chapter 6 RFID Antenna Design (a) Reader (b) Antenna Figure (a) 1: Connected to reader antenna; 2: Connected to a computer; 3: Connected to power supply. (b) 1: Connected to a reader; 2: Ferrite core to reduce noise (a) Reader (b) Antenna Figure (a) 1: Connected to reader antenna; 2: Connected to a computer. (b) Connected to a reader. ALR 9780 UHF RFID reader and ALR 9610-BC reader antenna, both from Alien Technology. It supports EPC C1G1 and EPC C1G (a) Reader (b) Antenna Figure (a) 1: Connected to a computer using serial connection; 2: Connected to power supply; 3: Connected to reader antennas; 4: Unused connection to antennas. (b) 3: Connected to a reader. Page 109

126 6.8 HF RFID Antenna Design 6.8 HF RFID Antenna Design This section presents the design of a HF RFID antenna. A HF RFID system runs at MHz. A HF RFID antenna is designed so that together with a RFID tag chip, it resonates at the operating frequency. Firstly, this section discusses the use of a simple loop antenna as a HF RFID antenna. In the later part, various novel HF antenna designs for various applications are shown. The literature, [93, 84, 94, 95] provide guidelines in designing an HF RFID antennas. The fundamental idea in designing HF antenna is the same, which is to have the deigned antenna providing sufficient inductance to tune with the HF chip capacitance, to have a resonant point at the operating frequency. This section is not meant to overlap with existing literature. This section touches lightly on the theory and focuses more on the application-driven engineering challenges; from understanding a unique RFID application, initial antenna simulation, to antenna prototyping and finally antenna testing Simple Loop Antenna A simple loop antenna acts as a magnetic dipole. The formulae used and the detailed design procedures are described by Balanis [77] and Bernhard [82]. Optimisation of an HF antenna is discussed by Chen [96] while Lee offers a design note on RFID antenna circuit design [93]. An HF electrically small circular loop antenna with radius a and wire radius b [77] has the radiation resistance given by: R r = 20π 2 (βa) 4 (6.32) and inductance, [ ( )] 8a L = µ o a ln b 2 (6.33) For a square or rectangular (not circular) loop antenna, its characteristics can be estimated using the area encompassed by the loop: Page 110

127 Chapter 6 RFID Antenna Design πr 2 = a b (6.34) where r is the radius of an equivalent circular loop antenna, a and b are the width and the length of the rectangular or square (where a = b) respectively HF Planar Spiral Coil Antenna The characteristics of a simple one-turn planar coil antenna can be estimated using the characteristics of a simple loop antenna. The equivalent of a simple one-turn planar coil antenna as a simple loop antenna can be found by using the table of conductor geometrical shapes and their equivalent circular cylinder radii [75]. A width of the planar coil, a, is related to the radius of the cylindrical coil, a e, though the relation a e = 0.25a. For a multi-turn coil antenna where the coil maintains a fixed radius, the total inductance can be estimated using: L total = n 2 L (6.35) where L total is the total inductance, L is the inductance of one loop, and n is the number of loops. However, it is impossible to have a multi-turn planar coil antenna with fixed radius. Hence to predict the total inductance of a multi-turn planar coil antenna, a more complex method from [97] is applied. (6.36) to (6.42) and their explanations are adapted from [97]. The exact expression for the inductance of a straight conductor is given as: L = 0.002l [ ( ) 2l ln AMD ( µ ) ] + T GMD l 4 (6.36) where Page 111

128 6.8 HF RFID Antenna Design L = inductance (µh) l = conductor length (cm) GMD = Geometric Mean Distance of the conductor cross section. GMD is defined as the distance between two infinite thin imaginary filaments whose mutual inductance is equal to the mutual inductance between the two original conductors. For a single conductor, GMD is defined as the distance between two imaginary filaments normal to the cross section, whose mutual inductance is equal to the self-inductance of the conductor. The computation of GMD is often lengthy but for common cross sectional shapes, the GMD can be obtained in research literature. In the case of planar coil, where the cross sectional shape is a very thin rectangle, the values for GMD is given by (a + b), where a and b are the width and height of the cross sectional rectangle. AMD = Arithmetic Mean Distance of the conductor cross section. AMD for a single conductor is the average of all possible distances within the cross section. In the case of planar coil, where the cross sectional shape is a very thin rectangle, the values for AMD is given as 1 3. µ = conductor relative permeability. T = frequency-correction parameter. Although T can be considered 1 for microwave frequency, we still assumed a value of 1 in HF. The argument for that is out of the study of this research and is an assumed value here. To reduce (6.36) for our inductance calculation, we use T = 1, µ = 1, and consider the copper strips (of the antenna loop) as having a very thin rectangular cross sectional shape (i.e. b 0): L = 0.002l { [ ] 2l ln a + b [ ]} (a + b) 3l (6.37) The aim is to compute the inductance of a planar spiral antenna. The total inductance is contributed by both self inductance and mutual inductance by all the segments of the coil. The self inductance can be obtained by calculating the self inductance of each segment using (6.37) and summing all the self inductances of the segments. Page 112

129 Chapter 6 RFID Antenna Design The computation of mutual inductance is more complex, as there are positive and negative mutual inductances. Positive mutual inductance is caused by two parallel conducting segment with same current direction while negative mutual inductance is caused by two parallel conduction segments with different current directions. We assume that there is negligible mutual inductance between two perpendicular conducting segments. Hence the total inductance of a planar coil antenna is given by (6.38): L Total = L sel f + M + M (6.38) and M = 2lQ (6.39) where L sel f is the total self inductance, M + is the total positive mutual inductance (in nh), M is the total negative mutual inductance (in nh), l is the length of the segment (in cm), and Q (do not confuse with quality factor, Q) is the mutual inductance parameter calculated using (5). ( ) [ l Q = ln GMD ( ) ] 1 [ l 2 2 GMD 1 + ( ) ] 1 GMD 2 2 ( ) GMD + l l (6.40) However, if the two parallel conducting segments are not the same in length (which is the case of a planar spiral antenna), the mutual inductance between them has to be calculated using (6): 2M a,b = (M b+p + M b+q ) (M p + M q ) (6.41) In the computation of the inductance of the planar spiral antenna, it is assumed that the length of p and q are the same and equation (6) is then reduced to: M a,b = M b+p M p (6.42) Page 113

130 6.8 HF RFID Antenna Design This planar spiral antenna must provide sufficient inductance so that the reactance provided by the inductance can cancel the reactance provided by the chip capacitance at the frequency of interest (which is MHz for HF operation). If we assume chip capacitance of approximately 1 pf in series, we would need a series inductance of 140 µh in series, calculated using: f 0 = 1 2π LC (6.43) This amount of inductance is difficult to achieve using a spiral planar coil structure. The normal practice is to increase the capacitance instead, by attaching a parallel capacitor to the chip s input capacitance. However, at this stage, it is not known what the impact of the additional parallel capacitance to the self-resonance frequency of the spiral planar coil will be. The computation of positive and negative mutual inductances is complex and lengthy. Hence a MATLAB script is written to compute these mutual inductances. The complete code is attached as Appendix D. Together with the self inductance (refer to (6.37)), the total inductance can be computed. Table 6.2 shows some of the results for spiral planar coil on different dimensions: (More discussion of the results.) HF Antenna for a New Wine Closure This section describes the design process for an RFID HF tag antenna for a special wine closure. This wine closure is a new invention from Zork [98]. It is made of plastic with a thin metal foil on top to optimise its functionality as a wine closure, which is to preserve the quality of the wine. Fig shows a model of the wine closure of interest. An RFID tag is confined in the space shown in the figure. The aims of embedding an RFID tag into the wine closure from the point of wine retailer include product authentication, tracking, and anti-theft and anti-counterfeiting. Page 114

131 Chapter 6 RFID Antenna Design Table 6.2. Inductance values for various spiral planar coil, with L = 20; L1=40; W =3; G=1 Turns L sel f M M + L Total (µh) Thin Metal Sheet Plastic Cover Plastic Stopper Space for Tag Figure A Zork wine closure model. This diagram is not to scale and does not represent an actual Zork wine closure. available for RFID tag. This diagram is to indicate the location of the spacing From the view of customers, they can scan a wine bottle embedded with an RFID tag in some specific area to obtain extra information about the wine they have just chosen, such as the the origin of the wine, ingredients, awards, and even recommended from wine critics around the world. The aim is to have a multi-turn coil antenna to provide inductance to resonate with the HF chip input capacitance. Besides the space limitation, it is expected that the thin metal foil will reduce the performance of the RFID tag. The final design of the HF tag antenna is as shown in Fig Table 6.3 shows the details and read range Page 115

132 6.8 HF RFID Antenna Design Figure RFID Tag for Wine Closure performance of the designed HF RFID tag for Zork wine closure. It should be noted that the maximum read Range is limited by the reader antenna size [84]. Table 6.3. Performance of HF RFID Tag for Zork Material Dimension (mm) Chip Reader Read Range (mm) Copper wire 25.0 (h) 7.5 (r) ISD72128 or C220 ISD L (Best) HF Antenna for a Pig Ear Tag This section describes the design process for an RFID HF tag antenna for a pig ear tag. The pig ear tag project is sponsored Pork CRC, Australia. The design of a UHF RFID tag antenna for pig ear tagging is presented in Section The pig ear tag is a relatively huge project, when as compared to the Zork wine closure tag antenna design. The common frequency used in livestock tagging is the Low Frequency (LF) band, spanning khz. LF RFID operation is standardised by ISO 11784/85 [49, 50]. Page 116

133 Chapter 6 RFID Antenna Design LF RFID system normally cannot handle dense tag environments. Also, LF requires large antenna components and hence is difficult to implement and is susceptible to electrical noise, which HF can handle [19]. Although a new enhanced version of LF chip is available [99] with anti-collision capability, the traditional LF tag which complies with ISO 11784/85 does not offer such anti-collision capability. Without anti-collision, a single LF tag must only be present in the reader antenna interrogation zone at any one time. For practical real life applications, animals are forced to pass by (or enter the interrogation zone) one by one. Where this is practicable is when identifying larger animals, such as cattle or sheep. Examples include cattle and sheep farming. However, in pig farming, a problem arises. Small baby piglets are smaller in size when compared to cattle. It is not easy to herd piglets through a gate mounted with an antenna. Hence it is the intention of this research work to investigate the feasibility of using either HF or UHF RFID tag for the tagging of pigs, by using the development of pig ear tag as the case study. Livestock ear tags for cattle deployed in Australia must be accredited by National Livestock Identification System (NLIS), Australia. Examples of these livestock ear tags with embedded RFID tag can be found in [100, 101], which is ISO 11784/85 compatible. These tags can be recovered at slaughterhouse and reused, and can withstand hazardous environment. A cattle livestock ear tag before any RFID tag is embedded is as shown in Fig. 6.19(a). One part of the ear tag has a hole in the middle to allow both parts to be pinned together on the ear of cattle livestock. The space to attached an RFID tag is hence restricted to a circular disc with a hole in the middle as shown in Fig. 6.19(b). The outer diameter of the space available is 28.3 mm, while the inner diameter is 12 mm. The initial design incorporated the following key features of an HF tag: A multi-turn loop to provide the required inductance. Allow as much space as possible in the middle of the coil to have a high inductance value. Page 117

134 6.8 HF RFID Antenna Design Part 1 Part 2 (a) (b) Figure (a) Two parts of a cattle livestock tag so that it can be pinned on the ears of cattle livestock. (b) The possible space for RFID tag allocation on a cattle livestock tag. An underpass and two via holes to complete the loop. The antenna has 6 loops on one side of a FR4 (16 mm thick) as shown in Fig The formation of the multi-turn loop is achieve by connecting several almost circular ring tracks together. In the simulation software, the circular shapes are formed by s polygon, which is visibly obvious in Fig Also, the connecting part is bent and is Pig1 - HFSSModel1-3D Modeler Thursday, February 15, 2007 not smooth. Figure First version of a pig tag with bend tracks to connect all the circular tracks together. An alternative way to model the multi-turn loop antenna is to use a spiral structure. A spiral multi-turn loop is shown in Fig There isn t any bend as compared to Page 118

135 Chapter 6 RFID Antenna Design Fig However, the simulation of a spiral model takes a comparatively longer time. Also, similar to circular shape, it is noticeable that the spiral track is formed by Pig2 - HFSSModel1-3D Modeler Thursday, February 15, 2007 polygon and the edges are not smooth. Figure Second version of a pig tag using spiral tracks. Hence, spiral is not chosen as the final choice for the modelling of an HF pig ear tag. Circular shapes are also avoided to prevent the polygonal representation. The solution is rather peculiar. Rather than using a polygonal circular shape, a higher edges polygon is defined. The choices of using high edges polygon can be seen in Fig. 6.22, which is also the figure Pig4 - HFSSModel1 for the - 3D Modeler first prototype HF pig ear tag. Thursday, February 15, 2007 Figure First prototype of a pig tag. Page 119

136 6.8 HF RFID Antenna Design After the simulation, the antenna was sent for fabrication. A serious problem was encountered during the fabrication. As shown in Fig. 6.23, both the gap between the coil track and the width of the track itself of the HF antenna are very small. This often results in coil track short-circuiting. Also, uneven track increases the loss resistance of the antenna. This will reduce the quality factor and reduce the performance of the tag. 1 2 Figure Problem of uneven track thickness when prototyping antenna with thin tracks: (1) Thinner tracks, (2) Thicker tracks. After a satisfactory antenna was fabricated despite the problem mentioned above, a simple test was carried out to check the resonance point of the fabricated antenna. A small capacitor was soldered onto the feed point of the antenna to represent the input capacitance of a HF RFID chip, as shown in Fig Figure First version prototype of a pig tag. No HF chip is attached. A capacitor of 100 pf is used instead to represent the input reactance of a HF chip. Page 120

137 Chapter 6 RFID Antenna Design The added capacitor had a capacitance value of 100 pf, the nearest value to the HF chip input capacitance of 100 pf (Section ), to mimic a HF chip. The tag, HF antenna with the added capacitor was then tested for its resonance frequency. It was then discovered that the resonance frequency was MHz. This shows that with slight modification (slight increment in inductance) of the HF tag antenna design shown in Fig. 6.24, the HF tag antenna will operate in the desired frequency of MHz. A suitable HF chip is the NXP-Phillips SL1 ICS31 01 ICODE1 Label IC [92] which has an input capacitance of 97 pf. It should be noted that the test on tag resonant frequency by substituting a HF chip with a capacitor is purely an early indication on whether the design approach is suitable or not, this is achieved by measuring the resonance frequency. A test using an actual HF chip is needed to confirm the functionality of the tag. As mentioned before in Section , there is no available HF chip straps which resemble the UHF chip straps. As discussed in Section , all the HF chips used in experiment are recycled chips from existing HF tags. Hence, it is not feasible to used recycled HF chips at the beginning of the designing cycle. After simulation and fabrication, an easy way to test the design is to use a tag chip with Tag Talk First (TTF) compatibility (ISD72128, refer to Section ). A tag (same design with the one shown in Fig attached with a tunable capacitor and a TTF HF chip is as shown in Fig Figure Pig Tag with TTF Compatibility Page 121

138 6.8 HF RFID Antenna Design Figure Testing a Pig Tag with TTF Compatibility: (1) To a signal generator, (2) To a reader antenna, (3) To a spectrum analyser. No RFID reader is required to test a TTF RFID tag. Once a TTF RFID tag is powered by a CW wave at the correct frequency, it will start generating signal, depending on the induced voltage at its input terminal (See Section for more details). Hence the testing environment as as shown in Fig. 6.26, where a signal generator, a spectrum analyser, an antenna and a three port directional coupler. The connection to the directional coupler is as followed: Out: Connected to Signal Generator. In: Connector to the antenna of HF reader. Cpl: Connected to spectrum analyser to monitor the response from a tag. This test shows a read range of approximately 0.2 m. The results prove that the tag is working as an RFID tag. No fine-tuning was carried out to optimised the performance of the tag as it was then decided to pick a HF chip with lower input capacitance. This Page 122

139 Chapter 6 RFID Antenna Design is due to the fact that the input capacitance value has great impact on the cost of a HF chip as the chip will need bigger area to provide the necessary capacitance. Hence it is most desirable to have a tag antenna design which can function on a lower input capacitance HF chip, such as the 23.5 pf input capacitance RF-HDT-SJME-G1 Tag-it HF-I transponder IC from Texas Instrument [91]. An HF antenna needs to provide sufficient inductance through its coil turns to resonate with the tag chip capacitance at the frequency of operation, which is MHz. In the case of a size constraint and reduced HF chip input capacitance, a HF tag would not have sufficient inductance, and extra capacitance would be added by making the HF antenna double sided and introducing capacitance through the two overlapping plates. To avoid the difficulties in fabrication (Uneven tracks as shown in Fig caused by the closely spaced tracks, the tracks are designed to be further apart as shown in Fig Simulation was carried out using Ansoft HFSS to predict the performance of the designed tag antenna. Pig88 - HFSSDesign1-3D Modeler Thursday, February 22, 2007 Figure Pig tag with external capacitance added. The simulation results are as shown in Fig The resonant point of the HF antenna alone is at approximately 40.2 MHz. The measurement results on a fabricated prototype antenna show a resonant point at the frequency of 41.3 MHz. Page 123

140 6.8 HF RFID Antenna Design 22 Feb 2007 Ansoft Corporation 18:10:02 XY Plot 1 HFSSDesign Y1 im(z(lumpport1,lum Setup1 : Sweep1 Y1 re(z(lumpport1,lum Setup1 : Sweep Y Freq [MHz] Figure Simulation results of To have a high capacitance value, it is desirable for the tag to be as thin as possible to reduce the distance between the overlapping coils and plates so that the capacitance between the parallel plate in the middle and also the capacitance between the upper and the lower rack of the HF coil are increased. The thickness of the FR4 was reduced from 1.6 mm to 0.8 mm. The simulation results are as shown in Fig The simulated resonant point is at approximately 29.2 MHz. The measurement results on a fabricated prototype antenna show a resonant point at the frequency of 28.6 MHz. 23 Feb 2007 Ansoft Corporation 16:13:05 XY Plot 1 HFSSDesign Y1 im(z(lumpport1,lum Setup1 : Sweep1 Y1 re(z(lumpport1,lum Setup1 : Sweep Y Freq [MHz] Figure Impedance of second version of a dual frequency antenna. Page 124

141 Chapter 6 RFID Antenna Design Several simulations, antenna fabrications and testing were carried out to optimise the performance of the tag. The final design of the HF antenna is shown in Fig The thickness of the substrate has been reduced to 0.2 mm. Also, the coil tracks separation cannot be increased due to size limitation. The chip that was used was the Texas Instrument Tag-it HF-I Standard Transponder IC [91], which is based on the ISO/IEC and ISO/IEC standards, and has an nominal input capacitance of 23.5 pf Figure Final version prototype of a pig tag: (1) Multi-turn coil to provide inductance, (2) TI HF chip, (3) Tunable capacitance plate for fine tuning. The front and the back of the un-tuned final design was as shown in Fig Pig110 - HFSSDesign1-3D Modeler Tuesday, January 09, 2007 Pig110 - HFSSDesign1-3D Modeler Tuesday, January 09, 2007 (a) Front view (b) Back view Figure HF RFID Tag to be embedded in a livestock ear tag. For testing purposes, the used ID ISC.LR2000 HF reader and ID ISC.ANT300 reader antenna, both from FEIG Electronics were used. The width and length of the reader antenna are both approximately 0.33 m. In HF operation, the maximum possible read Page 125

142 6.9 UHF RFID Antenna Design range is proportional to the size of the reader antenna [84]. As a rule of thumb in real life deployment, the maximum read range is approximately equals to the dimension of the reader antenna. We obtained a read range of 0.34 m in free space and of approximately 0.32 m when attached to a human hand, which simulates the environment where a tag is attached to the pig s ear. To increase the read range, we can increase the size of the reader antenna in a way presented in [84]. (Leader Product Encapsulation to be added.) (Actual Encapsulations process and field testing to be added.) Part of this section has been published in Investigation on the deployment of HF and UHF RFID tag in livestock identification [102]. 6.9 UHF RFID Antenna Design This section presents the design of a UHF RFID antenna. A UHF RFID tag operates within 860 MHz to 960 MHz around the world. A UHF RFID tag antenna is designed to have a conjugate match with a UHF RFID chip to allow maximum power transfer. Firstly, this section discusses the common commercial UHF RFID tags and their common features. In the later part, various novel UHF antenna designs for various applications are shown. UHF tag as discussed before are very susceptible to the environment. Hence, the design of UHF RFID tag antenna is very diverse, largely depending on its application. There are numerous research papers on the design UHF antenna for various applications. [103] The design of UHF RFID antenna is not of the main research interest of the author. As it can be seen later, most of the designs of UHF tag antenna come from joint researches. However, it is an important learning process as the design of a dual frequency antenna in Chap. 7 will require the design of UHF RFID tag antenna. Page 126

143 Chapter 6 RFID Antenna Design Common UHF RFID Tag A common commercial UHF RFID Tag has very distinctive features. It is either a dipole, or a small variation of a dipole, and has a simple matching network, most commonly a matching network very similar to a T-Match is used to transform the capacitive nature of a dipole into inductive to conjugately match to the input impedance of a UHF RFID tag chip. Also, a common UHF RFID tag is very thin and bendable. It is readily applied to a pallet in supply chain. It has a read range of more than 10 m using the air interface specified by EPCglobal C1G Simple Planar Dipole This section presents the fundamental of a simple planar dipole. Dipole or electric dipole is the counter part of a loop antenna or magnetic dipole. As suggested by the name, an electric dipole antenna is sensitive to electric field while a magnetic dipole antenna (loop antenna) is sensitive to magnetic field. As a common name, an electric dipole is often known as dipole while a magnetic dipole is known as loop antenna. A dipole and a loop antenna has very similar radiation pattern (doughnut shape) when aligned in the correct way. A half wave length dipole is theoretically purely resistive (zero in reactance). However in actual implementation, a half wave dipole has to be cut from 4-10 percent shorter than a half wavelength to present a purely resistive impedance to the feeder. From [75], input impedance of a half wave length dipole, Z in, is equal to 73 + j42.5ω with a directivity of In theory, as mentioned above, a half wave dipole is purely resistive. A dipole shorter than that is always capacitive. Increase the size of dipole pass the pure resistive point will push into inductive Smith Chart region. A inductive dipole is always longer than half wave length without any matching network. The UHF band for RFID operation around the world is from 860 MHz to 960 MHz. A half wave length dipole for this frequency band is between m and m. A Page 127

144 6.9 UHF RFID Antenna Design typical UHF RFID tag is normally around 0.1 m in length, and hence, a dipole without any matching network is not feasible to be used as a UHF RFID tag antenna. Some of the popular matching network for a dipole is the Omega and Gamma matching. However, a Omega matching network can only match loads less than 50 Ω resistive by stepping the resistance up. It can not step or transform antenna feed resistance downwards when using capacitors. Also, both Omega and Gamma matching networks cannot match capacitive antenna loads. A variation of dipole is the folded dipole. The directional characteristics of a folded dipole are the same as those of a simple dipole. However, the reactance of a folded dipole varies much more slowly as the frequency is varied from resonance. Because of this the folded dipole can be used over a much wider frequency range than is possible with a simple dipole. For RFID application, dipole is always in the form of planar dipole UHF antenna for Sheep This section presented a UHF antenna for sheep tagging. This section is an extract from a co-authored paper A small passive UHF RFID tag for livestock identification [104]. Similar to pig tagging as discussed in Section 6.8.4, animals such as cattle and sheep are tagged for purposes such as disease control, breeding management and also stock management. Also similar to the pig tagging, where a passive HF RFID tag is attached on a pig s ear, sheep tagging uses a small passive UHF RFID tag to be attached on a sheep s ear. The current RFID technology for livestock tagging, including both pig tagging and sheep tagging, operate in the low frequency (LF) region [100]. However, as discussed in Section 6.8.4, LF tags can only be read at close range and may not perform well when multiple tags are simultaneously present in the interrogation field. It is believed that the HF RFID tag designed for pig tagging as presented in Section can be used in sheep tagging as well. However, the focus of the research of sheep Page 128

145 Chapter 6 RFID Antenna Design tagging is on UHF operation. As compared to HF tags, UHF tags not only give better read range, but also support higher data rates. The design of UHF RFID sheep tag is based on a loop antenna, contrary to the common dipole built of UHF RFID tag. The main reason is that a loops antenna can be smaller in overall dimension as compared to a dipole antenna with equivalent performance. A quick test confirm that an electric dipole of approximately same size to difficult to be conjugately matched a UHF RFID chip. Also, a loop antenna does not have sharp edges and can be fitted on the sheep s ear relatively easily. This sheep tag is designed to operate in the frequency of 915 MHz. The UHF chip used is the Alien C1G1 while the reader used is the Alien UHF RFID reader. The front and rear view of the final product is shown in Fig The matching network is provided by the copper strips on the rear side, providing series and shunt capacitance to the circuit. This copper strips can be trimmed during experiment to fine tune the circuit so that the antenna will resonate at the desired frequency, which is 915 MHz. This antenna is designed to operate in the USA, but with a simple tuning (trimming of copper strips on the rear side of the antenna), it can operate in Australia while conforming to all the local regulations. The target impedance of the designed antenna is j133.44ω, as compared to the chip impedance of j149.2ω at 915 MHz. The maximum measured read range was about 1.42 m. However, as predicted by the radiation pattern, when the tag was aligned with its axis parallel with that of the interrogator antenna, the reading distance was only a few centimetres. A simple example on the deployment of this tag for sheep tagging is to locate the reader antenna on top of a door arch where tagged animals pass through, with the direction of antenna facing down UHF antenna for Pig This section is based on a published paper Investigation on the deployment of HF and UHF RFID tag in livestock identification [102]. Page 129

146 6.9 UHF RFID Antenna Design Figure (a) Front view; (b) Rear view of the sheep ear tag Two different UHF tag designs have been considered in our study. Both UHF tags have been designed for operation at 923 MHz in Australia. The tag chips used for both tags are from Alien Technology [105], and have impedance of approximately 20 j141 Ω at 923 MHz. The first UHF tag design consists of a circular loop antenna with a two element matching network to match the impedances of the tag antenna and tag chip. More detailed design steps of this tag can be found in [104]. However, the tag has been made slightly larger in diameter compared to that presented in [104] to suit the application studied in this paper. The structure and dimensions of this tag is shown in Fig. 6.33(a). The tag is made using a thin FR4 board with substrate thickness h = 0.36 mm and relative dielectric permittivity ε r = 4.4. This tag design utilizes both sides of the FR4 board. From Fig. 6.33(a), the front view is the circular loop antenna and the back view is the implementation of the matching network that consists of a series and a shunt capacitor. The tag antenna structure is simulated using Ansoft HFSS and the simulated impedance is Page 130

147 Chapter 6 RFID Antenna Design 1 mm substrate series capacitor 25 mm x y chip copper strip 2 mm 13 mm shunt capacitor FRONT BACK (a) Tag 1 (b) Tag 1: radiation pattern 2 mm 28 mm x y 0.5 mm 13 mm chip (c) Tag 2 (d) Tag 2: radiation pattern Figure Two UHF pig tags j148 Ω, which is the conjugate of the tag chip impedance. The radiation pattern of the antenna is as shown in Fig. 6.33(b), with maximum directivity of 1.65 db. The second UHF tag design considered consists of a curved electric dipole antenna with an inductance track across the dipole for impedance matching purpose. The structure and dimensions of this tag is shown in Fig 6.33(c). This tag is made using a FR4 board with substrate thickness h = 1.6 mm and relative dielectric permittivity ε r = 4.4. This tag design is single-sided. The tag antenna structure is simulated using Ansoft HFSS and the simulated impedance is 1 + j152 Ω. As can be seen, the impedance of the tag antenna structure is not exactly the conjugate of the tag chip impedance and hence, the tag antenna and chip impedances are not perfectly match. However, the inductance track across the dipole does provide sufficient inductance to tune with the capacitance of the tag chip. The radiation pattern of the antenna is as shown in Fig. 6.33(d), with maximum directivity of 1.83 db. As can be seen, although the general shape of the radiation pattern of the first and second UHF tags is almost the same, the maximum directivity of these two tags occur at different directions with respect to the Page 131

148 6.9 UHF RFID Antenna Design Table 6.4. Read Range for UHF Tags UHF Tag 1 UHF Tag 2 Tag in free space (m) Tag attached to hand (m) plane of the tags. Both UHF tags are tested and the read range results are shown in Table UHF antenna for Beer Keg This section presented a UHF antenna for a Beer Keg. This section is an extract from a co-authored paper A Small Passive UHF RFID Tag for Metallic Item Identification [106]. The tag presented here is not only for the deployment of RFID in beer keg tagging. The main purpose is to illustrate how RFID tag can be designed on a metallic surface. The idea of this tag design is to fully exploit the fact that magnetic field is doubled at a metallic surface. The loop antenna is oriented in a way to coupled to the magnetic filed at a metallic surface. The final design is as shown in Fig Tag antenna Tag chip Width (W rec) Length ( Lrec) Height (H rec) z x y Figure Structure of the RFID tag with a rectangular loop antenna. The simulated antenna impedance is as shown in Table 6.5. The read range of 0.83 m is obtained when a reader is operating at reduced power. It is expected the tag can be read up to 1.3 m when a reader is operating at maximum allowable power. Page 132

149 Chapter 6 RFID Antenna Design Table 6.5. Read Range for UHF Tag for Beer Keg Antenna Impedance Peak Directivity Free Space j91ω 1.5 db Above Metal j91ω 6.5 db UHF antenna for Wine Cork This section describes the design of a UHF antenna for a wine cork. This wine cork of interest is the same as presented in Section and illustrated in Fig Figure RFID Tag for Wine Cork The functionality of a UHF tag for wine cork is also the same with the a HF tag for wine cork as discussed in Section The final design is as shown in Fig and the details and results of actual testings is presented in Table RFID Tag Fabrication An RFID tag is the combination of an RFID tag antenna and an RFID tag chip, where the tag chip will be locate across the input terminal of the RFID tag antenna. For prototyping, the raw material for RFID tag antenna is the Printed Circuit Board (PCB). An example is the commonly used FR4 board. There are many types of PCB Page 133

150 6.10 RFID Tag Fabrication Table 6.6. Performance of HF RFID Tag for Zork Material Dimension (mm) Copper strip 6.0 (h) 7.0 (r) Chip Alien Class 1 Generation 1 Reader Alien ALR-9780 Read Range (mm) 300 boards used in the prototyping of the design RFID tag antenna. For the conductive layer, copper and aluminium is the most common. For the dielectric layer, FR4 is the most common. There is a selection of choices for the thickness of the dielectric layer. The commonly used one is 1.6 mm and 0.2 mm. Also, the thickness of metallic layer can be chosen within a few common values. Most of the time, two sided PCB boards are needed for most of the antenna design. For the HF coil antenna, the bottom layer is needed for the underpass, which is link to the top layer through via holes. Via holes are created using rivet if a pad size of at least 1.3 mm in diameter is possible. If due to space constraint and no pad with at least 1.3 mm is available, a via hole is created by first drilling a hole using mini drill and a fine conductor is passed through the hole and soldered at both ends. For the case of UHF antenna, the bottom layer is always used for matching purposes. The common way is to have the bottom layer providing series or parallel capacitance. No physical connection is needed between the top and bottom layers. A typical HF coil antenna, with low fabrication cost, is a planar spiral strip antenna on a dielectric slab, as shown in Fig The dielectric slab chosen to be use in this experiment in making the HF antenna is the FR4, with relative permittivity of 4.4. To attach a tag chip, 3M TM Z-Axis Electrically Conductive Tape 9703 is used. As suggested by its name, this conductive tape only conduct in the Z-axis. It has a low contact Page 134

151 Chapter 6 RFID Antenna Design Figure Typical HF Antenna: Planar spiral in shape on a dielectric slab, normally FR4. resistance. However it is not very clear from the specification sheet what is the capacitance it may introduce to the circuit. A simple experiment was carried out to test the effect of this conductive on a RFID tag. As shown in Fig. 6.37, a small part of the HF coil antenna is removed. A small copper tape is then attached using the Z-axis conductive tape to replace the removed HF coil track. A measurement using a Gemplus L120 reader shows that there is no degradation in performance. Hence it is concluded that the Z-axis conductive tape has minimal impact on the RFID tag antenna. Figure Effect of using Z-axis Conductive Tape on HF RFID Tag 6.11 Antenna Design and Simulation Page 135

152 6.11 Antenna Design and Simulation Ansoft HFSS High Frequency Structure Simulator (HFSS) is a 3D Electromagnetic-Field Simulation for High-Performance Electronic Design from Ansoft. This section explains the setup for all the simulations carried out for RFID antenna design, for both the HF and UHF cases. This section is based on the HFSS User Manual from Ansoft [107]. Ansoft HFSS uses finite element method (FEM) as the solver of its simulation. There are other types of solver, which include the popular finite-difference time-domain (FDTD) and method of moment (MOM). This thesis does not investigate the difference between these solvers. A good description of different solvers or computational electromagnetic techniques can be found in [108]. Below are the common HFSS settings for all the simulations carried out for antennas presented in this thesis: Solution Type: Driven Modal Solution For the calculation of model-based S-parameters of passive, high frequency structure (including microstrip, transmission line). Driven Terminal Solution For the calculation of terminal-based S-parameters of multi-conductor transmission line ports. Eigenmode Solution For the calculation of eigenmodes and resonances of a structure. For the purpose of antenna simulation, the Driven Modal Solution is the most appropriate choice. Excitation Point Excitations in HFSS are used to specify the sources of electromagnetic fields and charges, currents, or voltages on objects or surfaces in the design. Page 136

153 Chapter 6 RFID Antenna Design There are several types of excitation points in HFSS. The definition given by Ansoft for the two common types are listed below: 1. Wave Port Represents the surface through which a signal enters or exits the geometry. 2. Lumped Port Represents an internal surface through which a signal enters or exits the geometry. The difference between a wave port and lumped port is subtle. Both are represented by a geometrical square or rectangle shape as an opening, which allows a signal to enter or exit a structure (antenna in our case). As suggested by the name, wave port is often used to represent a waveguide or a transmission line and is characterised by an characteristic impedance. In simulation a wave port has to be defined at the edge of a structure. On the other hand, lumped port can be characterised by a user defined complex impedance and can be located internally. It is normally used for microstrip structure. Almost all of the antenna designed and presented in this thesis are fabricated on PCB boards and have good resemblance with microstrip structure. Hence, lumped port is chosen. The impedance of lumped port is set as 50 Ω instead of the impedance of the RFID chip used. A quick test shows that a change in the impedance of the lumped port will affect the simulation results. For example, the input impedance of the antenna will change from 20 + j170 at 950 MHz to 25 + j182 at 950 MHz if the lumped port impedance is changed from 15 Ω to 50 Ω. The reason a 50 Ω lumped port is chosen for simulation is that the network analyser used for input impedance measurement is a 50 Ω system. It will make the comparison between measurement results and simulation results more meaningful and more accurate. More in-depth discussion on measurement techniques and challenges is presented in Chap. 8. Page 137

154 6.11 Antenna Design and Simulation Also, experience shows that simulation using a capacitive lumped port (For example 5 j150ω to represent an capacitive UHF RFID tag) will results in much longer simulation time. The disadvantage of using a 50 Ω lumped port is that simulation results based on a 50 Ω lumped port may not be reflecting the actual situation when a UHF RFID tag is connected to the antenna. To overcome this short-coming, most RFID tag antennas are designed to have a structure with tuning flexibility. Since the difference in impedance is not significant, a quick fine-tuning can optimise the performance of a RFID tag. Boundary Condition Similar to excitation types, there are several options for setting up a boundary for a simulation. A boundary is required to specify the region of the problem. Without a boundary, HFSS would not be able to confine a problem to a reasonable size. The two common types of boundary for antenna simulation is as follow: 1. Radiation Boundary In HFSS, radiation boundaries are used to simulate open problems that allow waves to radiate infinitely far into space, such as antenna designs. HFSS absorbs the wave at the radiation boundary, essentially ballooning the boundary infinitely far away from the structure. 2. Perfectly Matched Layers (PML) Perfectly matched layers (PML) are fictitious materials that fully absorb the electromagnetic fields impinging upon them. These materials are complex anisotropic. In HFSS the radiation boundary must be at least quarter of wavelength away for the radiating structure of interest. Also, a radiation boundary must be convex with respect to a radiating source. It poises a lot of difficulties for the simulations of antennas presented in this thesis. For example, at HF (13.56 MHz), a quarter of wavelength is approximately 5.53 m. The antenna size is often in the size of up to 0.1 m. This means in the simulation, the antenna must be located in a huge empty region, and the simulation duration will increase significantly. Page 138

155 Chapter 6 RFID Antenna Design One may argue that a RFID HF antenna operates through inductive coupling and not radiation. Hence a small radiation boundary is sufficient. However, HFSS computes the power entering an antenna of interest and the power reflected from the antenna, in order to compute the impedance of the antenna. With a radiation boundary smaller than the recommended quarter wavelength, more power will be reflected, reducing the accuracy of the simulation. One can compare a simulation to an actual measurement. Having a radiation boundary in simulation is comparable to having absorbance material in an anechoic chamber for a measurement. The absorbance material must be a certain distance away depending on the frequency of operation of the antenna under test. Also, radiation boundary must be convex in shape with respect to the radiating source. It is easy to fulfill this requirement if the antenna under test is a simple loop antenna or a simple dipole. For a more complex antenna with unsymmetrical or irregular shape, it is very difficult to construct an appropriate radiation boundary. Project31 - HFSSDesign1-3D Modeler Friday, May 11, 2007 Figure PML Layer in HFSS: The larger cubical in the middle defines the working zone, and is surrounded by PML layers Scripting in HFSS Although HFSS working environment allows 3-D modelling and HFSS supports parametric optimisation, to construct spiral planar coil structure of many turns (> 4 turns) using HFSS is very cumbersome. The latest version of HFSS (version 10) supports Page 139

156 6.11 Antenna Design and Simulation modelling using mathematic equation. However, HF coil antenna is not readily defined mathematically. The alternative is to scripting function in HFSS to create HF coil. The comprehensive HFSS guide is provided by Ansoft [109] and Virtual Basic Script (VBS) is used as the programming language in the scripting of HFSS. However, VBS itself is lacking of analytical tools. To write a VBS programme from stretch is time consuming. Hence, it is decided to use MATLAB as the analytical tools. MATLAB and HFSS are linked together to produce the desirable results. MATLAB will compute the geometry of the structure using mathematic equations and compile HFSS script using Virtual Basic Script (VBS). The link between MATLAB and HFSS is shown in Fig below: MATLAB Script generate VB Script run HFSS Simulation export result Figure Linking MATLAB and HFSS: MATLAB script is written to generate VB script. The VB Script is then fed into the HFSS simulator to construct the antenna model and and set up simulation parameters. The simulated results are then exported from HFSS to MATLAB for analysis and plotting. The foundation of the work is largely based on the API produced by V. C. Ramasami [110]. Extensive Extension and modification of the API has been made to suit the work of the author. The complete MATLAB script can be found in Appendix C. This MATLAB script defines the function makecoil, which takes in the following parameters: turn: This is to specify the number of turns of the coil antenna. L: The length of the inter-most coil. L1: The width of the inter-most coil. Page 140

157 Chapter 6 RFID Antenna Design thick: The thickness of the copper film on top (or other conductive material). It is normally set as m. height: The thickness of the dielectric board. Normally a FR4 board of 1.6 mm thickness is used. w: The width of the HF coil track. g: The gap between the coil tracks. A gap of 0 mm will results in a rectangular patch as all the coil track will be merged together. save loc: The location and the filename of simulated results. The files will contain both the S 11 and the input impedance of the simulated antenna with respect to frequencies. An example of an automatic generated HF coil model is as shown in Fig testbox - testbox - 3D Modeler Thursday, May 10, 2007 Project31 - testbox - 3D Modeler Thursday, May 10, 2007 (a) 2-loop Coil (b) 10-loop Coil Figure The auto generated loop using VBS. A comparison between Table 6.2 and Table 6.7 shows that when the number of turns is small (< 6 turns), the simulated results is very close to the computed results. However, as the number of turns increases, the difference between these two results increases. In simulation, the impedance of the antenna is computed, not the inductance value. As an approximation, the impedance is transformed back to the inductance value, using Z = 2π f L. Page 141

158 6.11 Antenna Design and Simulation Table 6.7. Inductance values for spiral planar coil with different number of turns, with L = 20; L 1 =40; W=3; G=1. Turns Reactance at MHz (Ω) L (µh) Plotting in HFSS HFSS can provides results in several forms. The most common results obtained are the input impedances of the simulated results. In the simulation of HF RFID antenna, resonance frequency of an antenna can be observed from the plot of real and imaginary parts of the simulated impedances. At resonance the the real part of the impedance will increase towards infinity and the imaginary part will be zero. In the simulation of UHF RFID antenna, input impedance can be presented in normal plot of impedance versus frequency or Smith Chart. Also, in the simulation of UHF RFID antenna bandwidth is also very important. To obtain the bandwidth of the antenna itself is easy, by using a 50 Ω lumped port for simulation, and plot the S 11 of the simulation results. However, this S 11, as shown in Trace 1 in Fig. 6.41, does not offer any insight on the bandwidth of the RFID tag. The bandwidth of the tag is affected by both the antenna and the chip attached to the antenna. Normalising the simulation results using the target impedance of the designed antenna is one easy way. HFSS has a built in post processing function to normalise results based Page 142

159 Chapter 6 17 May 2007 Ansoft Corporation 15:25:19 XY Plot 3 HFSSModel RFID Antenna Y1 Design db10(p_lost) Setup1 : Sweep1 Y1 db20(s(lumpport1,l Setup1 : Sweep Y Freq [GHz] Figure Plots in HFSS (Y-axis in db): Trace 1: S 11 ; Trace 2: Eq. (6.21) in db term. on a new port impedance value. The arithmetic for the normalising process can be found in [107]. An example is shown in Fig Trace 1 is showing the S 11 plot after the post processing normalisation. 21 May 2007 Ansoft Corporation 14:55:34 XY Plot 3 HFSSModel Y1 db20(p_lost) [db] Setup1 : Sweep1 Y1 db20(s(lumpport1,lumpport1)) [db] Setup1 : Sweep Y Freq [GHz] Figure Plots in HFSS (Y-axis in db): Trace 1: Normalised S 11 ; Trace 2: Eq. (6.21) in db term. However, the problem with this method is that it assumes a constant RFID chip input impedance, which in fact is frequency dependent. Also, the arithmetic of normalisation results in very low S 11 value and it is very difficult to determine the bandwidth. Page 143

160 6.12 Conclusion The solution is to use (6.21) as the foundation in determining the bandwidth of an RFID tag. Firstly, from Section, an equivalent circuit for a UHF RFID tag chip consists of a resistor in parallel with a capacitance. Using this equivalent circuit, the resultant input impedance of a UHF RFID tag chip within a frequency band can be computed. Note that the input impedance changes as the frequency increases. By substituting this computed input impedance of the chip and also the simulated antenna impedance from HFSS into (6.21), the ratio between the power loss and the power available can be computed. This ratio plotted in db term is shown as Trace 3 in Fig The bandwidth is defined as the half power point, which is -3 db point ISO-Pro ISO-Pro is a software from T-Tech [111] to control the T-Tech s quick prototyping system. T-Tech s quick prototyping system is actually a milling machine. It mills away unwanted metal from a PCB, and the remaining metal forms a designed antenna. The 3-D antenna model is drawn in HFSS for simulation. The same simulation model is used to avoid extra work to redraw the designed antenna in Protel (Protel is a PCB layout software, which can export a designed layout to ISO-Pro with ease). From HFSS, the antenna model is exported in the format of AutoCAD DXF format. However, a DXF format is a 2-D format. Hence, if a model has several layers, several DXF files are needed Conclusion This chapter presents the fundamental of HF and UHF tag designing, novel HF and UHF RFID tag antenna designs for new applications, antennas modelling, simulation and prototyping,and tag fabrication process. This provides a strong foundation to advance into the design of dual frequency antenna for RFID application, which is essentially a single antenna operating as both a HF antenna and a UHF antenna. Page 144

161 Chapter 7 Dual Frequency RFID Antenna W ITH good read range and fast data rates, UHF Radio Frequency Identification (RFID) is being deployed in the supply chain to uniquely identify each item. However, UHF transmission power is easily absorbed by ionised liquids, such as water. Hence, normally, in a environment with lots of liquid products, an HF RFID system is chosen instead. This chapter investigates the feasibility of designing a dual-frequency (HF and UHF) RFID tag antenna, with frequency ratio of up to 70, to embrace the benefits offered by both the UHF and HF RFID systems. Also, this chapter shows that with careful design, a single feed dual-frequency RFID tag antenna can be achieved. Prototypes of the dual-frequency RFID tag antenna are presented in this chapter, with validating simulations and measurement results. Page 145

162 7.1 Introduction 7.1 Introduction Radio Frequency Identification (RFID) is a technique used to identify objects by means of electromagnetic waves. An electronic code responding label, also known as a tag, consists of an antenna and an integrated circuit. Upon receiving any valid interrogating signal from any interrogating source, such as an RFID reader, the tag will respond according to its designed protocol. RFID has been recently applied into supply chain tracking systems, revolutionising the conventional ways of how an object is traced along the supply chain and how an inventory system works. Not only does RFID bring benefits in saving costs, it also has the potential of increasing security levels along the supply chain, especially in genuine goods authentication. To reduce the cost of RFID deployment, passive tags are used. There are 4 common bands for RFID applications; these are the LF band (less than 135 khz), the HF band (13.56 MHz), the UHF band ( MHz) and the so-called microwave (2.45 GHz). LF is not suitable for supply chain deployment, as to uniquely identify every object in the supply chain, each tag has to bear a unique EPC number, which must be at least 64 bits in length [18]. A low frequency operation, such as in the LF band, would suffer very slow reading in a heavily populated tag environment. Also, LF requires large antenna components and hence difficult to implement and is susceptible to electrical noise, which HF can handle [19]. UHF and microwave tags can offer comparatively very fast reading, but their performance will suffer more than the other bands described above in the presence of liquid or metal [17]. It is very difficult to conclude whether HF or UHF is better for RFID application in supply chains. This is because in any supply chain, there are many different scenarios, and in some HF is better and in the others, UHF outperforms HF operation. Table 7.1 summarises comparison between HF and UHF as presented in [17, 23, 13]. To embrace both the advantages contributed by HF and UHF, this chapter investigates the feasibility of designing a dual-frequency antenna for an RFID tag. The next section reviews current technology on dual-frequency antennas, and their shortcomings in designing an RFID HF and UHF dual-frequency antenna. Design aims ensuring Page 146

163 Chapter 7 Dual Frequency RFID Antenna Table 7.1. HF versus UHF in RFID Operation HF UHF Read Range 1 m Up to 10 m Cost (in large volume) Medium Low Read Rate Low High Metallic Area Bad Bad Liquid Surrounding Better Worse that the antenna is feasible for real life application are presented in Section 7.3. Section 7.4 describes the design process in detail, including the problems encountered and the solution of those problems using simulation results. Section 7.5 presents the actual measurement results for the designed dual-frequency antenna, together with some discussions and comments. 7.2 Current Dual-Frequency Antenna Design The intended dual-frequency RFID antenna is planned to function at MHz (HF) and also within the UHF band for RFID operation ( MHz). If using the highest frequency in the UHF band, which is 960 MHz, the frequency ratio for this dualfrequency antenna is slightly more than 70. Wong has done extensive research on dual-frequency microstrip antennas, with frequency ratio ranging from [112]. However, the frequency ratio of this technique is too low to be applied on the intended RFID dual-frequency antenna. A common aperture, dual-feed dual-frequency antenna for 900 MHz and 60 GHz was presented by Menzel [113]. The Frequency ratio achieved is approximately 70. However, the antenna has dual feeding points. One of the antennas which has a high frequency ratio and used in RFID is the one produced by IPICO [114]. Under the IPICO dual-frequency system, the RFID reader transmits a signal using the LF spectrum (125 khz to 135 khz) in order to power the Page 147

164 7.3 Design Aims tag and the tag uses the HF spectrum (3 to 30 MHz) to transmit its signal back to the reader. The tag produced by IPICO is using LF and HF, while the aim of this chapter is to combine both HF and UHF. 7.3 Design Aims The aim is to design, simulate and fabricate a dual-frequency RFID antenna, which can demonstrate the following characteristics: 1. Antenna impedance is equal to the complement of the input impedance of the RFID chip at UHF operation. This is to ensure a matching network to have maximum power transfer from the antenna to the RFID chip. The RFID chip used in the fabrication process has an approximate input impedance of jΩ at RFID UHF band. Hence the final antenna design must have an input impedance of approximately j Ω. In actual fact, the chip impedance will change within the RFID UHF band, but in this chapter the UHF antenna is designed to be matched at the highest RFID operating frequency (960 MHz). We believe that if an antenna can work at the highest frequency (having the highest frequency ratio), we can also tune the antenna to operate at the lowest frequency (860 MHz). 2. A resonance point at HF. It is expected that a dual-frequency antenna will behave like a parallel resonance circuit at HF. To ensure maximum power transfer, the antenna should have a resonance point at MHz, and a high quality factor. 3. A single feed antenna. The final antenna design is planned to be attached to commercially available RFID chips. Hence a single feed antenna is desired to avoid any modification required on the currently available RFID chips. 4. Reasonable antenna size and cost. Page 148

165 Chapter 7 Dual Frequency RFID Antenna The material used for fabrication is FR4, with relative permittivity, ε r, equal to 4.4. The antenna area should not exceed mm 2. However, it is not the focus of this chapter to minimise the size of the antenna. Hence, the final design may not be of the smallest possible size. 7.4 Dual-Frequency Antenna Design The basic idea in making a dual-frequency antenna (HF and UHF), is to design HF and UHF antenna separately as in Section 7.4.1, and merge these two antennas together with a single feed. However, since these two antennas are joined together in parallel, the UHF antenna will affect the HF operation of the HF antenna, and vice versa. A few experiments were carried out to verify this merging idea as presented in Section and Section Independent HF and UHF Antenna Design For the operation of HF, a multi-turn planar loop antenna is designed. The calculation of the inductance of a multi-turn planar loop antenna is done using [97], which includes the computation of the positive and negative mutual inductances. From 1 f 0 = 2 π, with an intended resonance frequency of MHz, we know that a LC large inductance in required. The problem is at the later stage, when this loop antenna is merged with the UHF dipole, the resonance frequency will change. Hence, the number of turns of the multi-turn loop is not fixed at this stage, and will only be finalised during the fine-tuning stage after the merging of the HF and the UHF antennas. An example of a generic multi-turn planar loop antenna is as shown in Fig For the operation of UHF, a dipole is used. An example of a generic half wave-length (of 960 MHz) dipole is designed as shown in Fig Page 149

166 7.4Merge Dual-Frequency - HFSSModel - 3D Modeler Antenna Design Monday, February 26, 2007 Merge - HFSSModel - 3D Modeler Figure 7.1. A generic HF coil antenna. Monday, February 26, 2007 Figure 7.2. A generic UHF dipole antenna Quick Feasibility Test For proving the idea of merging a HF antenna together with a UHF antenna, a HF antenna is merged with a UHF antenna in HFSS to observe the feasibility and to form the Page 150

167 Chapter 7 Dual Frequency RFID Antenna foundation of building a dual frequency RFID tag antenna which will operate according to specifications. Generic models from Section are used. A simple four-loop planar coil antenna is used as the HF antenna and a simple dipole is used as the UHF antenna. Two via holes and a underpass is required for the connectivity of the simple four-loop planar coil antenna. The model of this dual frequency antenna is shown in Fig The width of the track is arbitrarily chosen as 3 mm with the gaps between the HF planar coil loop chosen as 1 mm. The length of the transmission line is 80 mm, with the HF antenna connected to one side, and the UHF antenna on the other side. This length includes the 3 mm track of both the HF and the UHF antennas at the junctions where these two antennas are connected to the transmission line. The value is chosen to be approximately λ 4, which is 78 mm (for a frequency of 960 MHz). Although the permittivity of the substrate will affect this length, it is not considered at this stage, but will discussed in details in later stages. The length of dipole including the gap is 132 mm. The choice of length is based on the fact that a practical half-wave dipole is shorter than its theoretical half-wavelength which is 156 mm (at a frequency of 960 MHz). Again, the effect of the substrate is not taken into account at this stage. This initial simulation of a dual frequency antenna with the proposed method of merging two different antennas together shows promising results. We obtained a resonant at HF, though not exactly at the frequency we want and also, the impedance characteristics at UHF can be easily matched for the frequency we want. With fine-tuning, we believe we can obtain a dual frequency antenna which operates at the bands we want Tunability Test The second design is basically the first design with slight modification to test its tunability. In other words, we want to see how the characteristics of the antenna change when slight modification are applied. Page 151

168 7.4 Dual-Frequency Antenna Design Merge - HFSSModel - 3D Modeler Tuesday, February 06, 2007 Ansoft Corporation Figure 7.3. First version of a dual frequency antenna. 19 Feb :21:29 XY Plot 3 HFSSModel Y1 im(z(lumpport1,lum Setup2 : Sweep2 Y1 re(z(lumpport1,lum Setup2 : Sweep Y Freq [MHz] Figure 7.4. Impedance of the first version of dual frequency antenna at HF. In this part, the dipole is bent to reduce of the overall size of the dual frequency antenna. Also, the length of the transmission line linking the HF and UHF antennas is varies (10, 20, 30, 40, and 80 mm), to observe the changes in impedance. An example of the simulation model used is as shown in Fig. 7.6 where the dipole is bent and the length of the transmission line is 10 mm. An example of the graph showing the results obtained through simulation are shown in Fig. 7.7, which corresponds the simulation of the model in Fig Prediction of Impedance Page 152

169 Chapter 7 19 Feb 2007 Ansoft Corporation 18:17:30 XY Plot 1 HFSSModel Dual Frequency RFID Y1 Antenna im(z(lumpport1,lum Setup1 : Sweep1 Y1 re(z(lumpport1,lum Setup1 : Sweep Y Freq [MHz] Figure Merge HFSSModel1 Impedance - 3D Modeler of the first version of dual frequency antenna Tuesday, February at 06, 2007 UHF. Figure 7.6. Second version of dual frequency antenna. Since the HF and the UHF antenna are connected in parallel, it might be possible to predict the overall impedance if the individual impedance of both the HF and UHF antennas are known. The idea is to assume the antenna impedances are lumped and a simple parallel circuit can be formed and a combined impedance can be calculated using Z total = Z HF Z UHF. Table 7.2 shows the calculated combined impedance of a dual frequency antenna when given individual impedances of a HF antenna and a UHF antenna while Table 7.3 compares the calculated combined impedance of Page 153

170 7.4 Dual-Frequency Antenna Design 02 Feb 2007 Ansoft Corporation 15:01:53 XY Plot 1 HFSSModel Y1 im(z(lumpport1,lum Setup1 : Sweep1 Y1 re(z(lumpport1,lum Setup1 : Sweep Y Freq [GHz] Figure 7.7. Impedance of the second version of dual frequency antenna at UHF. a dual frequency antenna with the simulated impedance of the same dual frequency antenna. Fig. 7.8 shows clearly the accuracy of the calculated impedance value as compared to the simulated impedance based on the values shown in Table Impedance (Ohm) Re(Simu) Re(Cal) Im(Simu) Im(Cal) Freq (GHz) Figure 7.8. Graph showing the calculated impedance and simulated impedance of a dual frequency antenna. This experiment confirms that the total impedance of a single feed dual frequency antenna is dependent on both the impedance of the combined antennas following simple parallel circuit theory. Test of Transmission Line Length Page 154

171 Chapter 7 Dual Frequency RFID Antenna Table 7.2. Calculation of combined impedance based on individual impedances. Freq. Re(Z dipole ) Im(Z dipole ) Re(Z coil ) Re(Z coil ) Re(Z total ) Re(Z total ) (GHz) (Ω) (Ω) (Ω) (Ω) (Ω) (Ω) Table 7.3. Comparison between calculated impedances and simulated impedances Freq. Re(Z cal ) Im(Z cal ) Re(Z simu ) Re(Z simu ) (GHz) (Ω) (Ω) (Ω) (Ω) Page 155

172 7.4 Dual-Frequency Antenna Design Coil Antenna Dipole Antenna Merging of Antenna (Feed Point at B) A l B l (A) (B) Figure 7.9. (A) A transmission line is added between the two antennas. (B) Feed point located at B. The transmission line between the two antennas of interest is shown in Fig The single feed point is to be located at one of the original feed points. Feed point B is chosen, and l equals λ 4 (UHF). The loop antenna (for HF) can be designed to have low impedance at UHF, to be transformed into high impedance using the λ 4 transmission line, so it will not interfere with the UHF dipole. The UHF dipole has small capacitance and will not interfere with the HF antenna at HF. The influence of the length of the transmission line on the overall impedance of a dual frequency antenna was investigated. The overall impedance of a dual frequency antenna with different transmission line length is plotted in a Smith Chart shown as Fig Simulations were carried at various discrete frequency points and hence the plots are jagged and not smooth. However, Fig provides sufficient insight to shows that there is a pattern in the changes of impedance when the length is varied Redesign of the UHF dipole In the previous section, we have shown two important facts: A single-feed dual frequency antenna can be obtained by merging two independent antennas, and an adjustable transmission line can be used to controlled the overall impedance. Page 156

173 Chapter 7 Dual Frequency RFID Antenna 10mm 20mm 30mm 40mm 80mm Figure Smith Chart showing the simulated impedance of the dual frequency antenna with different transmission line length. As can be seen, As the transmission line length changes, the impedance will change accordingly, following a vague pattern. Now, we will redesign the UHF dipole so that it operates in the frequency band as intended. The frequency band for RFID around the world is within MHz. The first design is as shown in Fig It is just a normal dipole with a matching network added. The matching network chosen is modified version of the T-match as discussed in [75]. The designed antenna is tuned carefully to have a good match for a large bandwidth. The design is shown in Fig. 7.12, and the impedance within the UHF RFID band is shown in Fig However, the real part is slightly higher than expected. Also, the dimension of the dipole is too long in the x-axis direction. Hence, the two sides of the dipole are bent as shown in Fig When comparing with Fig. 7.12, the length at the x-axis is reduced from 168 mm to 78 mm, in the expense of the overall length in the y-axis. However, this will not be a problem, as we have to Page 157

174 7.4 Dual-Frequency Antenna Design Merge3-1 - HFSSModel1-3D Modeler Friday, February 09, 2007 Merge3-3 - HFSSModel1 Figure - 3D Modeler A UHF dipole with a matching network. Friday, February 09, 2007 Figure A UHF dipole with tuned matching network. position the transmission line along the y-axis as well. The impedance of this new design is shown in Fig The resistance of this new design has been lowered to well below the 20 Ω line for the entire UHF band for RFID ( MHz) Compatibility of dipole in HF The dipole shown in Fig has a good match with an RFID UHF chip within the UHF band. However, the matching network of the UHF dipole has provide a short circuit path for HF. The initial idea is to add a series capacitance which acts like an Page 158

175 Chapter 7 02 Feb 2007 Ansoft Corporation 15:45:44 XY Plot 1 HFSSModel Dual Frequency RFID Y1 Antenna im(z(lumpport1,lum Setup1 : Sweep1 Y1 re(z(lumpport1,lum Setup1 : Sweep Y Freq [GHz] Merge3-4 - HFSSModel1-3D Modeler Thursday, February 15, 2007 Figure Impedance of the UHF dipole in Fig Figure A UHF dipole with bent sides. open circuit at HF and acts like a short circuit at UHF. The additional series capacitance takes the form of a gap, with an underlying track, as shown in Fig To maintain a balance current flow, it is better to reposition the gap so that the gap is symmetric along the y-axis. The end result is as shown in Fig. 7.17, (top view in Fig. 7.18). The patch at the bottom of the gap can be adjusted easily for fine-tuning purposes. The simulation results of the final design of the UHF dipole with matching network (Fig. 7.18) are shown in Fig and Fig which show the impedance in the HF Page 159

176 7.4 Dual-Frequency Antenna Design 02 Feb 2007 Ansoft Corporation 15:49:39 XY Plot 1 HFSSModel Y1 im(z(lumpport1,lum Setup1 : Sweep1 Y1 re(z(lumpport1,lum Setup1 : Sweep Y Freq [GHz] Figure Impedance of the UHF dipole in Fig Merge3-5 - HFSSModel1-3D Modeler Thursday, February 15, 2007 Figure A UHF dipole with bent sides and gap. band and UHF band respectively. At HF, the antenna has a relatively huge impedance. Though the real part is significantly smaller than the imaginary part, at MHz, its impedance is Ω. At UHF, its impedance is a bit higher at the 860 MHz end, but can be adjusted easily, by changing the width of the underpass track below the gap. As mentioned before, the design is focussed on 960 MHz, which is the highest end of the RFID operating frequency. At 960 MHz, we have Ω. Page 160

177 Chapter 7 Dual Frequency RFID Antenna Merge3-6 - HFSSModel1-3D Modeler Thursday, February 15, 2007 Figure A UHF dipole with bent sides and repositioned gap. Merge3-7 - HFSSModel1-3D Modeler Thursday, February 22, 2007 Figure Final version of the UHF dipole, with matching network, bent sides and gap Merging of the new UHF dipole with the HF coil Once we have the dipole providing a conjugate match with a UHF RFID chip at UHF, the attention is now is to obtain a resonance at HF. We are assuming that a RFID chip will not provide any input capacitance at HF. Hence The antenna itself must have a self-resonance at HF. Following f 0 = 1 2π LC, where L and C are the inductance and the capacitance of the antenna respectively, we must increase either L or C or both to achieve this HF resonance. As mentioned before, The Page 161

178 7.4 Dual-Frequency Antenna Design 22 Feb 2007 Ansoft Corporation 13:09:29 XY Plot 1 HFSSModel Y1 im(z(lumpport1,lum Setup2 : Sweep1 Y1 re(z(lumpport1,lum Setup2 : Sweep Y Freq [MHz] Figure HF Impedance of the final version of the dipole shown in Fig Feb 2007 Ansoft Corporation 16:16:03 XY Plot 3 HFSSModel Y1 im(z(lumpport1,lumpport1)) Setup1 : Sweep1 Y1 re(z(lumpport1,lumpport1)) Setup1 : Sweep Y Freq [GHz] Figure UHF Impedance of the final version of the dipole shown in Fig UHF dipole has a gap so that it will look like an open circuit at HF. This is not exactly true as Fig shows that the impedance of the UHF dipole at HF is not near infinity. The UHF dipole will be acting like a capacitor in parallel with the HF loop antenna (which is an inductor). However, the inductance and the capacitance is not high enough to have a resonance point at 13.56MHz, which is actually a very low frequency to achieved with respect to the size of the antenna. The solution is to have a back to Page 162

179 Chapter 7 Dual Frequency RFID Antenna back multi-turn loop antenna on the both sides of the FR4, to provide the extra capacitance required. Via holes are used to connect the planar tracks on both sides. Having a back to back multi-turn loop antenna also solves the determination of l. The back to back tracks allows UHF signals to bypass most of the inductive loop, making the HF loop antenna appear to be a short circuit. Using l equals a quarter of a wave-length, the short circuited loop antenna is transformed to an open circuit. The final design of the antenna is as shown in Fig The size of the antenna is approximately 110 mm 100 mm. Even with the UHF dipole attached, the four turn HF planar coil antenna is found to have too high a self-resonance frequency as shown in Fig Hence the first step is to increase both the inductance, L, and capacitance, C, of the antenna. The idea is to have a two sides HF planar coil antenna. Not only the length of the coil is double (which increases the L value), the overlapping coil tracks on the both sides of the antenna will provide extra C to the antenna. 22 Feb 2007 Ansoft Corporation 16:31:27 XY Plot 8 HFSSModel Y1 im(z(lumpport1,lum Setup4 : Sweep1 Y1 re(z(lumpport1,lum Setup4 : Sweep Y Freq [MHz] Figure Self resonance frequency of HF planar coil antenna shown in Fig The design is as shown in Fig It can be seen that the overall dimension in the x-axis is caused by the HF antenna. Hence HF antenna is redesigned to reduce the overall dimension. The idea is to have the number of loops reduced by one. To cater for the reduction in L and C caused by the reduction of the number of loop, the coil Page 163

180 7.5 Antenna Fabrication and Testing track is made thinner, reduced from 3 mm to 2 mm. The final design is as shown in Fig Merge9-1 - HFSSModel1-3D Modeler Thursday, February 22, 2007 Figure Merging of double-sided HF antenna with a dipole. Fig shows the simulated impedance values of the dual-frequency antenna within the UHF band. At 960 MHz, the antenna impedance is j Ω, which is very near to the intended impedance of j Ω. For HF, since the antenna represent a parallel resonance circuit, the resistance is theoretically infinite at MHz, while the reactance is 0 Ω. 7.5 Antenna Fabrication and Testing The final design for the dual-frequency antenna (Fig. 7.23) is fabricated on a doublesided FR4 material and the actual measurement is taken for both HF and UHF operation. In the case of HF, the antenna resistance is too high at resonance to have a proper impedance measurement using a network analyser. The unbalanced input of the analyser also poses problems of measurement. To demonstrate the fabricated antenna has a resonant point at HF, a transmission measurement is used, where a wide-band loop antenna is used as the transmitting antenna and the fabricated antenna is used as the receiving antenna. Rotation of the loop confirms that coupling is via magnetic field. Page 164

181 Chapter 7 Dual Frequency RFID Antenna Via Via Via Via (Front) (Back) Figure Final design for dual-frequency antenna: (A) Front look (B) Rear look The result is shown in Fig The exact resonance point is at MHz, which is very near to the designed frequency of MHZ. Impedance measurement was carried out from MHz, and compared with the simulated results, as shown in Fig It was discovered that the measured resistance is higher than the simulated results, but the measured reactance is lower than the simulated results. The impedance at 960 MHz is measured to be Ω. This discrepancy is probably due to balanced load measuring problem, where the antenna is a balanced load while the network analyser is unbalanced. This can be solved by having a RF transformer or by connecting a BALUN to the antenna or employing other means of suppressing common mode currents. However this will not be discussed further in this chapter but in the next chapter of this thesis. Nonetheless, the measurements have shown that a dual-frequency antenna has been successfully designed and fabricated, which could be used in an RFID dual-frequency operation. So far we have presented a detailed design for a high frequency ratio dual-frequency antenna. Simulated and measured results are shown to confirm its functionality at both HF and UHF, as specified in the design aims. This shows that with proper design, a single feed dual-antenna of very high frequency ratio can be achieved. So far we haven t focussed on the optimisation of the antenna size, which will be presented in the next section onwards. Page 165

182 7.5 Antenna Fabrication and Testing Figure Transmission measurement using network analyser to locate the resonance point at HF. Figure The simulated and measured results for the frequency response of the fabricated dualfrequency antenna within the RFID UHF band from MHz. Page 166

183 Chapter 7 Dual Frequency RFID Antenna 7.6 Miniaturisation of Dual Frequency RFID Antenna with High Frequency Ratio With a functioning dual frequency antenna, the attention is now focussed on the design of a miniaturised dual frequency Radio Frequency Identification (RFID) antenna, which can support both the HF (13.56 MHz) and UHF bands ( MHz). A novel dual frequency RFID antenna with high frequency ratio is presented in Fig The dimensions of this antenna are 114 mm (l) by 98 mm (w), giving a total area of (mm) 2. It has a resonant point at HF and has an impedance match to an RFID UHF chip impedance (17-150j Ω at 915 MHz). 7.7 Improved Novel Design There are some of the characteristics from previous design which will be retained: 1. It uses a similar idea to obtain a dual frequency antenna, which is by designing a HF and a UHF antenna separately. These two antennas are then merged together with a single feed. The HF antenna used is a HF coil antenna and has a resonance point at MHz. The UHF antenna chosen is a UHF electric dipole with a matching network so that it has a impedance match with the chip impedance at the operating frequency. 2. A transmission line (with length l t in Fig. 7.38(a)) is used to link the HF and the UHF antenna together with a single feed point, while preventing the HF antenna from affecting the UHF antenna. 3. It is double sided. The HF coil on the top overlaps with the HF coil on the bottom to provide enough capacitance in order to obtain a resonant point at HF. 4. A series capacitor is added at the matching network of the UHF antenna (Gap provided by g in Fig. 7.38(a)). This series capacitor will have negligible impedance at UHF but will have a very high impedance at HF. This is to prevent the UHF antenna from shorting the HF antenna. Page 167

184 7.7 Improved Novel Design 5. The HF antenna will provide a DC path, and this is required in certain rectifier circuit implementations. In order to have a miniaturised version of a dual frequency RFID tag antenna, some major changes have been made on top of those retained characteristics mentioned above. Also, some modifications were carried out to improve the overall performance of a dual frequency RFID tag antenna: 1. Relocation of UHF dipole The improvement idea is to locate the UHF dipole inside the HF planar coil antenna to reduce overall size of the dual frequency antenna. The model is as shown in Fig The UHF dipole had to be shrunk slightly to fit. The impedance of this antenna is as shown in Fig and Fig for HF and UHF respectively. The HF frequency is slightly off the target frequency of MHz, but this can be fine-tuned reasonably easily. However, the most challenging is the UHF impedance graph, where the impedances fluctuate as the frequency increases. This means the antenna can only be matched to the chip at a single frequency point only. 2. Miniaturisation of HF coil Since in the new design, the overall size of the dual frequency antenna is determined by the size of the HF coil, we want a smaller HF coil. The new HF coil is reduced from 98 mm by 65 mm to 81 mm by 58 mm, gap is 0.5 mm and track width is 1 mm. Width of transmission line is 6 mm. The rule of thumb in designing a HF coil antenna is to maximise the dimension of the interior area (w inner l inner ) to maximise the total inductance (including self inductance and mutual inductance) of the coil. With the reduction of the size of the HF coil, we have the problem of lower inductance and higher resonance frequency than the intended frequency. This problem can be solved by careful design and by introducing an additional loop into the HF coil as shown in Fig Miniaturisation of UHF dipole Page 168

185 Chapter 7 Dual Frequency RFID Antenna InMerge10 - HFSSModel1-3D Modeler Friday, February 23, 2007 Figure A different version of a dual frequency antenna where the dipole is located in the inside of the HF coil, Ansoft to reduce Corporationoverall area. 23 Feb :35:58 XY Plot 2 HFSSModel Y1 im(z(lumpport1,lum Setup2 : Sweep1 Y1 re(z(lumpport1,lum Setup2 : Sweep Y Freq [MHz] Figure The HF impedance of the dual frequency antenna in Fig With the size reduction of the HF coil antenna, the UHF dipole antenna can be reduced as shown in Fig Merging the new HF coil and UHF dipole together, we have a new dual frequency antenna as shown in Fig HF not affecting UHF over a broader bandwidth Page 169

186 7.7 Improved Novel Design 23 Feb 2007 Ansoft Corporation 16:35:21 XY Plot 1 HFSSModel Y1 im(z(lumpport1,lum Setup3 : Sweep1 Y1 re(z(lumpport1,lum Setup3 : Sweep Y Freq [MHz] Figure The UHF impedance of the dual frequency antenna in Fig new-merge1 - HFSSModel1-3D Modeler Tuesday, February 06, 2007 Figure A miniaturised HF coil antenna. Initially, it was believed that the HF antenna would behave like a short circuit at the UHF band as the overlapping HF coil provides a short circuit path for UHF signal. Hence, it was decided to use a transmission line of length λ 4 to transform this short circuit into an open circuit, so that the HF coil would not affect the operation of the UHF dipole. Page 170

187 Chapter 7 Dual Frequency RFID Antenna new-merge50 - HFSSModel1-3D Modeler Wednesday, February 07, 2007 Figure A miniaturised UHF dipole. new-merge2 - HFSSModel1-3D Modeler Tuesday, February 06, 2007 Figure A dual frequency antenna with miniaturised HF coil and UHF dipole. However, it was discovered later that the HF antenna is not exactly a short circuit for the entire band of UHF as shown in Fig. 7.32(a). If we simply add λ 4 transmission line of any characteristic impedance, the HF antenna will still affect the UHF antenna over a large portion of UHF band. Page 171

188 7.7 Improved Novel Design 06 Oct 2006 Ansoft Corporation Smith Plot 2 HFSSModel1 17:44: Z MHz M S(LumpPort1,LumpP MHz M M1= P1= MHz 10 Oct 2006 M2= Ansoft Corporation 12:00:29 (a) P2= Smith Plot MHz HFSSModel Z MHz MHz M1= P1= MHz (b) Figure Impedance transformation in Smith Chart where all the traces cover from 860 to 960 MHz: (a) Trace 2 is the reflection coefficient of the impedance of the antenna normalised to 50 Ω while Trace 1 is the reflection coefficient of the impedance of the antenna re-normalised to 290 Ω. (b) Trace of the reflection coefficient when Trace 1 in (a) is transformed using a λ 4 transmission line. Page 172

189 Chapter 7 Dual Frequency RFID Antenna Hence, the idea is to design a transmission line with high characteristic impedance, Z 0. The type of transmission line chosen is coplanar strips (CPS). CPS has a theoretical physical limit on the lowest and highest Z 0 of 45 and 280 Ω respectively when ε r is 10 [76]. The highest Z 0 limit can be increased by reducing the ε r value. If we design our CPS to have a very high Z 0, by changing w t and s t, we can re-normalise the impedance of the designed antenna in Smith Chart (by default normalised by Z 0 of 50 Ω) before we transform the impedance of the antenna using a CSP. The S parameter of an antenna is given by: S = Z L Z 0 Z L + Z 0 (7.1) where the characteristic impedance (or normalising impedance), Z 0, is purely resistive. Rearranging (7.1) will give us: Z L = Z S 1 S (7.2) And with a new characteristic impedance (of the CPS), the new S parameter can be computed using: S new = 1+S Z 0 1 S Znew 0 1+S Z 0 1 S + Znew 0 (7.3) The effect is shown in Fig. 7.32(a), and with the impedance transformation of the λ 4 CPS, the HF coil impedance is plotted as Fig. 7.32(b). It can be seen that the HF coil behaves like a high impedance for most of the UHF band of interest. The final design of the transmission line linking the HF and the UHF antenna together has the following design values (Refer Fig. 7.38(a)): s t = 6 mm, w t = 0.8 mm and ε r = 4.4, resulting a Z new Ω. The effective dielectric constant value, ε e f f is approximately Hence, the length of the λ 4 CSP is: Page 173

190 7.7 Improved Novel Design l t = λ 4 ε r (7.4) At 920 MHz, λ is approximately m. Hence l t is approximately m or 56 mm. 5. The coupling effect between UHF and HF antenna As mentioned before, we designed and simulated a UHF electric dipole antenna with matching impedance to the RFID tag chip. A transmission line is used to separate the HF and UHF antennas. In our previous design, the UHF dipole antenna is located beside the HF coil. Although the overall size is increased, the coupling between the HF and UHF antennas is negligible. In our new design, it is decided to move the UHF dipole inside the HF coil to reduce overall size. However, the coupling between the antennas changes the UHF dipole impedance quite significantly and we no longer have a impedance match between the UHF dipole and the UHF chip at UHF. A fine tuning was carried out by readjusting g, l uhf and l under. 6. Reduction of HF antenna size To fit both the transmission line and the UHF antenna within the HF coil antenna, the HF coil antenna is rotated 90 degree from the previous design (compare Fig and Fig. 7.33). With the HF coil rotated, the simulation results are as shown in Fig and Fig for the antenna impedance in HF and UHF respectively. It can be seen that it is still functioning well as a dual frequency antenna. However, the antenna in Fig does not contribute to the aim of minimising our antenna. Hence, ultimately, the UHF antenna has to be minimised and relocated inside the HF coil antenna as discussed before. The impact on the size limitation of the UHF dipole antenna is significant. The w uhf is shortened from 95 mm to 44 mm. Although we can design this dipole antenna to be matched in impedance, its simulated directivity drops from 2.9 to 2.2. This is a trade-off between size and performance. Page 174

191 Chapter 7 Dual Frequency RFID Antenna new-merge80 - HFSSModel1-3D Modeler Wednesday, February 07, 2007 Figure Dual frequency Ansoft Corporation antenna with rotated HF coil. 07 Feb :20:33 XY Plot 2 HFSSModel Y1 im(z(lumpport1,lum Setup4 : Sweep1 Y1 re(z(lumpport1,lum Setup4 : Sweep Y X1= 12.30MHz Y1= Freq [MHz] Figure Point 1 at MHz. 7.8 Testing The prototype of the designed dual frequency antenna was tested and found to be resonating at MHz while having reasonable impedance match with a generic RFID tag chip at UHF. Taking a further step towards putting our dual frequency antenna into real life testing, the biggest challenge in real life testing for a dual frequency RFID tag is that there is no single tag chip available that would work on both HF and UHF. It is our intention to Page 175

192 7.8 Testing 08 Feb 2007 Ansoft Corporation 14:42:41 XY Plot 1 HFSSModel Y1 im(z(lumpport1,lum Setup3 : Sweep1 Y1 re(z(lumpport1,lum Setup3 : Sweep Y Freq [MHz] Figure UHF simulated results. A B Figure Figure to illustrate the position of RFID chip attached to the antenna. In the initial testing both HF and UHF chips are attached to feed point A. In the final prototype, UHF chip is located at point A while HF chip is located at point B. have a single feed system. However, for the purpose of testing, we used an UHF RFID C1G1 chip and a Tag Talk First HF chip. Referring to Fig. 7.36, we first put both the HF and UHF chip at position A. In HF operation, the read range is limited by the size of the reader antenna [84]. A quick measurement shows that this dual frequency RFID tag offers a maximum possible read range at HF when using a fixed size reader antenna. However, the HF chip affects the UHF chip significantly and a poor read range of only 0.3 m is observed at UHF. If we move the HF chip to the B position in Fig. 7.36, the HF chip will be transformed to a very high impedance at UHF by the transmission Page 176

193 Chapter 7 Dual Frequency RFID Antenna line. Hence the UHF chip will perform normally and a read range of more than 2 m is observed. A quick check again shows that this dual frequency RFID tag still offers a maximum possible read range at HF. Although the final prototype dual frequency RFID tag has two chips and has two feed points, we believe that our prototype will work with equivalent efficiency in both HF and UHF bands when a single dual frequency chip is used on a single feed. To compare our dual frequency RFID tag with commercial tags, we placed a commercial HF RFID tag and a commercial UHF RFID tag side by side and carried out a read range measurement in both the HF and UHF bands. We discovered that the read range performances are comparable when the tags are alone or placed near to each other. The conclusion is that, our prototype antenna is superior over the commercial tags because it can handle operations in both frequency bands with only one feeding point. 7.9 Final Design new-merge100 - HFSSModel1-3D Modeler Wednesday, February 07, 2007 Figure A miniaturised dual frequency antenna. The new design is to relocate the UHF electric dipole antenna inside the HF coil antenna to reduce the overall size as shown in Fig The total dimension is 81 mm (l hf ) by 58 mm (w hf ), giving a total area of 4698 mm 2. Page 177

194 7.10 Conclusion w hf w uhf g w inner Feed A l under l hf w t l uhf s t l t l inner Feed B Via Via Via (a) Front view Via (b) Back view Figure Final antenna design with design parameters. The final prototype has the following design values in mm: l hf = 81; w hf = 58; l inner = 70; w inner = 48; l uhf = 37; w uhf = 44; g = 3; w t = 0.8; s t = 6; l under = Conclusion We have presented novel designs for dual frequency RFID antenna with high frequency ratio by the method of merging a HF antenna together with a UHF antenna. This antenna works well in both the HF and UHF bands for RFID operation with a single feed. A functioning dual frequency RFID tag proves that having a dual frequency RFID system is feasible. Although no dual frequency RFID chip is available at the moment, we hope that this successful design of a compact dual frequency RFID antenna will catalyse the development of a dual frequency RFID chip. Page 178

195 Chapter 8 Measurement T HE performance of a passive Radio Frequency Identification (RFID) tag depends on the antenna and the chip efficiency of the tag, and also depends on the matching between the antenna and the chip. The design of the tag antenna often involves the measurement of its impedance. This paper investigates the RF cable effect on the measurement of RFID tag antenna impedance. Measurement results obtained through several common methods are compared with the simulated results. Ways of minimising measurement error are discussed and an analysis on the effect of mismatch between tag antenna and chip are presented. Page 179

196 8.1 Introduction 8.1 Introduction (Draft chapter only!) A basic Radio Frequency Identification (RFID) system comprises of three major parts; RFID readers or interrogators, RFID tags and RFID network. An RFID reader uses radio link to communicate with RFID tags to obtain information stored in RFID tags, such as unique Electronic Product Code (EPC) of a tag or any data collected through any sensor attached on a tag. This information is then passed to RFID network for storage and processing. For wide deployment of RFID system in supply chains, where every object has a RFID tag attached, passive RFID tags are chosen because of their low cost. These passive tags do not have any built-in power source, and solely rely on the RFID reader RF signal for operation. To enable maximum power transfer from the tag antenna to the tag chip, the tag antenna should have an impedance conjugate of the input impedance of the tag chip. Also, a tag antenna should have at least unity gain with wide antenna gain pattern, to enable easy absorbtion of RF energy from any RFID reader. A RFID tag antenna can be designed using simulation software, such as Ansoft HFSS. During simulation, fine-tuning can be carried up to adjust antennas parameters, such as the operating frequency, bandwidth and impedance, to the desired values. Most of the time, with the restriction of antenna size and cost, the designing stage often involve trade-offs between some of the antenna parameters. The designed antenna is then fabricated. The biggest challenge is on how a prototype tag antenna is tested to confirm that it functions as designed, before large scale production is carried out. This paper aims to investigate the challenge of measuring a small RFID tag antenna using several different ways. Suggestions are presented to improve the measurement results to reflect the actual characteristic of an RFID tag antenna. There is no literature which explicitly discusses the measuring methods of the performance of an RFID tag antenna. The closest literature is on the measurement of small antennas used by mobile handset, which will be discussed in details in the next section. Section 8.3 shows the experimental settings while Section 8.4 presents the experimental Page 180

197 Chapter 8 Measurement results, together with some results interpretations. Further modifications and improvements on measurement were carried out and discussed in Section 8.5 while conclusion is presented in Section Background and Literature Review An RF feed cable will affect the measurement of input impedance, current distribution, and radiation characteristic of the antenna. The source of the error is the appearance of common mode current at the outside surface of the outer conductor of a coaxial cable (Fig. 8.1 (b)). In the case of antenna measurement, when the antenna is a balanced load, this common mode current is caused by the balanced to unbalanced connection between an antenna under test (AUT) and the coaxial cable as shown in Fig. 8.1 (b). S 11 will be the only antenna parameters to be measured and compared with the simulation results in this paper. Another useful antenna parameters which can be compared is the radiation pattern as discussed in [115]. However, the authors believed that S 11 is more important than radiation pattern in the case of RFID tag antenna as S 11 affects the power transfer from the antenna to the tag chip, and power transfer from the antenna to the tag chip is directly proportional to the read range of an RFID tag. Although it is good to have the verification of simulated radiation pattern, a tag antenna gain is averagely unity in all direction and the orientation of a tag antenna with respect to an RFID reader antenna is random in real life, which reduce the importance of radiation pattern of an RFID tag antenna in predicting its read range. DeMarinis [116] proposed the use of a ferrite choke as an effective solution while Icheln [117, 118] proposed the use of Balun to transfer the unbalanced nature of a coaxial cable to a balanced antenna under test (AUT) to minimise the effect of RF cable. However, Icheln stressed that the use of Balun outperform the use of ferrite choke despite the fact that Balun is often band limited, as ferrite choke is lossy and will affect the measurement of radiation pattern. Alternative methods includes a measurement scheme using a microstrip transmissionline to connect a coaxial cable to a AUT as suggested by Chen [119], and also, the Page 181

198 8.3 Settings and Connection Figure 8.1. Common Mode Current: (a) flow at the outside surface of the outer conductor of a coaxial cable (b) caused by the direct connection of a balanced load to an unbalanced connecting cable. use of a stack of metal rings to suppress the appearance of common mode current as presented by Kahng [120]. 8.3 Settings and Connection An RFID tag antenna is normally made of thin copper tracks on an adhesive sheet. This is to make an RFID tag easily and readily attached to an object. However, for convenience, our prototype tag antenna is fabricated on a solid dielectric material to have a solid foundation. The material chosen is FR4. A SMA connector is then used as the feed point of the AUT in order to connect the AUT to our network analyzer for measurement. Measurement results are then compared with the simulated results obtained through Ansoft HFSS simulation software. Three types of coaxial cable configurations is used for testing purposes: 1. Using RG223 coaxial cable. 2. Using RG223 coaxial cable with ferrite sleeve. 3. Using Gore GMCA (now known as G2) coaxial cable. Two types of antenna are tested using some or all of the coaxial configurations stated above: Page 182

199 Chapter 8 Measurement 1. A balanced bow tie antenna. A balanced bow tie antenna is fabricated and is shown in Fig The feed point of the antenna (a SMA connector) is from the back and through the FR4 material as shown in Fig 8.2 (a). Figure 8.2. A bow-tie antenna on FR4: Side view (a) With SMA connector (b) With modified SMA connector to enhance measurement results. 2. Half of the balanced bow tie antenna on ground plane. The dimension of the balanced bow tie antenna (Fig. 8.2) is referred to fabricate this special antenna, which is half of the balanced bow tie antenna on ground plane (Fig This unbalanced version of the bow tie antenna is soldered on a SMA connector, which is mounted on a ground plane. The idea is to combine the unbalanced bow tie antenna with its image created by the ground plane, to simulate a complete balanced bow tie antenna. It has been proven in theory that a half unbalanced bow tie antenna on a ground plane will have half of the input impedance of a complete balanced bow tie antenna. Page 183

200 8.4 Testings and Results Figure 8.3. Half bow-tie antenna on ground plane To illustrate the effect of SMA connector of the balanced bow-tie antenna on the results, a comparison between the simulated and measured reactance of a simple bow tie antenna is shown in Fig As the edges of the SMA connector overlap with part of the AUT, extra capacitance is added. The solution is to minimise the edge of the SMA connector as shown in Fig 8.2 (b) and the improved results is plotted in Fig In this paper, all the experiment carried out were with a modified SMA connector, unless otherwise stated. Figure 8.4. The effect of SMA connector on measurement results. 8.4 Testings and Results A balanced bow tie antenna The inaccuracy of measurements done using Cable (1), Cable (2) and Cable (3) are on average similar as shown in Fig. 8.5 and Fig However, by hand touching both of the cables and monitoring the impedance of the AUT, it is confirmed that Cable (1) and Page 184

201 Chapter 8 Measurement Cable (3), without the addition of ferrite sleaves, are more susceptible to environment effect. Figure 8.5. Simulated and measured resistance of AUT. Figure 8.6. Simulated and measured reactance of AUT Half bow-tie antenna on ground plane It is observed that the ground plane which the half bow-tie antenna is mounted on provided sufficient shielding between the cable and the AUT. The results obtained using any of the cable shows negligible difference. In all comparisons and graphs presented in this paper, measured values for a half balanced bow tie antenna on ground plane has been doubled for the reason explained in previous section to enable easy comparison. In comparison of resistance values, the measured results are constantly higher than the simulated results. This is most probably due to the loading effect of the ground plane. Nonetheless, in comparison of reactance values, the measured results are slightly better than all the results obtained Page 185

202 8.5 Discussion and Improvement on a balanced bow-tie antenna, with the exception of Cable(3), which is a high quality shielded coaxial cable. Hence, the ground plane has effectively shielded the cable and the measuring instrument from any unwanted coupling. The challenge in using this method is that the AUT must be symmetrical in nature, which is not always the case for a RFID tag antenna. Also, the distance of the half bow-antenna from the ground plane must be half of the distance between the two feed points of a balanced bow tie antenna. 8.5 Discussion and Improvement Using coaxial cable with high shielding can improve the measurement results by reducing the coupling between the cable and the AUT. However, not all error can be eliminated as inaccuracy may be introduced by the balanced to unbalanced issue. Hence a Balun is designed by using coplanar strips with an RF transformer (Fig. 8.7 (a)). The model of the chosen commercial RF transformer is TC1-1-13M with with characteristic impedance of 50 Ω and with operating frequency range of 4.5 to 3000 MHz. The biggest challenge in using RF transformer is to have a 50 Ω coplanar strips. According to [76], the impedance of coplanar strips is physically limited to above 45 Ω. Even for a 50 Ω coplanar strips, the fabrication process is extremely difficult. Since the AUT is small, the track size W1 and W2 cannot be too wide (W1 and W2 < 5 mm). To obtain 50 Ω coplanar strips with W1 and W2 less than 5 mm, the width of separation, S, has to be around 0.05 mm for FR4 material with ε r equals 4.4. The final design uses a more expensive composite dielectric, consists of polytetraflouroethylene (teflon) and ceramic, with a ε r equals W1 and W2 are both 2 mm in width and the separation is 0.12 mm. The calculated characteristic impedance is Ω. The results obtained using Balun are summarised in Table 8.1. It can be seen that with the use of Balun, the measurement of resistance is the best, though not much better, than the rest. On the other hand, a high quality co-axial cable provides the best results for reactance measurement. Page 186

203 Chapter 8 Measurement Figure 8.7. A Balun: (a) The use of balun to connect a balanced load to unbalanced coaxial cable (b) The design parameters of coplanar strips. Table 8.1. Average Error and Highest Error Average Error (Ω) Highest Error (Ω) Re Im Re Im Cable (1) Cable (2) Cable (3) With Ground Plane With Balun It is inconclusive which measurement methods is the best among all, as a direct measurement using a normal coaxial cable offers comparably acceptable results. Note that the results is only applicable for small RFID antenna. Big scaled antenna, such as a half-wave dipole, will most certainly required a Balun or ferrite sleeve for proper measurement. Also, in all our comparison, simulated results are used as the standard values. We are assuming that with proper fabrication process, an actual antenna with similar characteristics with the modeled antenna, can be obtained. Fig. 8.8 and Fig. 8.9 serve two purposes. Firstly, they show the power transfer loss when there is a mismatch in either the real part or imaginary part between the AUT and the chip impedances. It can then be used to project the impact of inaccuracy in measurement. Although mismatch in both real and imaginary is common, it can be approximated, even not very accurately, from both of these graphs, by multiplying two power transfer losses. Page 187

204 8.5 Discussion and Improvement The second purpose is to show the impact when the characteristics of the modeled antenna is not accurately translated into the fabricated antenna. If the initial antenna design is entirely carried out using simulation program, this problem will most likely happen. It shows that by using a higher resistance value chip, the power transfer loss can be minimised. Figure 8.8. The power transfer efficiency between AUT and chip when reactance are equal and opposite, given the chip input resistance, with respect to resistance difference between the AUT and chip. Figure 8.9. The power transfer efficiency between AUT and chip when resistance are equal, given the chip input resistance, with respect to reactance difference between the AUT and chip. The investigation of RF cable effect on RFID tag antenna impedance measurement exposes the difficulties in getting accurate measurement and obtain the best match between AUT and chip. In actual RFID tag design, it is best to: 1. Design by simulation. Adjust the dimension of the antenna so that the input impedance equals to the conjugate of the chip impedance. Page 188

205 Chapter 8 Measurement 2. Fabricate the designed antenna. Measure the antenna input impedance. 3. Attach tag chip on antenna. Fine-tuning antenna, such as reducing the dimension of certain part of the antenna, while monitoring the read range of the tag. 8.6 Conclusion Detailed investigation of RF cable effect on RFID tag antenna impedance measurement is presented in this paper. 8.7 Calculation of Coplanar Strips Variable defined in Fig. 8.7 (b): S, W1, W2, t, h. Defined new variable: a = S 2 (8.1) b = W1 + W2 + S 2 (8.2) Characteristic Impedance, Z 0 : Z 0 = 120π εe f f K(k ) K(k) (8.3) where: ε e f f = (ε r 1) K(k) K(k h ) K(k ) K(k h ) (8.4) ( a ) 2 k = 1 (8.5) b k = 1 k 2 (8.6) k h = 1 sinh2 ( πa 2h ) sinh 2 ( πb 2h ) (8.7) Page 189

206 8.8 Calculation of S parameters and Power Transfer Efficiency where K is the complete elliptical integral of the first kind. k h = 1 k 2 h (8.8) 8.8 Calculation of S parameters and Power Transfer Efficiency Defined: Z L is the impedance of chip; Z S is the impedance of AUT. Z L = a + jb (8.9) Z S = c + jd (8.10) If reactance are equal and opposite (b = d): If resistance are equal (a = c): s = Z L Z S Z L + Z (8.11) S s = a c a + c (8.12) s = (b + d) 2 + j2a(b + d) 4a 2 + (b + d) 2 (8.13) Note that b is always negative as an RFID tag chip is always capacitive in nature due to the reservoir capacitor located at the rectifier circuit. Hence the term (b + d) is actually the difference between the magnitude of Im(Z L ) and Im(Z S ). This term should be minimised for maximum power transfer as maximum power transfer occurs when b equals d (when (b + d) equals zero), or in other words Z L is the conjugate of Z S. Power transfer efficiency: η = 1 s 2 (8.14) Page 190

207 Chapter 8 Measurement 8.9 Raw Experiment Data (To be removed in final version) The initial measurement shows that the SMA feed point of the balanced bow-tie antenna has significant impact on the results. The size of the SMA connector is comparable with the size of the AUT. The edges of this SMA connector provides an alternate path for displacement current. The solution is to minimise the edge of the SMA connector as shown in Fig 8.2 (b). The simulated and measured results for real and imaginary parts of the impedance for the bow-tie antenna, as shown in Fig. 8.2 and Fig. 8.3, are presented in Fig and Fig respectively. Figure Simulated and Measured Resistance of Bow-tie Antenna: Simu Re: Simulated results; Mod SMA Cable (1): Measurement using Cable (1) on Modified SMA connector (Fig. 8.2 (b)); Mod SMA Cable (2): Measurement using Cable (1) on Modified SMA connector (Fig. 8.2 (b)); Mono Cable (1): Using antenna (2) with Cable (1). Figure Simulated and Measured Reactance of Bow-tie Antenna Page 191

208 8.9 Raw Experiment Data (To be removed in final version) In the real part (resistance) comparison in Fig. 8.10, the half bow-tie antenna on ground plane configuration gives the best results, but is the only measurement that gives a higher resistance as compared to measurement configurations. (Loading effect? No Sure. To be discussed.) For the imaginary part (reactance), All measurement configurations matched well in the region 900 to 920 MHz, where the reactance values fall within a 5 Ω range. The position of ferrite beads affects the measurement results according to [121]. A small test showed that ferrite beads should not be located to near to the feed point of the antenna. It will increase the measured resistance, as the ferrite beads will act as additional load. The simulated and measured results for real and imaginary parts of the impedance for the dual frequency antenna, as shown in Fig. X are presented in Fig and Fig respectively. Figure Simulated and Measured Resistance of Dual Frequency Antenna Figure Simulated and Measured Reactance of Dual Frequency Antenna Page 192

209 Chapter 8 Measurement Using ferrite sleeve cable (Cable (2)), improves the measurement if resistent quite significantly. However, not much improvement in the measurement of reactance. Using the same configuration, simulation and measurement were carried out on bowtie antenna at a higher frequency (from 1.1 GHz to 1.4 GHz). The results are shown in Fig.8.14 Figure Results It is shown that at a higher frequency, the accuracy of the measurement of reactance drops, but the accuracy of resistent is still very acceptable Notes (To be removed in final version) It is observable that a special shorting connection, which resembles AUT, but with a shorting pin across the positive and negative terminal, does not have zero impedance (not a complete short circuit) at high frequency. In our case, where the AUT is a bow-tie dipole, the special shorting connection has a net inductance value. Hence, if a special shorting connection is used for calibration, hopefully the results can be improved. 1. Measurement accuracy drops as frequency increases. Page 193

210 8.10 Notes (To be removed in final version) 2. From the results shown, ferrite sleeve does improve the measurement results. However, it is discovered that the placement of ferrite sleeve may affect the measurement results, especially on the distance away the ferrite sleeve from the antenna under test. 3. Half of the AUT on ground plane gives best improvement overall. However, only limited antenna can be cut into half, such as the AUT must be symmetrical at the feed point. 4. In progress: Have a shorted antenna for calibration. Page 194

211 Appendix A Calculation of Coplanar Strips The characteristic impedance of a coplanar strips (CPS) is computed with reference to [122]. The variables defined in Fig. 8.7 (b) are S, W1, W2, t, h. To simplify the mathematical expression in calculating the characteristic impedance, two new variables are defined: a = S 2 (A.1) b = W1 + W2 + S 2 (A.2) Characteristic Impedance, Z 0 : Z 0 = 120π εe f f K(k ) K(k) (A.3) where: ε e f f = (ε r 1) K(k) K(k h ) K(k ) K(k h ) (A.4) ( a ) 2 k = 1 (A.5) b k = 1 k 2 (A.6) Page 195

212 k h = 1 sinh2 ( πa 2h ) sinh 2 ( πb 2h ) (A.7) k h = 1 k 2 h (A.8) where K is the complete elliptical integral of the first kind. Page 196

213 Appendix B MATLAB Code for Path Loss Calculation This Appendix includes all the MATLAB code required to produce the simulation results shown in Chap. 3. Detailed explanation on how the code works is also included in Chap. 3. Section B shows the main body of the simulation code, where the settings for any output plot is located. Also, the area of simulation can be changed there. Theoretically, there is no limit on the size of the area, except the limitation of the computer used for simulation. Section B has a data grid to represent the antenna gain pattern for the reader antennas used in the simulation. If a different antenna is used, a new data grid is required. The data grid is obtained through actual antenna gain measurement in an anechoic chamber. Section B computes the path loss based on path loss model discussed in Chap. 3. Main Code c l e a r ; 2 warning o f f MATLAB: dividebyzero ; 4 antenna1 = antgainv2 ( 5 0 0, 4 5 0, 5 0 ) ; antenna2 = antgainv2 ( 5 0 0, 4 0 0, 5 0 ) ; 6 t o t a l = 1 0 log10 ( 1 0. ˆ ( 0. 1 ( antenna1 ) ) ˆ ( 0. 1 ( antenna2 ) ) ) ; Page 197

214 8 figure ( 1 ) =pcolor ( t o t a l ) ; hold ; 10 color1 = [ ] ; c olor2 = [ ] ; 12 color3 = [ ] ; c olor4 = [ ] ; color5 = [ ] ; c olor6 = [ ] ; 14 colormap ( usercolormap ( color1, color2, color3, color4, color5, c olor6 ) ) ; 16 c a x i s ( [ ] ) ; 18 colorbar ; s e t ( gca, PlotBoxAspectRatioMode, manual ) ; 20 s e t ( figure ( 1 ), EdgeColor, none ) ; clabel ( contour ( t o t a l, 2 0, k ), manual ) ; 22 hold ; Antenna Gain Pattern function x=antgainv2 ( size, x, y ) disp ( Loading antenna gain pattern & 2 path l o s s.... ) ; n=size ; %a r r a y s i z e 4 a=y ; %antenna l o c a t i o n row 6 b=x ; %antenna l o c a t i o n column %antenna f a c i n g down 8 gain=ones ( n, n ) ; 10 %d a t a g r i d 12 ddata =[ Page 198

215 Appendix B MATLAB Code for Path Loss Calculation ] ; 38 ddata back =[ ]; Page 199

216 58 %c o m p u t a t i o n 60 for i =1:n for j =1:n 62 r a t i o = abs ( ( j b ) /( i a ) ) ; r a t i o = 2 (90 atan ( r a t i o ) 180/ pi ) ; 64 r a t i o = round ( r a t i o ) ; i f i a =0 66 i f a<=i gain ( i, j ) = ddata ( 1, r a t i o +1) ; 68 else gain ( i, j ) = ddata back ( 1, r a t i o +1) ; ; 70 end else 72 gain ( i, j ) = ddata ( 1, 1 ) ; end 74 i f j b==0&&a<=i gain ( i, j ) = 6 ; 76 end end 78 end x=gain pathloss ( n, b, a, ) ; %r e t u r n gain 80 disp ( Done ) ; Path Loss function x=pathloss ( size, x, y, u n i t s ) 2 n=size ; %a r r a y s i z e 4 ploss = ones ( n ) ; d i s t a n c e = ones ( n ) ; 6 a=y ; %antenna l o c a t i o n row b=x ; %antenna l o c a t i o n column 8 %G e n e r a t e D i s t a n c e 10 for i = 1 : n for j = 1 : n 12 d i s t a n c e ( i, j ) =( sqrt ( ( a i ). ˆ 2 + ( j b ). ˆ 2 ) ) ; Page 200

217 Appendix B MATLAB Code for Path Loss Calculation 14 end end d i s t a n c e ( i, j ) =distance ( i, j ). u n i t s ; 16 for i = 1 : n 18 for j = 1 : n 20 i f d i s t a n c e ( i, j ) ==0 22 end d i s t a n c e ( i, j ) =1. u n i t s ; 24 i f d i s t a n c e ( i, j )<8 26 ploss ( i, j ) =32+25 log10 ( d i s t a n c e ( i, j ) ) ; else 28 ploss ( i, j ) =23+35 log10 ( d i s t a n c e ( i, j ) ) ; end 30 end end 32 x = ploss ; Page 201

218 Page 202

219 Appendix C MATLAB Code for HFSS VB Script Generation This Appendix shows the MATLAB code written to generate HF coils of different size to be simulated in Ansoft HFSS. function makecoil ( turn, L, L1, thick, height,w, g, s a v e l o c ) 2 % add p a t h s t o t h e r e q u i r e d m f i l e s. 4 addpath C: \ Matlab\ h f s s a p i \3dmodeler ; addpath C: \ Matlab\ h f s s a p i \ general ; 6 addpath C: \ Matlab\ h f s s a p i \ a n a l y s i s ; addpath C: \ Matlab\ h f s s a p i \boundary ; 8 % Temporary F i l e s. These f i l e s can be d e l e t e d a f t e r t h e o p t i m i z a t i o n i s c o m p l e t e. We have t o s p e c i f y t h e c o m p l e t e path f o r a l l o f them. 10 tmpprjfile = C: \ Matlab\ h f s s a p i \temp. h f s s ; tmpdatafile = s t r c a t ( C: \ Matlab\ h f s s a p i \, s a v e l o c ) ; 12 t m p S c r i p t F i l e = C: \ Matlab\ h f s s a p i \temp. vbs ; 14 % HFSS E x e c u t a b l e Path. hfssexepath = C: \ Progra 1\ Ansoft \HFSS9\ h f s s. exe ; 16 % C r e a t e a new temporary HFSS s c r i p t f i l e. 18 f i d = fopen ( tmpscriptfile, wt ) ; 20 % C r e a t e a new HFSS P r o j e c t and i n s e r t a new d e s i g n. hfssnewproject ( f i d ) ; 22 h f s s I n s e r t D e s i g n ( fid, t e s t b o x ) ; Page 203

220 24 % C r e a t e t h e C o i l. 26 % v a r i a b l e s : n = turn ; 28 length0 = L ; length1 = L1 ; 30 t = t h i c k ; h = height ; 32 width = w; gap = g ; 34 % c r e a t e s u b s t r a t e 36 sub x = 2 length1 +2 (n+2) ( gap+width ) ; sub y = 2 length0 +2 (n+2) ( gap+width ) ; 38 hfssbox ( fid, Sub,[ sub x/2, sub y / 2, 0 ], [ sub x, sub y, h ], mm, FR4 epoxy, true ) ; 40 % f u n c t i o n h f s s C y l i n d e r ( f i d, Name, Axis, Center, Radius, Height, Units ) hfsscylinder ( fid, Via 1, Z, [ width / 2, length0+width / 2, 0 ], width /3, h, mm, vacuum, true ) ; 42 hfsscylinder ( fid, Via 2, Z, [ width / 2, length0+width/2+n ( width+gap ), 0 ], width /3, h, mm, vacuum, true ) ; h f s s S u b t r a c t ( fid, { Sub }, { Via 1, Via 2 }) ; 44 % c r e a t e c o i l 46 Count=n 4+1; 48 % Setup S t a r t L o c =[0, length0, 0 ] ; Curr Loc= S t a r t L o c ; Ct =1; 50 while ( true ) Arm = s t r c a t ( Arm, i n t 2 s t r ( Ct ) ) ; 52 i f Ct==1 54 %1 s t Arm temp length=length1 ; 56 temp width=width ; hfssbox ( fid,arm, Curr Loc, [ length1, width, t ], mm, copper, f a l s e ) ; Page 204

221 Appendix C MATLAB Code for HFSS VB Script Generation 58 e l s e i f Ct>=Count %f i n a l arm 60 temp length =(n+1) width+length1 +(n 1) gap ; temp width=width ; 62 hfssbox ( fid,arm, Curr Loc, [ temp length, temp width, t ], mm, copper, f a l s e ) ; break ; else 64 %t o p arm temp length = ( ( Ct 1)/2) width+2 length1 +( Ct 3)/2 gap ; 66 temp width=width ; hfssbox ( fid,arm, Curr Loc, [ temp length, temp width, t ], mm, copper, f a l s e ) ; 68 end 70 Ct=Ct + 1 ; Curr Loc=Curr Loc +[ temp length, temp width, 0 ] ; Arm = s t r c a t ( Arm, i n t 2 s t r ( Ct ) ) ; 72 %r i g h t arm 74 temp length=width ; temp width =( Ct /2) width+2 length0 +( Ct/2 1) gap ; hfssbox ( fid,arm, Curr Loc, [ temp length, temp width, t ], mm, copper, f a l s e ) ; 76 Ct=Ct +1; 78 Curr Loc=Curr Loc +[ temp length, temp width, 0 ] ; Arm = s t r c a t ( Arm, i n t 2 s t r ( Ct ) ) ; 80 %down arm 82 temp length = ( ( Ct 1)/2) width+2 length1 +( Ct 3)/2 gap ; temp width=width ; 84 hfssbox ( fid,arm, Curr Loc,[ temp length, temp width, t ], mm, copper, f a l s e ) ; 86 Ct=Ct + 1 ; Curr Loc=Curr Loc+[ temp length, temp width, 0 ] ; Arm = s t r c a t ( Arm, i n t 2 s t r ( Ct ) ) ; 88 %l e f t arm 90 temp length=width ; temp width =( Ct /2) width+2 length0 +( Ct/2 1) gap ; Page 205

222 hfssbox ( fid,arm, Curr Loc,[ temp length, temp width, t ], mm, copper, f a l s e ) ; 92 Ct=Ct +1; Curr Loc=Curr Loc+[ temp length, temp width, 0 ] ; 94 end 96 % f u n c t i o n h f s s C y l i n d e r ( f i d, Name, Axis, Center, Radius, Height, Units ) hfsscylinder ( fid, Via Cu 1, Z, [ width / 2, length0+width / 2, 0 ], width /3, h, mm, copper, f a l s e ) ; 98 hfsscylinder ( fid, Via Cu 2, Z, [ width / 2, length0+width/2+n ( width+gap ), 0 ], width /3, h, mm, copper, f a l s e ) ; 100 %Underpass hfssbox ( fid, Underpass 1, [ 0, length0, h ], [ width, width, t ], mm, copper, f a l s e ) ; 102 hfssbox ( fid, Underpass 2, [ 0, length0+gap+width, h ], [ width, n width +(n 1) gap, t ], mm, copper, f a l s e ) ; 104 %Merging Cu temp str= ; 106 for i = 1 : ( n 4+1) 108 temp str= s t r c a t ( temp str, Arm, i n t 2 s t r ( i ) ) ; i f i = ( n 4+1) 110 temp str= s t r c a t ( temp str,, ) ; end end hfssunite ( fid, { Underpass 1, Underpass 2, Via Cu 1, Via Cu 2, temp str }) ; %Lump p o r t 116 hfssrectangle ( fid, Rect 1, Z, [ 0, length0+width, h ], width, gap, mm ) ; 118 hfssassignlumpedport ( fid, LumpedPort, Rect 1, [ width /2, length0+width, h ], [ width /2, length0+width+gap, h ], mm, 5 0, 0 ) ; 120 % Add an AirBox. Page 206

223 Appendix C MATLAB Code for HFSS VB Script Generation hfssbox ( fid, AirBox, [ 1.1 sub y, 1.1 sub y, 1.1 sub y ], [ sub y, sub y, sub y ], mm, vacuum, true ) ; 122 % Add PML Layer 124 %f p r i n t f ( f i d, omodule. CreatePML Array ( UserDrawnGroup : =, f a l s e, \n ) ; %f p r i n t f ( f i d, PMLFaces : =, Array ( , , , , , ), \n ) ; 126 %f p r i n t f ( f i d, C r e a t e C o r n e r O b j s : =, true, \n ) ; %f p r i n t f ( f i d, T h i c k n e s s : =, 8 0mm, RadDist : =, 2 0mm, \n ) ; 128 %f p r i n t f ( f i d, UseFreq : =, true, MinFreq : =, GHz ) \n ) ; 130 % Add a S o l u t i o n Setup. %h f s s I n s e r t S o l u t i o n ( f i d, Setup13 56MHz, / 1 e9 ) ; 132 %h f s s I n t e r p o l a t i n g S w e e p ( f i d, Sweep10to20MHz, Setup13 56MHz, / 1 e9, / 1 e9, ) ; 134 % Save t h e p r o j e c t t o a temporary f i l e and s o l v e i t. 136 %h f s s S a v e P r o j e c t ( f i d, t m p P r j F i l e, t r u e ) ; %h f s s S o l v e S e t u p ( f i d, Setup13 56MHz ) ; 138 % Export t h e d a t a 140 %hfssexportnetworkdata ( f i d, tmpdatafile, Setup13 56MHz, Sweep10to20MHz ) ; 142 % C l o s e t h e HFSS S c r i p t F i l e. f c l o s e ( f i d ) ; 144 disp ( VBS f i l e ready. ) ; 146 %h f s s E x e c u t e S c r i p t ( h f s s E x e P a t h, t m p S c r i p t F i l e ) ; Page 207

224 Page 208

225 Appendix D MATLAB Code for Inductance Calculation This Appendix includes all the MATLAB code required to compute the inductance of planar metal track. Section D.1 is the core of the program. It will accept parameters which will define the dimensions of a square loop antenna, including number of turns. It will then construct a square loop antenna using segment of straight planar track. The inductance of each planar track will then be computed using the function shown in Appendix D.2. Positive and negative mutual inductances between each track are computed by functions shown in Appendix D.3 and Appendix D.3 respectively. The total inductance is computed by adding the inductance of each track with the total positive mutual inductances and subtracting the total negative mutual inductances. Section D.2 calculates the inductance of a track, not including the positive and negative mutual inductances. Section D.3 calculates the positive mutual inductance between all the tracks. Section D.4 calculates the negative mutual inductance between all the tracks. D.1 Main Code c l e a r a l l ; 2 %i n cm global temp length ; 4 global d e l t a ; Page 209

226 D.1 Main Code global length ; 6 global length1 ; 8 width = 0. 1 ; gap = ; 10 length =1; length1 =2; 12 t h i c k = ; n=1; 14 d e l t a =gap+width ; 16 % c r e a t e c o i l Count=n 4+1; 18 % Setup 20 S t a r t L o c =[0, length, 0 ] ; Curr Loc= S t a r t L o c ; 22 Ct =1; 24 while ( true ) 26 i f Ct==1 %1 s t Arm 28 temp length ( Ct ) =length1 ; d i r e c t i o n ( Ct ) =1; 30 e l s e i f Ct>=Count 32 %f i n a l arm temp length ( Ct ) =(n ) width+length1 +(n 1) gap ; 34 d i r e c t i o n ( Ct ) =1; break ; 36 else %t o p arm 38 temp length ( Ct ) = ( ( Ct 1)/2) width+2 length1 +( Ct 3)/2 gap ; d i r e c t i o n ( Ct ) =1; 40 end Page 210

227 Appendix D MATLAB Code for Inductance Calculation 42 Ct=Ct +1; 44 %r i g h t arm temp length ( Ct ) =( Ct /2) width+2 length +( Ct/2 1) gap ; 46 d i r e c t i o n ( Ct ) =2; 48 Ct=Ct +1; 50 %down arm temp length ( Ct ) = ( ( Ct 1)/2) width+2 length1 +( Ct 3)/2 gap ; 52 d i r e c t i o n ( Ct ) =3; Ct=Ct +1; 54 %l e f t arm 56 temp length ( Ct ) =( Ct /2) width+2 length +( Ct/2 1) gap ; d i r e c t i o n ( Ct ) =4; 58 Ct=Ct +1; 60 end 62 %c a l c u l a t e t h i n f i l m i n d u c t a n c e L 0 64 for i =1: Count 66 t h i n f i l m ( i ) =induc ( temp length ( i ), width, t h i c k ) ; 68 end 70 L=sum( t h i n f i l m ) ; 72 %Compute ( + ) mutual p a i r temp ct = 1 ; 74 for i =1: Count 4 76 j = i ; 78 Page 211

228 D.1 Main Code while j <=Count 4 80 j = j +4; p pair ( temp ct, 1 ) = i ; 82 p pair ( temp ct, 2 ) = j ; temp ct=temp ct + 1; end end 88 %Compute ( ) mutual p a i r temp ct = 1 ; i =Count ; while i >0 90 for j =1: Count ; 92 i f d i r e c t i o n ( j ) ==3 n pair ( temp ct, 1 ) = i ; 94 n pair ( temp ct, 2 ) = j ; temp ct=temp ct + 1; 96 end 98 end i =i 4; 100 end 102 i =Count 3; 104 while i >0 106 for j =1: Count ; i f d i r e c t i o n ( j ) ==4 108 n pair ( temp ct, 1 ) = i ; n pair ( temp ct, 2 ) = j ; 110 temp ct=temp ct + 1; end 112 end i =i 4; 114 end Page 212

229 Appendix D MATLAB Code for Inductance Calculation 116 %compute p p a i r sum p pair =0; 118 for i =1: size ( p pair ) 120 arm1=p pair ( i, 1 ) ; 122 arm2=p pair ( i, 2 ) ; 124 sum p pair=sum p pair+mutual pair pos ( arm1, arm2 ) ; end %compute n p a i r sum n pair =0; 130 for i =1: size ( n pair ) 132 arm1=n pair ( i, 1 ) ; 134 arm2=n pair ( i, 2 ) ; sum n pair=sum n pair+mutual pair neg ( arm1, arm2 ) ; 136 end 138 L ; 2 sum n pair ; sum p pair ; L t o t a l =L+2 ( sum p pair/1000 sum n pair /1000) ; D.2 Inductance Calculation %C a l c u l a t e i n d u c t a n c e o f t h i n f i l m i n d u c t o r 2 % %l e n g t h in cm 4 %width in cm %t h i c k in cm 6 % 8 function x = induc ( length, width, t h i c k ) Page 213

230 D.3 Mutual Inductance - Positive 10 l =length ; w=width ; t = t h i c k ; 12 temp =0.002 l ( log (2 l /(w+ t ) ) (w+ t ) /3/ l ) ; 14 x=temp ; D.3 Mutual Inductance - Positive %C a l c u l a t e Mutual I n d u c t a n c e f o r a p a i r o f s t r i p s 2 % %l e n g t h in cm 4 %gmd in cm % 6 function x = mutual pair pos ( arm1, arm2 ) 8 global temp length ; 10 global d e l t a ; 12 l t h 1 =temp length ( arm1 ) ; l t h 2 =temp length ( arm2 ) ; 14 %a v e r a g e and compute s i d e d i s t a n c e 16 s i d e d i f f =abs ( lth1 l t h 2 ) /2; 18 %f i n d t h e s h o r t e r l e n g t h min length=min( lth1, l t h 2 ) ; 20 %compute gmd v a l u e 22 gmd temp=abs ( arm1 arm2 ) /4 d e l t a ; 24 i f s i d e d i f f ==0 x=mutual ( min length, gmd temp ) ; 26 else x=mutual ( min length+ s i d e d i f f, gmd temp ) mutual ( s i d e d i f f, gmd temp ) ; 28 end Page 214

231 Appendix D MATLAB Code for Inductance Calculation D.4 Mutual Inductance - Negative %C a l c u l a t e N e g a t i v e Mutual I n d u c t a n c e f o r a p a i r o f s t r i p s 2 % %l e n g t h in cm 4 %gmd in cm % 6 function x = mutual pair neg ( arm1, arm2 ) 8 global temp length ; 10 global d e l t a ; global length ; 12 global length1 ; 14 l t h 1 =temp length ( arm1 ) ; l t h 2 =temp length ( arm2 ) ; 16 %a v e r a g e and compute s i d e d i s t a n c e 18 s i d e d i f f =abs ( lth1 l t h 2 ) /2; 20 %f i n d t h e s h o r t e r l e n g t h min length=min( lth1, l t h 2 ) ; 22 %compute gmd v a l u e 24 gmd temp=abs ( arm1 arm2 ) /4 d e l t a ; 26 % adding gap t o gmd % i f odd 28 i f rem ( arm1, 2 ) ==1 gmd temp=gmd temp+2. length ; 30 else gmd temp=gmd temp+2. length1 ; 32 end 34 i f s i d e d i f f ==0 36 else x=mutual ( min length, gmd temp ) ; Page 215

232 D.4 Mutual Inductance - Negative 38 end x=mutual ( min length+ s i d e d i f f, gmd temp ) mutual ( s i d e d i f f, gmd temp ) ; Page 216

233 Appendix E Path Loss Experiment E.1 Preliminary Setting Up Procedure 1. Each of the two antennas was mounted on a tripod stand as shown in Fig. E.1. Both antennas were positioned at the same height and the distance from the ground to the centre point of the antenna, h, was measured. The value obtained was h = 1.29 m. Figure E.1. Antennas on tripods. 2. A signal of 10 dbm at a frequency of 902 MHz from the HP ESG-3000A Signal Generator was fed into one antenna. While 2 W ERP is equivalent to +33 dbm, 10 dbm was a convenient output power, well within the capability of the Signal Generator. In addition, while the frequency of operation in Europe is from 865 Page 217

234 E.2 Experiment Procedure and Results to 868 MHz, the antennas that we had available were rated for operation from 902 to 928 MHz. The lower 902 MHz was chosen so as to be within the operating parameters of our antennas and be as close to the European frequency band. 3. The spectrum analyser was set to a range of 900 to 904 MHz. The other antenna was connected to the HP 8594E Spectrum Analyser with a 10 db attenuator at the input to the spectrum analyser. E.2 Experiment Procedure and Results E.2.1 Map Map showing the locations of measurements: To EM building Antenna (Transmitter) To EM building Antenna (Transmitter) RFID Lab Office RFID Lab Office Antenna (Receiver) Office Antenna (Receiver) Office (a) (b) Figure E.2. Measuring signal strength (a) within a room, and (b) between rooms in Engineering North building. E.2.2 Signal strength at different distance (within and between buildings) 1. With the antennas setup as described above, the second antenna (connected to the spectrum analyser) was moved around various points in the building, with propagation through walls and typical building materials. The second antenna was also moved to other nearby buildings. (Please refer to previous for the path Page 218

235 Appendix E Path Loss Experiment To Maths building To EM building Antenna (Transmitter) RFID Lab Office ENGINEERING NORTH (West Side) Path taken when measuring signal strength Office To Engineering South Figure E.3. Path taken when measuring signal strength from the west side of Engineering North building. To EM building Antenna (Transmitter) RFID Lab Office ENGINEERING NORTH (East Side) Office Path taken when measuring signal strength Figure E.4. Path taken when measuring signal strength from the east side of Engineering North building. Antenna (Transmitter) RFID Lab From Engineering North ENGINEERING & MATHS (EM) BUILDING Path taken when measuring signal strength Figure E.5. Path taken when measuring signal strength from Engineering and Maths (EM) building. Page 219

236 E.2 Experiment Procedure and Results taken when measuring the signal strength.) For all measurements, the orientation of Antenna 1 was adjusted so as to point approximately towards the expected location of Antenna 2, see map in previous section. 2. The strength of the received signal was recorded and the straight-line distance from the transmitting antenna to the receiving antenna was approximated at each point. ( ) 3. The free space path loss was calculated using the equation, P loss = 20 log 4πd 10 l, where d is the distance between the transmitting and receiving antennas and l is calculated to be m for a frequency of 902 MHz. Considering a signal of amplitude 10 dbm, antennas of gain 8 dbi each, cable loss of 1.5 dbm (on both the transmitter and the receiver sides respectively), a 10 db attenuation from the attenuator, and using the free space path loss obtained, the expected signal strength at various points was calculated. 4. The measured signal strengths obtained from various points were then compared with their respective calculated (expected) signal strength. The results are shown in Table E.1 below. The graph of measured and calculated signal strengths against distance, d, was plotted and is as shown below in Fig. E.6. Comments on results: From the results in Table E.1, it can be observed that the measured strength of the received signal was stronger on the east side of the Engineering North building as compared to the west side. The reason a weaker signal strength was detected on the west side may be due to the reason that an anechoic chamber is located in between the propagation path of the signal. It can also be observed from the results that, when measurement was done in the east end of the Engineering and Maths (EM) building, the strength of the received signal was slightly better compared to the measurement taken in front of the lift in that same building. This is because, in the east end, there was an open space in the middle of the signal propagation path, and lots of glass windows, as opposed to brick walls. Page 220

237 Appendix E Path Loss Experiment Table E.1. Measured and calculated signal strengths for various locations. d is the distance between antennas; P cal is the calculated received power; P mea is the measured received power. d (m) P cal (dbm) P mea (dbm) Location/Comments Within room (N203) Within room (N203) Between rooms (N203 - N220b) West of engineering north building. Attenuator pad removed Further west of engineering north building Further west of engineering north building. Frequency span of spectrum analyser reduced to to MHz.** Further west of engineering north building Further west of engineering north building East of engineering north building Further east of engineering north building Further east of engineering north building At north of EM building (3rd Floor, in front of lift) At north of EM building (3rd Floor, east end, propagation through glass and open area) At north of EM building (4th Floor, in front of lift, propagation through floor) At north of EM building (4th Floor, east end, propagation through glass, open area and floor) Page 221

238 E.2 Experiment Procedure and Results Comparison Between Calculated and Measured Received Signal Strength In Building Received Signal Strength (dbm) Calculated Measured Straight line approximations Distance of Separation Between Antennas, d (m) Figure E.6. Comparison between calculated and measured received signal strength (in building). E.2.3 Reflection from typical wall and a conductive fence 1. The antennas were setup according to the description in Part (A) above. The settings for the signal generator remained the same, and the spectrum analyser was set to a frequency range of to MHz with the 10 db attenuator at the input removed. 2. Both antennas were arranged and positioned to face the wall (Fig. E.7) and the strength of the received signal was measured. It has to be noted that the antennas should not be placed too close to each other to avoid cross coupling. 3. Step (2) was repeated except that this time, both the antennas were positioned to face to the same point on the wall (Fig. E.8). 4. Step (2) was again repeated, but this time with the antennas facing a conductive fence instead of the wall (Fig. E.9). Page 222

239 Appendix E Path Loss Experiment Transmitting Antenna WALL Receiving Antenna (a) (b) Figure E.7. Antennas facing directly to the wall. Transmitting Antenna WALL Receiving Antenna (a) (b) Figure E.8. Antennas facing directly to the wall. Figure E.9. Antennas on tripods. 5. The results are as shown in Table E.2 below. Page 223

240 E.2 Experiment Procedure and Results Table E.2. Measured signal strength for antennas facing a wall or a conductive fence. D s is the straight line distance between antennas; D p is the perpendicular distance from wall (or fence); P rec,d is the measured received power when antennas are pointing directly to wall; P rec,0 is the measured received power when antennas are pointing at same point on wall. D s D p P rec,d P rec,0 Antennas facing wall (1st) Antennas facing wall (2nd) Antennas facing conductive fence E.2.4 Propagation Loss Outdoors 1. This part of the experiment was performed outdoor, on the lawn in front of the Mathematics building. 2. The antennas were setup in a similar way. The settings for the signal generator remained the same, and the spectrum analyser was set to a frequency range of to MHz with the 10 db attenuator at the input removed. 3. The antennas were arranged and positioned to face each other (Fig. E.10) and the distance of separation between the two antennas, d, was measured. It was found that d = 11 m. The height of the antennas from the ground, h, was also measured and it was found to be 1.29 m. (a) (b) Figure E.10. Measurement of signal strength in outdoors. Page 224

241 Appendix E Path Loss Experiment Table E.3. Calculated and measured signal strength (outdoor) for different antenna orientations. d is the distance between antennas; P rec is calculated received power; P mea is the measured received power. Antenna orientation Distance d (m) P rec (dbm) P mea (dbm) Antennas facing directly at each other Antennas on the ground, facing upwards The strength of the received signal was measured. 5. The procedures were repeated with both antennas placed on the ground and facing upwards (with the distance, d, maintained the same) (Fig. E.11). Figure E.11. Antennas on the ground and facing upwards. 6. The recorded results are as shown in Table E.3 below. E.3 Conclusion As expected, the measured values of the received signal strength were found to be quite different from the calculated values, as obtained from the Free Space Loss Equation. Fortunately, the observed path loss was greater than the path loss calculated in free space, due to building materials and objects in various rooms. Our transmitter operated at +10 dbm, whereas RFID transmitters are allowed to operate at 2 W ERP, or 35 dbm. The path loss required would then be 25 db greater than Page 225