DEVELOPMENT AND APPLICATIONS OF SENSOR INTEGRATED

HYBRID RFID SYSTEM

By

Saikat Mondal

A DISSERTATION

Submitted to

Michigan State University

in partial fulfillment of the requirements

for the degree of

Electrical and Computer Engineering — Doctor of Philosophy

2021

ABSTRACT

DEVELOPMENT AND APPLICATIONS OF SENSOR INTEGRATED HYBRID RFID

SYSTEM

By

Saikat Mondal

Today, Radio Frequency Identification (RFID) has become a multibillion-dollar industry.

RFID is primarily used for numerous object tagging and tracking applications such as road

toll collection service, facility access control, asset tracking in a supply chain to name a few.

Batteryless tag architecture and modern fabrication technology has enabled the miniatur-

ization and cost reduction of ultra-high frequency (UHF) passive RFIDs. There is growing

interest in the use of RFIDs for supply chain sensors (e.g., agricultural products and pharma-

ceuticals), underground object tagging (e.g., plastic pipes), low-power wearable devices, etc.

However, the RFID designs in their current form cannot meet the stringent requirements

of these new applications. Four key challenges that hinders the direct adoption of existing

RFIDs for these applications are: (a) clutter effect, (b) integration of sensors, (c) response

time, and (d) prolonged RF transmission from RFID reader for continuous sensor monitor-

ing. In this work, the fundamental limitations of conventional RFID system are described

in detail first, followed by proposed solutions leading to new RFID designs.

First, a dual frequency harmonic RFID system is proposed and demonstrated to mitigate

the clutter effect. Second, a low power digital interface sensing platform is demonstrated

for electrochemical pH sensor, targeted towards biochemical applications as an example.

Third, detailed analysis is performed at the component level to understand the efficiency

and response time dependence of the energy harvester within the RFID designs. Based on

the analysis, a model was proposed to estimate the response time of a conventional RFID

tag. Fourth, a dual mode RFID is proposed and demonstrated to reduce the transmit time of

RFID reader, hence reducing the effective RF transmission. Furthermore, the compatibility

of integrating these solutions together in a single platform are presented. This is important

as more than one challenge can be present in a single application. Finally, a hybrid RFID

configuration is proposed and demonstrated that is capable of simultaneously mitigating all

these four challenges.

This dissertation is dedicated to my beloved family. . .

iv

ACKNOWLEDGMENTS

I am grateful to a number of people who has contributed to help me complete my PhD

study at Michigan State University. First and foremost, I would like to thank one of the

essential contributors to this thesis, my advisor Dr. Premjeet Chahal for providing me the

privilege to be his student and then for his constant guidance and support throughout my

Ph.D. journey, for which I am forever indebted.

I also thank him for introducing me to

a broad spectrum of research ideas and providing me the flexibility to pursue them over

the years. Next, I am thankful to the rest of my committee members Dr. Yiming Deng,

Dr. Jeffrey A. Nanzer, and Dr. Arun Ross for their valuable suggestions and feedback

through my PhD work. I would also like to thank Dr. Edward J. Rothwell, Dr. Shanker

Balasubramaniam, and Dr. Tim Hogan for providing me valuable insights in the various

courses and encouraging me professionally.

I would like to thank the Department of Transportation (DoT), The AXIA Institute,

Michigan Translational Research and Commercialization (MTRAC) Innovation Hub, and

Ford Motor Company for sponsoring my research and making all this work possible. Special

mention for Dr. Mahmoud Ghannam, Mark Cuddihy, Dr. Chris Oakley, and Dr. Ali Attaran

from Ford, for their suggestions, which shaped many parts of this thesis. I would also like to

thank MSU College of Engineering for providing me multiple fellowships and travel awards

throughout my PhD program.

I appreciate all ECE department administrators and staff members for their help, spe-

cially Mr. Brain Wright, Mr. Gregg Mulder, Mr. Cameron Crump from MSU ECE shop, and

Mr. Karl Dersch from ECE cleanroom, without whose support most of the work presented

in this thesis would not be possible.

v

I would like to thank my fellow colleagues and alumni from the Electromagnetics Research

Group (EMRG). I am thankful to our previous members Dr. Amanpreet Kaur and Dr. Mohd

Ifwat Mohd Ghazali, Dr. Jennifer Byford for all their wisdom and advices. Further, I would

like to mention about my special friends and co-workers, who made my PhD journey a

memorable one: Dr. Saranraj Karuppuswami, Dr. Suhas Vidhate, and Deepak Kumar.

I would like to mention my other colleagues and friends Yihang Chu, Yamini Kotriwar,

Kanishka Wijewardena, Pratik Chatterjee, Dr. Yuxiao He, Dr. Michael Craton, Dr. Vinny

Gjokaj, Stavros Vakalis, Serge Mghabghab, Anton Schlegel, Omkar Ramachandran, Abdel

Alsnayyan, Zane Crawford, Dr. Scott OConnor, Dr. Sean Ellison, Eric Klinefelter, Elliot

Lu and everyone else in EMRG. I would like to thank my friends Dhrubajit, Atri, Anmol,

Rahul, Ritesh, Kunal, and my old kgp buddies Debashis, Shobhit, Sourin, and Subhosit.

Finally, I would like to thank my parents specially my mother for all their endless love,

support and motivation not only through my PhD but also throughout my life. I am also

thankful to Oindrilla Dutta for all her continuous support during my PhD journey.

vi

TABLE OF CONTENTS

LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

x

LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xi

Chapter 1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.1 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.1.1 Electronic toll system . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.1.2
Supply chain application . . . . . . . . . . . . . . . . . . . . . . . . .
1.1.3 Financial banking . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Sensing with identification . . . . . . . . . . . . . . . . . . . . . . . .
1.1.4
1.2 RFID System and Components
. . . . . . . . . . . . . . . . . . . . . . . . .
1.3 Backscatter Modulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.3.1 Coupled antenna impedance . . . . . . . . . . . . . . . . . . . . . . .
1.4 Thesis Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 2 Challenges with Conventional RFIDs . . . . . . . . . . . . . . . .
2.1 Clutter and Multipath Effect . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1.1 Clutter formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1.2 Clutter measurement . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2 Sensor Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3 Low Power Operation and response Time . . . . . . . . . . . . . . . . . . . .
2.4 Prolonged RF transmission from reader . . . . . . . . . . . . . . . . . . . . .
2.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Simulation and measurement

3.3.1 Antennas

Chapter 3 Harmonic RFID . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2 Nonlinear Transmission Line: NLTL . . . . . . . . . . . . . . . . . . . . . . .
3.2.1 Design principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
3.2.2
3.3 Harmonic RFID: Gen I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3.1.1 Meandered dipole antenna . . . . . . . . . . . . . . . . . . .
3.3.1.2 Yagi-Uda antenna . . . . . . . . . . . . . . . . . . . . . . .
3.3.2 Harmonic tag design . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3.2.1 HD using NLTL . . . . . . . . . . . . . . . . . . . . . . . .
3.3.2.2 Energy Harvesting Unit (EHU) . . . . . . . . . . . . . . . .
3.3.2.3 Digital Modulation Unit (DMU)
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
3.3.2.4 Power Splitter (PS)
3.3.3 Reader design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . .
3.3.4 Wireless measurement
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.4 Harmonic RFID: Gen II

1
3
3
4
6
7
8
10
12
14

16
17
19
21
22
26
29
30

31
31
34
35
40
42
42
45
46
47
47
50
51
53
58
59
60

vii

3.4.1 Antenna . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.4.2 Tag circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.4.3 Wireless measurement
. . . . . . . . . . . . . . . . . . . . . . . . . .
3.5 Harmonic RFID Tag Efficiency . . . . . . . . . . . . . . . . . . . . . . . . .
3.6 Harmonic RFID Performance Under Clutter . . . . . . . . . . . . . . . . . .
3.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 4 RFID With Integrated Sensor . . . . . . . . . . . . . . . . . . . .
4.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2 Design and Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.1 RF interrogator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.2 Readout circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.3
pH sensor fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3.1 Electrode performance . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3.2 Elemental analysis of sensor electrodes . . . . . . . . . . . . . . . . .
4.3.3 Multiple sensor electrode response . . . . . . . . . . . . . . . . . . . .
4.3.4 Readout circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . .
4.3.5 Wireless measurement
4.4 Application: Wearable Mouthguard . . . . . . . . . . . . . . . . . . . . . . .
4.4.1 Antenna design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.4.2 Mouthguard design . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

61
65
68
70
71
74

77
78
81
81
82
83
86
86
89
91
91
93
94
96
97
99

Chapter 5 Energy Harvester Modeling . . . . . . . . . . . . . . . . . . . . . . 101
5.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
5.2 Modified Dickson Charge Pump . . . . . . . . . . . . . . . . . . . . . . . . . 103
5.3 Effect of Matching Network . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
5.3.1
Single-stage formulation . . . . . . . . . . . . . . . . . . . . . . . . . 107
5.3.2 Multi-stage formulation . . . . . . . . . . . . . . . . . . . . . . . . . 109
5.4 Transient Model for Arbitrary Load . . . . . . . . . . . . . . . . . . . . . . . 109
5.5 Measurement Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
5.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

Chapter 6 Effective Transmit Time Reduction of RFID Reader . . . . . . 115
6.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
6.2 Finite State Machine and Event Driven Modes . . . . . . . . . . . . . . . . . 118
6.2.1 M0 state: Tag backscatter communication . . . . . . . . . . . . . . . 118
6.2.2 M1 state: Tag active transmit . . . . . . . . . . . . . . . . . . . . . . 119
6.2.3 Events and driven modes . . . . . . . . . . . . . . . . . . . . . . . . . 119
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
6.3.1 Tag design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
6.3.1.1 Energy harvester . . . . . . . . . . . . . . . . . . . . . . . . 122
6.3.1.2 Reader code demodulator
. . . . . . . . . . . . . . . . . . . 123
6.3.1.3 Oscillator . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

6.3 System Architecture

viii

6.3.1.4 Charge control module . . . . . . . . . . . . . . . . . . . . . 124
6.3.1.5 Micro controller . . . . . . . . . . . . . . . . . . . . . . . . . 126
6.3.1.6 Antenna and front end . . . . . . . . . . . . . . . . . . . . . 126
6.3.2 Reader design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
6.4 Charging and Discharging Operation . . . . . . . . . . . . . . . . . . . . . . 129
6.4.1 Charging time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
6.4.1.1 Controlled current charging . . . . . . . . . . . . . . . . . . 129
6.4.1.2 Bi-directional charging . . . . . . . . . . . . . . . . . . . . . 131
6.4.2 Discharging time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
6.5 Measurement Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
6.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137

Chapter 7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
7.1 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
7.1.1 Miniaturization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
7.1.1.1 Electrically small antenna . . . . . . . . . . . . . . . . . . . 141
7.1.2
Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
7.1.3 Packaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143

APPENDICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
APPENDIX A Modulation Depth Formulation of RFID Tag Impedance . . . . . . 145
APPENDIX B Copper Etching Process . . . . . . . . . . . . . . . . . . . . . . . . 159

BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161

ix

LIST OF TABLES

Table 1.1: Advancement of RFID technology through decades.

. . . . . . . . . . . .

2

Table 3.1: The dimension of different antenna designs

. . . . . . . . . . . . . . . . .

43

Table 3.2: Comparison with other work reported in literature. . . . . . . . . . . . . .

58

Table 4.1: Performance comparison with other works.

. . . . . . . . . . . . . . . . .

99

Table 5.1: Comparison with relevant literature.

. . . . . . . . . . . . . . . . . . . . . 113

Table 6.1: Current consumption branches. . . . . . . . . . . . . . . . . . . . . . . . . 134

x

LIST OF FIGURES

Figure 1.1: Widely used RFID for toll system (a) an EZ-Pass tag, (b) inside of an
. . . . . . . . . . . . .

EZ-Pass, and (c) the inside circuit of an EZ-Pass.

Figure 1.2: Low cost example UHF RFID tags for supply chain applications.

. . . .

Figure 1.3: Contactless credit card with NFC reader. . . . . . . . . . . . . . . . . . .

Figure 1.4: A commercial moisture sensor with interdigited capacitance for sensing. .

Figure 1.5: A typical RFID system with reader and tag. . . . . . . . . . . . . . . . .

Figure 1.6: RFID tag antenna and IC at the center.

. . . . . . . . . . . . . . . . . .

Figure 1.7: Block diagram of RFID internal circuit.

. . . . . . . . . . . . . . . . . .

3

5

6

8

8

9

9

Figure 1.8: Two cylindrical dipole antennas with different dimensions.

. . . . . . . .

12

Figure 1.9: Normalized scattered impedance change at the reader antenna for different
cases according to Equation (A.23). Normalization was performed with
respect to case I.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 1.10: Normalized percentage change in impedance at the reader antenna for
the same six different cases as shown in Fig. 1.9. Normalization was
performed with respect to case I.
. . . . . . . . . . . . . . . . . . . . . .

Figure 2.1: Different sources of clutter (a) self-jamming [1], (b) multi-reader jamming
. . . . . . . . .

[1], (c) reflection from human body in a BAN sensor [2].

13

14

17

Figure 2.2: Single frequency RFID tag and background reflector.

. . . . . . . . . . .

19

Figure 2.3:

(a) Tag modulated signal, (b) Received tag signal at the reader in absence
of any reflector and (c) Received tag signal at the reader in presence of a
reflector. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

20

xi

Figure 2.4: Front-end of a standard single frequency RFID reader.

. . . . . . . . . .

23

Figure 2.5: Phase noise after FFT analysis for cases when the reader antenna is (a)
25 cm, (b) 40 cm away from the clutter source, and (c) 40 cm away from
an absorber. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 2.6: Wireless and batteryless sensing system: (a) a pH sensor with sensor
output embedded in phase information [3]; (b) a food volatile sensor with
sensor output embedded in the resonance frequency shift [4]. . . . . . . .

Figure 2.7: Example of sensor integrated with conventional RFID IC: (a) the sensing
element is integrated on top of the antenna [5]; (b) the sensor tag antenna
and in the inset sensing element integrated in between antenna and RFID
IC [6].
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

23

24

25

Figure 2.8: Schematic block diagram of the RFID tag during energy harvest.

. . . .

27

Figure 2.9: Analysis on energy harvest at the tag vs. distance with varying consumption. 28

Figure 2.10: (a) RF radiation on human body, and (b) Desired effective RF radiation
. . . . . . . . . . . . . . . . . . . . . . .

time reduction from the reader.

Figure 3.1:

(a) Harmonic phase based localization in presence of a scatterer [7] (b)
Harmonic phase based sensing with better immunity to background clutter
[3]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

30

32

Figure 3.2: Fundamental and harmonic antenna setup in between the tag and reader.

34

Figure 3.3:

(a) Example of NLTL: (a) Discrete diode with CPW line [8] (b) Discrete
inductors and diodes shown at the top [9].
. . . . . . . . . . . . . . . . .

Figure 3.4: Lumped element representation of the NLTL with a single cell NLTL at
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

the bottom.

35

37

Figure 3.5:

(a) S11 and (b) S21 for different number of NLTL stages. . . . . . . . . .

40

Figure 3.6:

(a) S11 and (b) S21 for three number of NLTL stages at different DC bias. 41

xii

Figure 3.7:

(a) S11 and (b) S21 for four number of NLTL stages at different DC bias.

41

Figure 3.8: Antenna structures for (a) Meandered dipole (434 MHz) for tag side, (b)
Meandered dipole (868 MHz) for tag and interrogator sides and (c) Yagi-
. . . . . . . . . . . . . . . . . . . .
Uda (434 MHz) for interrogator side.

Figure 3.9: Effect of T-match on input impedance shown on Smith chart from 100
MHz to 600 MHz. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

43

44

Figure 3.10: Radiation pattern of the Yagi-Uda with and without the back reflector. .

44

Figure 3.11: Simulation and measurement results for (a) 434 MHz meandered dipole,
. .

(b) 434 MHz Yagi-Uda and, (c) 868 MHz meandered dipole antenna.

45

Figure 3.12: Harmonic tag architecture with different components. . . . . . . . . . . .

47

Figure 3.13: A three stage nonlinear transmission line (NLTL) with matched termina-
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

tion at 434 MHz.

Figure 3.14: Measured second harmonic output power of a three stage NLTL as func-
. . . . . . . . .

tion of DC bias for input power of -10 dBm at 434 MHz.

Figure 3.15: Measured reflection coefficient of the matched termination with and with-
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

out 2f0 antenna.

Figure 3.16: 4-stage charge pump circuit with matching network and DC voltage reg-
ulator. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

48

49

50

51

Figure 3.17: The unregulated and regulated DC voltage generated by the EHU.

. . .

52

Figure 3.18: Input impedance of the EHU with marker at 434 MHz. . . . . . . . . . .

52

Figure 3.19: The different stages of signal generated by the DMU. Description from
top to bottom: 1) Serial output of the shift register, which is the ID, 2)
LD signal, and 3) Clock signal.
. . . . . . . . . . . . . . . . . . . . . . .

54

Figure 3.20: The D flip-flop setup to generate the LD signal from the Clock signal. . .

54

xiii

Figure 3.21: DC power consumption and harvest by HD, DMU, and EHU independently. 55

Figure 3.22: Total power consumption and harvest for equal and unequal power splitting. 56

Figure 3.23: Power splitter design with unequal power division. . . . . . . . . . . . . .

56

Figure 3.24: The tag circuit photograph with different components as compared to a
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

penny.

57

Figure 3.25: Integrated interrogator system.

. . . . . . . . . . . . . . . . . . . . . . .

59

Figure 3.26: Demodulated received signal (a) after digitization, (b) after averaging,
. . . . . . . . . . . . . . . .

and (c) after pulse shaping and bit recovery.

Figure 3.27: (a) The block diagram and (b) fabricated part of the single antenna har-
monic tag. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

60

62

Figure 3.28: Dual frequency designed antenna. All the dimensions are in mm.

. . . .

63

Figure 3.29: The reflection co-efficient of the dual frequency antenna.

. . . . . . . . .

63

Figure 3.30: The step by step procedure for dual frequency single antenna design.

. .

64

Figure 3.31: Measured results for (a) S11 and (b) S21 for different number of NLTL
stages at no bias condition; (c) S11 and (d) S21 for two stage NLTL at
different bias.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 3.32: Measured results for (a) conversion loss and (b) DC power consumption
. . . . . . . . . . . . . . . . . . . .

for different number of NLTL stages.

Figure 3.33: Measured reflection co-efficient of the energy harvester with and without
NLTL. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

66

67

68

Figure 3.34: Measured harmonic voltage at different distance and modulation frequency. 69

Figure 3.35: Efficiency calculation block diagram of harmonic RFID for (a) harmonic
. . . . . . . . . . . .

efficiency and (b) energy harvester of charge pump.

70

xiv

Figure 3.36: Measurement setup for performance comparison.

. . . . . . . . . . . . .

72

Figure 3.37: Measurement setup with commercial RFID tag at the center.

. . . . . .

72

Figure 3.38: Correlation in between transmitted power and tag detection count.

. . .

72

Figure 3.39: Power profile for 10 dBm transmitted power at 926 MHz with (a) no tag
.

or reflector, (b) only tag is present, and (c) tag in presence of reflector.

73

Figure 4.1: High level picture of analog sensor integration with ADC and microconroller. 78

Figure 4.2:

Interaction of a handheld reader with identical pH sensors with different
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
IDs.

Figure 4.3: The block diagram of the measurement setup with an interrogator and
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

the RFID sensor.

80

82

Figure 4.4: The sensor voltage readout digital circuit with ID generation.

. . . . . .

83

Figure 4.5: Fabrication procedure of pH combination electrodes a) Cu plated RT/duroid

5880 substrate; b) patterning of Cu layer to create ‘T’ shaped Cu backbone
interconnect; c) sputtering followed by lift-off of Sb metal; d) oxidation of
Sb layer; e) sputtering followed by lift-off of Ag metal; f) chlorination of
Ag layer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

85

Figure 4.6: Voltage response of the fabricated electrode in different pH solutions.

. .

86

Figure 4.7: Stability of the voltage response of the electrode in different pH solutions.

87

Figure 4.8: Repeatability test of the electrode in different pH solutions.

. . . . . . .

87

Figure 4.9: Locations for EDX elemental characterization: (a) reference electrode with
highlighted black box for Ag/AgCl layer and red box for Nafion layer, (b)
pH sensitive electrode with highlighted black box for Sb/Sb2O3 layer. . .

89

Figure 4.10: Elemental characterization for Ag/AgCl layer. . . . . . . . . . . . . . . .

90

xv

Figure 4.11: Elemental characterization for Nafion layer.

. . . . . . . . . . . . . . . .

90

Figure 4.12: Elemental characterization for Sb/Sb2O3 layer.

. . . . . . . . . . . . . .

90

Figure 4.13: Voltage response of the three different fabricated electrode in different pH
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

solutions.

Figure 4.14: From top to bottom: a) The ID bit ‘101101001’ b) 8 bits ADC output
‘10010000’ of analog bias 1 V, and c) combined ID and ADC output bits.
. . . . . .
The ’idle time’ in between the bit streams is also highlighted.

Figure 4.15: The complete sensor tag with zoomed in view of the electrode pair in the
In the inset, the left electrode is
. . . .

inset. The scale is shown in inches.
Nafion coated Ag/AgCl and the right one is Sb/Sb2O3 electrode.

Figure 4.16: Digital bit response of the tag at different pH solutions in wireless setup.
From top to bottom: a) bit response of ‘111010’ at pH 4.3, b) bit response
of ‘1001100’ at pH 7, and c) bit response of ‘1010110’ at pH 8.6. . . . . .

92

93

94

95

Figure 4.17: Antenna dimensions for (a) inside human mouth, and (b) in air medium.

96

Figure 4.18: Simulated reflection co-efficient for in-air and in-mouth antenna designs.

97

Figure 4.19: The sensor tag integrated on mouthguard (a) bare mouthguard, (b) dig-
ital part of the transponder at the right side and (c) RF circuit of the
transponder at the left side of the mouthguard.
. . . . . . . . . . . . . .

98

Figure 4.20: Detected signal in Oscilloscope in buffer pH solution of 5.3. . . . . . . . .

98

Figure 4.21: Detected signal in Oscilloscope in buffer pH solution of 9.5. . . . . . . . .

98

Figure 5.1:

(a) The block diagram of a wireless batteryless system with a multi-stage
modified Dickson charge pump; (b) Measurement setup with a fabricated
5-stage charge pump. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103

Figure 5.2:

(a) A single stage CP, and it’s equivalent circuit for (b) negative half cycle,
and (c) positive half cycle. . . . . . . . . . . . . . . . . . . . . . . . . . . 103

xvi

Figure 5.3: Effect of pump capacitor CP on steady state voltage.

. . . . . . . . . . . 105

Figure 5.4:

(a) A Schottky diode model with pad parasitics. (b) ’L-match’ as two-
element matching network for n-stage CP.

. . . . . . . . . . . . . . . . . 105

Figure 5.5: A typical matching trajectory (dotted line) of single stage CP. For case (a),
the matching network uses series inductor and shunt inductor (X1 > 0,
X2 > 0). For case (b), the matching network uses series inductor and
shunt capacitor (X1 > 0, X2 < 0).

. . . . . . . . . . . . . . . . . . . . . 107

Figure 5.6:

(a) n stage charge pump, and (b) it’s simplified impedance model. . . . . 108

Figure 5.7: Equivalent transient model for (a) single stage CP, and (b) n stage CP. . 110

Figure 5.8:

(a) Extracted pump resistance (RP ) of a single stage CP at different input
power. (b) Model vs. discrete circuit based single stage CP simulation
time comparison.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110

Figure 5.9: Measurement results of reflection co-efficient (S11) (a) single stage CP at

different input power, and (b) multiple stages CP at -10 dBm input power. 112

Figure 5.10: Voltage normalized transient response of CPs with different stages with

Cload = 100 nF and no external Rload.

. . . . . . . . . . . . . . . . . . . 112

Figure 5.11: (a) Voltage normalized transient response of a 2-stage CP at different
load conditions. (b) Measured efficiency of CPs with different stages. (c)
Scaling factor of Trn and Vssn compared to a single stage CP.

. . . . . . 113

Figure 6.1: Dual mode RFID tags for patients’ sleep monitoring system. . . . . . . . 116

Figure 6.2: Different modes of communication in between the reader and tag.

. . . . 117

Figure 6.3: Different sub-modules of the tag circuit.

. . . . . . . . . . . . . . . . . . 121

Figure 6.4: Energy harvester with 3-stage Dickson charge pump configuration. . . . . 122

Figure 6.5: Reader code demodulator to decode reader instruction at the tag.

. . . . 123

xvii

Figure 6.6: Oscillator circuit to generate 900 MHz RF carrier. . . . . . . . . . . . . . 124

Figure 6.7: Charge control module with (a) switches and a 2.2 mF storage capacitor,

and (b) voltage comparator.

. . . . . . . . . . . . . . . . . . . . . . . . . 125

Figure 6.8:

(a) Input and output ports connections of micro controller, and (b) sensor
circuit configuration.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126

Figure 6.9: Antenna dimensions with front and side views. . . . . . . . . . . . . . . . 127

Figure 6.10: Simulated and measured PIFA antenna reflection co-efficient comparison. 127

Figure 6.11: The reader to interrogate the tags in two modes.

. . . . . . . . . . . . . 128

Figure 6.12: Charging of the capacitor using controlled current.

. . . . . . . . . . . . 129

Figure 6.13: VM C and Vcap voltage level for 3 cases: input power is -1 dBm and α = 1,
input power is -3 dBm and α = 1, input power is -3 dBm and α = 0.86
for initial 9.2 s and later α = 1.

. . . . . . . . . . . . . . . . . . . . . . . 132

Figure 6.14: (a) Dual mode RFID tag circuit with SMA, and (b) front and back side
of the PIFA antenna with SMA outlet, compatible to be embedded within
seat.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135

Figure 6.15: Storage capacitor and micro-controller voltage level at different switching

signals.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136

Figure 6.16: Activation of a dual mode tag due to specific code sent from the reader.

136

Figure 6.17: The information received at the reader sent by a dual mode RFID tag.

. 137

Figure A.1: Two cylindrical dipole antennas with different dimensions.

. . . . . . . . 146

Figure A.2: Four-terminal impedance network for two antenna system.

. . . . . . . . 146

Figure A.3: A single center-fed dipole antenna.

. . . . . . . . . . . . . . . . . . . . . 147

xviii

Chapter 1

Introduction

Radio Frequency Identification (RFID) has become an integral part of numerous appli-

cations such as road toll collection service, access control, tracking assets in supply chain,

credit cards, batteryless sensors to name a few [10–12]. RFID system works on a wide range

of RF spectrum starting from 100 kHz to few GHz. Among available RFID bands, the most

popular bands are 125 kHz at low frequency (LF), 13.5 MHz at high frequency (HF), 433

MHz, 868 MHz, 915 MHz at ultra-high frequency (UHF), and lastly 2.4 GHz at super high

frequency (SHF). RFIDs operating at 125 kHz, 13.5 MHz, 868 MHz, 915 MHz are generally

batteryless or passive in nature. Whereas, RFIDs operating at 433 MHz and 2.4 GHz are

usually battery powered or active in nature. Passive tags have limited range of operation as

they can be only powered by the reader, hence distance from the reader will greatly impact

the operation range. On the other hand, active tags can have long range more than 100 m

as the available power at the tag is not a major issue. The UHF RFIDs working in 868 MHz

or 915 MHz band has become the most commercially popular due to long range operation

in passive mode, miniaturized antenna with acceptable radiation efficiency, available band

with high power transmission capability, structured and defined regulatory protocols, low

cost manufacturing capability, and finally longer period of research and development. The

most recent UHF RFID protocol is EPC Gen 2, which primarily outlines the frequency of

operation, effective transmit power limit, two-way modulation specification, data encoding

1

Table 1.1: Advancement of RFID technology through decades.

Decade
1940-1950 Radar refined and used, a major World War II development effort.

Event
Concept of RFID proposed in 1945∼48.

1950-1960 Development of RFID technology, primarily in laboratory.
1960-1970 Development of RFID theory with field trials.
1970-1980 Large companies such as Raytheon, RCA started development of

RFID technology.

1980-1990 Early implementation of RFID started.
1990-2000 RFID started being widely deployed. E-Z pass was formed.
2000-2010

Introduction of EPC Gen 2 protocol, a unified protocol for RFIDs.
Started being implemented for supply chain, animal tracking.
RFID industry worth around $8.5 billion in 2016.

2010-

scheme, timing and memory requirement, instruction set architectures (ISAs), anti-collision

protocols among many others [13]. The development history of RFID technology, as sum-

marized in Table 1.1, is quite fascinating and can be traced back around World War II

era [14].

An important development of RFID tag with practical demonstration was reported from

Los Alamos Scientific Laboratory [15]. A subcarrier was introduced at the tag side to perform

the backscatter modulation of the tag antenna. The subcarrier was introduced to remove

the clutter, which are caused from background reflection. Other notable development of

RFID tags from early days are Raytheon’s Raytag, Richard Klensch’s electronic identification

system from RCA. Apart from the system level design, there was also requirement of CMOS

fabrication technology miniaturization for small circuitry, development of EM theory for

efficient backscatter modulation, development of required software for adoption in large

scale business, development of printing techniques for manufacturing the RFID tags at very

low cost. All of the factors played a major role in creating a multi-billion dollar RFID

industry. Today’s RFID industry has become successful in numerous supply chain, electronic

2

toll system, financial banking, e-passport, library book tracking, building access, inventory

management as well as anti-counterfeiting for different parts in car industry. Each of the

major applications are discussed very briefly in the following section.

1.1 Applications

1.1.1 Electronic toll system

Figure 1.1: Widely used RFID for toll system (a) an EZ-Pass tag, (b) inside of an EZ-Pass,
and (c) the inside circuit of an EZ-Pass.

RFID technology has shaped public transportation largely by making the transportation

system seamless. Among all RFID applications, electronic toll system was the first appli-

cation to be commercially implemented. As an RFID tag embedded car reaches at a toll

booth, the overhead RFID reader scans the specific unique information from the tag. In the

US, electronic toll collection as efficient and effective method that eliminates long lines of

traffic. EZ-Pass is widely used in US and the inside circuit is shown in Fig. 1.1 (c). The

RFID tag works at 900 MHz but requires an external power supply for example an on-board

battery. Hence, the RFID tag in EZ-Pass is an active type.

3

1.1.2 Supply chain application

The major problem in supply chain management today is to locate a weak node in the

chain for proper functioning of the complete network. The problems can be leaving a package

or product behind during loading, misplaced during transportation, or proper storage in

warehouse. There can be multiple reasons for those problems, primarily of them is related to

human error due to manual interaction. Although major advancements in technology have

happened over the past decade, still partially manual driven supply chain is a part of any

modern supply chain.

Using RFID technology, an end to end monitoring and accounting is possible. The added

benefits of using RFID in supply chain are increased product visibility and transparency

with added location accuracy. Hence, RFID technology is proposed for multiple supply chain

sectors such as healthcare, food industry, agriculture etc. [16–18]. Among those supply chain

sectors, in food industry RFID was proposed earlier due to its low cost and small size to

ensure food safety by increasing the traceability of the food products in the food supply chain

[17, 19]. The importance of RFID-based temperature sensor tags for cold food supply chain

has been described in [19]. The sensor enabled tags can record the temperature of individual

food packages and track the real time shelf life in case of any deviations in environmental

conditions during storage or transportation. According to [19], if food products with reduced

shelf lives arrive unexpectedly at a retailer, priority should be given to sell those items

to reduce the amount of food wastage.

In [17], a detailed review has been performed to

discuss the potential of RFID technology in logistic development of different sectors for the

food supply chain. However, the visibility of the food products is limited by the amount

of information shared by each supply chain stage to its next stage.

Implementing RFID

4

technology along with cloud based services would provide a digital trace of the food products

during their lifetime and thus making the food supply chain visible to everyone.

Auto-ID Labs network was initiated in 1999 in the field of networked RFID and emerging

integrated sensing technologies. Auto-ID was responsible to design the architecture for the

Internet of Things (IoT) and Electronic Product Code (EPCglobal) for identification of

the items in the supply chain [20, 21]. The primary objective was to develop hardware,

software, protocol, communication interface, specification, and business research related to

RFID [22, 23]. The business case was very important for the supply chain to succeed as a

huge cost scaling down was necessary, which was demonstrated in [23] making a tag cost

around 5 cents.

Figure 1.2: Low cost example UHF RFID tags for supply chain applications.

In recent days, RFID has huge application in modern days supply chain industry. To

compete with existing barcode or QR code based optical system for object identification,

RFID tag price required to be significantly cheaper. By late 2000, with screen printing and

availability of low cost miniaturized RFID IC, bulk manufacturing of cheap RFID tags was

possible. Re-usability, long read range, and electronic surveillance made possible to monitor

products in a supply chain seamlessly. Embedding the RFID tags within products will also

help to fight counterfeiting as well as maintain inventory in numerous industries [24,25]. The

tags primarily used in supply chain are usually passive and work at UHF band. An example

screen printed RFID tag is shown in Fig. 1.2.

5

1.1.3 Financial banking

Another major application of RFID technology is in financial banking for contactless

transaction. Multiple contactless transaction systems such as Google pay, Android pay,

wireless credit card has been introduced [26, 27]. The technology although based on RFID,

is a little variant due to requirement of enhanced security [28]. More precisely, the technology

is Near Field Communication (NFC) which uses a near field antenna and NFC chip. The

communication range is limited to very short distance to avoid confusion in transactions

and protect the cards from stealing sensitive information from long distance. The NFC

chip along with antenna is embedded within the wireless credit cards, and the smartphones

during manufacturing of the devices. Since, a separate and completely independent chip is

embedded, the communication is independent of SIM card or other wireless connection of

the smartphone. In Fig. 1.3, a NFC credit card with a NFC reader is shown. There can be

few security issues with the NFC based communication [28]. The primary security issues are

1) eavesdropping, 2) data corruption and modification, 3) data insertion, 4) middle-man-

attack. Multiple schemes have been proposed to counter the attacks such as encryption,

token system etc.

Figure 1.3: Contactless credit card with NFC reader.

Usually, the NFC card uses a token system to restrict the use of sensitive data. In the

6

token system, the NFC card sends a token to the NFC reader and the NFC reader sends

the token to Token Service Provider (TSP). The TSP maps the token to find where the card

information is stored and the stored information is fetched from Card Data Vault (CDV).

Once the card transaction is verified, a confirmation is sent back to the NFC reader, which

confirms the transaction. For the contactless cards, no additional pin or identification is used

from the user side, which makes the transaction fast compared to other traditional magnetic

strip or chip based cards.

1.1.4 Sensing with identification

Recently, another interesting topic of research has been the integration of sensors along

with identification. As the RFID ICs already have harvested energy, the power can be used

for additional sensing operation.

In literature, multiple cases of sensor integrated RFID

is demonstrated [29–31]. The conventional RFID architecture primarily relied on system

concepts which exclusively serve one-way and low data rate traffic, and was designed for

sending short message consisting only ID and was never intended for streaming sensor data.

Proposed RFID sensor network design required extra functionality and processing power at

the tag. Also, as the harvested power is limited at the passive tags, the read range would be

limited for higher data-rate sensor data transmission.

The architecture of the sensor integrated RFID is more or less similar with the following

components: 1) the RF front end including the antenna and the circuity for rectification

and voltage regulation, and 3) a (low power) digital back end. when a carrier is sent from

the reader, the induced RF voltage at the tag antenna is rectified and regulated for a fixed

DC voltage. Usually all the functionality are performed using a single IC in commercial

RFID chip. However, for sensor coupled RFID, there are not any additional pins for the

7

sensor integration. Hence, in literature the proposed RFID based sensors are based on

microcontroller and discrete circuit components.

Figure 1.4: A commercial moisture sensor with interdigited capacitance for sensing.

Recently, few commercial RFID based sensor tags are active in market for different sensing

applications such as temperature, moisture, positioning etc. [32]. However, majority of the

sensor tags are non-customizable, which means the sensor based RFIDs will work only for

a specific application. An example moisture sensor is shown in Fig. 1.4 with interdigited

capacitance for sensing at the center. A RFID platform (SLA) for different type of sensors is

commercially available. However, the SLA architecture consumes lot of power, which limits

the read range of the sensor in passive mode of operation.

1.2 RFID System and Components

Figure 1.5: A typical RFID system with reader and tag.

A RFID system primarily consists of a reader and multiple tags, where the reader obtains

information from the tags wirelessly. A typical RFID system with a reader and a tag is

shown in Fig. 1.5.

In RFID communication, the reader provides the RF carrier signal,

8

which is backscatter modulated by a tag according to a specific bit stream. The bit stream

is called digital address or ID of the RFID tag and the address is saved in a non-volatile

memory inside RFID tag. The digital address is either programmed as factory set condition

or reprogrammable during later operation. This address is non-programmable in a read-

only tag and rewritable in a read-write tag. Like any other wireless communication, the

backscatter modulation technique requires an antenna for transmission and reception of RF

energy. Hence, A RFID tag contains two building block components: an antenna and an IC

for providing the digital address as shown in Fig. 1.7.

Figure 1.6: RFID tag antenna and IC at the center.

Figure 1.7: Block diagram of RFID internal circuit.

A generic basic block diagram of the internal circuitry of a RFID chip is given in Fig. 1.7.

As mentioned earlier, the RFID tag contains an antenna and an IC. Within the IC, there are

primarily two major components: 1) energy harvester, and b) digital microcontroller. Once

a RFID tag is illuminated with RF power, the energy harvester converts the RF power into

DC, which drives the digital microcontroller. The efficiency of the energy harvester is very

9

important as it will primarily dictate the read range of the tag from the reader. When the

DC power is available, the detector circuitry is activated within the microcontroller. The

detector circuitry detects the specific code sequence, sent by the RFID reader. Based on the

code, the microcontroller decides what task to perform. A programmable tag can perform

multiple actions based on the reader query: 1) write specific digital address or ID at the tag,

2) send the ID data to the reader, 3) perform a specific action such as writing at specific

registers, etc. The digital controller contains non-volatile EEPROM register, where the ID is

stored. Some advanced microcontroller includes a password protection scheme to access the

registers and hence to change the ID address. The microcontroller can be more complicated

based on the intended applications. Among other important analog sub-circuit parts, two

noteworthy components are a) voltage regulator, and b) random number generator. The

voltage regulator maintains the rectified DC voltage at a defined level to protect from over

voltage. When multiple tags are present in the environment, tag collision happens if more

than one tag reply simultaneously. To avoid this scenario, the microcontroller activates

the random number generator to decide when to transmit back to reader. The tag performs

backscatter modulation by switching on and off the high frequency switch, which is connected

parallel to the antenna as shown in Fig. 1.7. The mathematical formulation and physical

interpretation of backscatter modulation is provided in next section.

1.3 Backscatter Modulation

The conventional single frequency RFID operates on back-scattering principle. The early

development of back-scattering principle started around in early 1950’s after Yagi-Uda an-

tenna was reported. In back-scattering, impedance change in one antenna can be measured

10

by another antenna coupled to it. A few notable works on coupling of multiple antennas

were reported in [33–35].

In those works, as a general approach, a simultaneous integral

equation using Hallen’s method and variational expression is formulated. Then the expres-

sions are further simplified and solved for different purposes for example a) determining

broadside back-scattering cross section for a center loaded cylindrical dipole with arbitrary

load impedance [33], b) analyzing impedance parameters of coupled antennas [34], and c)

a generalized scattering formulation applicable for back-scattering and bi-static scattering

with active or passive loads extendable to N loads [35].

In back-scattering, a single antenna is used as transmitter and receiver for measuring the

impedance change of the target antenna. And in bi-static scattering two separate antennas

are used as transmitter and receiver for impedance change of target antenna. Although

back-scattering RFID is predominantly used, bi-static RFID is not very uncommon [36].

Bi-static RFID is usually helpful for localizing a RFID tag in 3D space. If there is only a

reader antenna and a tag antenna, then the formulation is a two port network. However, the

network changes to N port based on number of reader antennas or tags present. To keep the

formulation simple, single reader and tag antenna is considered with the medium in between

them reciprocal. Also, only passive load is considered for the RFID tag loading.

The detailed mathematical formulation of backscatter modulation working principle is

provided in Appendix A with two antenna system as shown in Fig. 1.8. The objective of the

analysis was to find the analytical expression of modulation depth due to two different load

impedance at the RFID tag antenna during modulation. Maximum modulation depth is

always desired as that would ensure the detectable signal level over the inherent noise level.

Now, with the help of modulation depth expression (1.1), it would be easy to find the realistic

load impedance combination during modulation that would provide maximum modulation

11

Figure 1.8: Two cylindrical dipole antennas with different dimensions.

depth. In the analysis, the tag antenna was considered a dipole antenna with infinitesimally

small width and the reader antenna was considered infinitesimally small dipole for simplified

expression.

(cid:20)

r (2) − V s
V s

r (1) = Z2
12

(cid:21)

1

Z22 + ZL2

−

1

Z22 + ZL1

I1(0)

(1.1)

1.3.1 Coupled antenna impedance

Computation and simulation are performed to verify the antenna coupling effect with

different load impedance at one of the coupling antennas. Using small reader antenna ap-

proximation described above, computation is performed for different combination of load

impedance for maximum modulation depth. For computation, six different combinations

are considered. The combination pairs for load impedance considered are [0, 73], [0, 30],

[30, 73], [30, 200], [73, 200], and [0, 200]. Each of the cases are termed as case I, II, III, IV,

V, and VI and the modulation depth is computed for those six cases according to Equation

(A.23). The simulation was performed in Ansys HFSS. In the simulation, the percentage

12

change in impedance at the reader antenna was measured for load impedance change at the

tag antenna for the previously mentioned six test cases. In the simulation, both the reader

and tag antennas were considered as quarter-wavelength antennas. For both computation

and simulation, l/a is considered as 560 to make sufficiently thin-wire approximation while

maintaining smooth meshing during simulation. The operating frequency considered is 900

MHz, and the separation distance between two antennas is 1.2λ. For computation, the small

reader antenna length is considered as dl/l = .025.

Figure 1.9: Normalized scattered impedance change at the reader antenna for different cases
according to Equation (A.23). Normalization was performed with respect to case I.

In Fig. 1.9, the normalized scattered impedance is plotted due to two different load

conditions.

In Fig. 1.10, the percentage change of reader antenna input impedance for

different cases of load impedance pair across the tag antenna is tabulated. From Fig.s 1.9

and 1.10, it is evident that the simulation result closely follows the pattern of computed

result. The best modulation depth is found for the load impedance pair of 0 Ω and 200

Ω. As the load impedance increases further from 200 Ω towards infinity, the modulation

13

depth becomes more stronger. In the test case analysis, 73 Ω was chosen because of the

real impedance of a standard dipole antenna at resonance is 73 Ω. However, as the RFID

tag is supposed to do energy harvesting, one of the antenna impedance choice should be

the conjugate of energy harvester input impedance, which is standard impedance as 50 Ω.

The another impedance choice can be either a short or open circuit. Circuit-wise performing

short operation at the antenna input is easier compared to open operation. Hence, the best

suited load impedance pair for the coupled antenna based modulation architecture are 0

Ω and 50 Ω.

In chapter 4, a standard RFID with sensor is demonstrated with two load

impedance pair as 0 Ω and 50 Ω for two different modulation states.

Figure 1.10: Normalized percentage change in impedance at the reader antenna for the same
six different cases as shown in Fig. 1.9. Normalization was performed with respect to case I.

1.4 Thesis Organization

The dissertation thesis is organized as followed:

• Chapter 2 presents the four primary challenges in conventional RFID. Solution to each

14

of the problems is presented from Chapter 3 to Chapter 6.

• Chapter 3 presents the harmonic RFID, capable of mitigating the clutter effects.

• Chapter 4 presents a low power generalized digital RFID platform, which can be used

to integrate analog sensors with digital RFID interface.

• Chapter 5 presents component level study on energy harvester for efficiency and re-

sponse time analysis.

• Chapter 6 presents a dual mode RFID, capable of reducing RF transmission time

significantly from RF reader.

• Chapter 7 finally concludes the thesis and proposes the possible future works.

15

Chapter 2

Challenges with Conventional RFIDs

In the previous chapter, the applications and simple working principle of conventional

RFID is demonstrated. Undoubtedly, RFID has become a multi-billion dollar industry today,

and applicability of conventional RFIDs in new areas is being researched. One of the primary

success reasons for passive RFID is the batteryless and small footprint architecture. This

specific feature enabled RFID desirable for applications such as: 1) underground or buried

object tagging and sensing, 2) RFID as a sensing platform (RaaSP) for a broad range of

sensors, 3) Low power operation for longer range or operating in a lossy medium, 4) Sensing

and monitoring in an environment where prolonged RF power transmission is not desired.

In spite of having many advantages, conventional RFID in its current form is not capable of

working for those applications due to fundamental working principle. The primary challenges

outlined are as followed:

• Clutter and multipath effect

• Sensor integration

• Low power operation and response time

• RF transmission from the reader

Each of the challenges are analyzed in detail and the best possible solutions are proposed

in the upcoming chapters. This chapter is primarily dedicated towards understanding the

16

physical limitations of conventional RFID. One chapter is dedicated for each of the four

challenges to propose a solution or understand the fundamental limits. All those solutions

are combined into a single new type of hybrid RFID, which is proposed in the last chapter,

mitigating all these four primary challenges.

2.1 Clutter and Multipath Effect

Clutter is described as an unwanted signal, which can obscure the desired signal and hence

diminish the system performance. Clutter source in conventional RFID can be several: 1)

self-jamming due to mismatch at the antenna port, 2) multi-reader jamming when multiple

RFID readers are transmitting simultaneously and one can jam the another, 3) reflection

from nearby objects [1, 2, 37]. The primary clutter objects are usually different in different

applications. For example, primary clutter object is biological body for a body area network

(BAN) sensor or sensor monitoring body vitals [38].

In underground object tagging, the

ground is the primary source of clutter [39, 40]. Similarly, in industrial environment, metal

objects are source of clutter [41]. The different clutter sources are shown pictorially for

multiple scenarios in Fig. 2.1.

(a)

(b)

(c)

Figure 2.1: Different sources of clutter (a) self-jamming [1], (b) multi-reader jamming [1],
(c) reflection from human body in a BAN sensor [2].

In [41], mathematical formulation is provided to show how the clutter phase noise can

17

shadow the low power tag return modulation signal. Also, experimentally it was demon-

strated that the read rate of a conventional RFID reduces in presence of a clutter source.

The read rate is directly correlated to signal to clutter ratio (SCR), and hence, read rate can

be monitored to quantify the clutter effect. The reflection clutter was measured in [2] using

the conventional RFID front end. In the reader front end, the baseband amplifier should be

of very low noise in order to measure the system phase noise [42]. If the tag returned signal

is of very low power and low frequency, it can get buried under the strong clutter phase

noise.

RFID has become widely accepted in enterprise supply chain management system as

it improves the efficiency of inventory tracking. In the supply chain, RFID tagged object

localization using RF signal is a desired feature. The RF localization is primarily done using

received signal strength (RSS), or phase based, or combination of both [43, 44]. However,

precise localization in conventional RFID system is a challenge due to a) multipath effect, b)

undesired interference, c) presence of multiple tagged objects [45]. In RSS based localization,

the signal attenuation level is monitored and the distance is estimated from Friis transmission

equation. Multipath effect is stronger in this method due to received signal power level

measurement. In phase based measurement, the phase of arrival of the backscattered signal

is measured. However, accurate phase based localization is possible when the bandwidth

is wider, which is difficult for conventional RFID system with only 30 MHz bandwidth.

Interference among multiple readers is possible and can contribute for distance uncertainty.

Also, when the RFID tags are densely populated, it becomes difficult to remove localization

ambiguity among the tags. The inherent problem lies at the frequency of operation and the

ambiguity reduces with increase in the operating frequency of operation.

18

Figure 2.2: Single frequency RFID tag and background reflector.

2.1.1 Clutter formulation

RFID systems operating at single frequency are susceptible to clutter noise or self-

jamming in multiple readers setup and an analysis on signal to interference ratio (SIR)

was shown in [1] in case of self-jamming when multiple readers would be transmitting to-

gether. Additionally, single frequency RFID tags can be disadvantageous while operating in

a cluttered environment. In Fig. 2.2, a single frequency RFID tag is shown in presence of a

background reflector. If the tag and the reflector are at a distance R1 and R2 respectively,

the received power at the reader due to tag and the reflector represented as Prt and Prr can

be shown as in (2.1) and (2.2) respectively.

Prt =

t G2

rλ4Γ

PtG2

(4πR1)4

Prr =

PtG2
t λ2σ
(4π)3R4
2

(2.1)

(2.2)

In (2.1), Pt and Gt represent the transmitted power and antenna gain of the reader, Gr

represents the tag antenna gain and Γ depends on the on-off switching of the tag. When Γ

is 1, the tag behaves like an ideal reflector and when Γ is 0, the tag is an ideal absorber.

From the presence and absence of reflected power, the ID sequence is decoded at the reader.

19

Figure 2.3: (a) Tag modulated signal, (b) Received tag signal at the reader in absence of
any reflector and (c) Received tag signal at the reader in presence of a reflector.

In (2.2), σ represents the radar cross section (RCS) of the reflector.

When a RFID tag receives a single frequency signal f0, it modulates the signal at f0 +∆f
and f0 − ∆f , which appears as the sideband of f0 as shown in Fig. 2.3(a). Once the tag

modulated signal reaches to the reader, the reader extracts the tag encoded ID at ∆f by

heterodyne mixing. In absence of any reflector, the dynamic range (DR) of the tag DRab

would be determined by (2.3), where Γf und represents the conversion efficiency for a practical

tag at fundamental frequency and RS represents the reader’s sensitivity level at f0 + ∆f or
f0 − ∆f . However, the scenario would be different in presence of any reflector within the

field of view of reader antenna. As the reader oscillator is not an ideal one, the oscillator
would have phase noise (P N ) at f0 + ∆f or f0 − ∆f , which would be backscattered by a

reflector. As a result, there would be reduction in DR as shown in Fig. 2.3(c) leading to poor

tag detection rate when P N dominates over RS. The clutter signal effects more strongly

when the received tag signal is close to RS level. The DR in presence of reflector DRpr as

given in (2.4) would depend on the distance, dimension and shape of the reflector. In this

scenario, even though the RS level is improved, it would not improve the tag read range in

presence of strong reflectors.

20

DRab =

PtG2

rλ4Γf und

t G2
(4πR1)4

− RS(f0 + ∆f )

DRpr =

− P N (f0 + ∆f )

PtG2

rλ4Γf und

t G2
(4πR1)4
PtG2
t λ2σ
(4π)3R4
2

where, P N (f0) =

2.1.2 Clutter measurement

(2.3)

(2.4)

It is important to quantify the clutter effect after mathematical formulation. A standard

single frequency RFID reader system is shown in Fig. 2.4. In general, any oscillator has

its standard phase noise around the carrier center frequency. In the presence of a strong

clutter object, the background clutter will appear along with the tag modulation signal. In

absence of clutter, the noise level will be primarily determined by the reader’s minimum

sensitivity. In presence of strong clutter signal, the noise level will be primarily determined

by the reflected clutter signal, which may dominate under strong reflections.

Based on the conventional RFID reader front-end, a RFID reader was constructed as

shown in Fig. 2.4 to measure the effect of clutter. In the reader, first the single frequency

signal is amplified and transmitted through a transmitter antenna. Now the back-scattered

tag signal or clutter will be received at the same antenna. The return signal is downconverted

using the carrier signal to obtain the baseband signal. To ensure the reader noise figure

√
minimization, an ultra-low noise amplifier with input voltage noise density of 2.4 nV/

Hz

was used to amplify the baseband signal before sampling and performing FFT analysis. In

the baseband signal, there are two parts: 1) tag modulation, and 2) return clutter signal. The

phase noise of the transmitted signal will dominate under strong cluttered environment. The

21

experimental setup to demonstrate the clutter effect is shown in Fig. 2.4. In the experiment,

a metallic cabinet was used for the clutter source. Three set of measurements were taken

at different distance from clutter source to show how clutter can effect the low frequency

tag-modulated signal.

At first, the RFID reader was kept at a distance of 25 cm from the clutter object and

the downconverted baseband signal is shown in Fig. 2.5(a). Due to specific phase noise

characteristics of the oscillator, when transmitted signal from the reader gets reflected from

the clutter object and after downconversion the phase noise was captured as shown in Fig.

2.5. When the distance was increased from 25 cm to 40 cm, the pattern of the phase noise

remains same with decreased amplitude, as expected, as shown in Fig. 2.5(b). The response

without any clutter source was measured using a broadband microwave absorber. In absence

of strong reflected signal, the background noise dominates over the phase noise as shown in

Fig. 2.5(c). The spikes in all the three graphs consist of even and odd harmonics of 60

Hz noise, which can be eliminated using better DC source. It is clearly visible that in the

presence of a clutter object, the phase noise level impacts significantly at lower frequency.

Hence, single frequency conventional RFID system can have disadvantage specifically in

biological systems under strong clutter for low frequency tag modulation (lower or equivalent

to 200 Hz).

2.2 Sensor Integration

RFID technology has become a good choice for sensing and monitoring at remote location,

where it is difficult to probe or wired future maintenance access such as battery change is

not possible. Many applications are in the field of structural health monitoring (SHM)

22

Figure 2.4: Front-end of a standard single frequency RFID reader.

(a)

(b)

(c)

Figure 2.5: Phase noise after FFT analysis for cases when the reader antenna is (a) 25 cm,
(b) 40 cm away from the clutter source, and (c) 40 cm away from an absorber.

[46, 47], intra-vehicular sensing network [48], utility such as gas pipeline sensing [40, 49],

remote monitoring of hazardous environment [3, 50], low-cost sensing for end to end supply

chain [4, 51–53], non-invasive human vitals sensing for health monitoring [54]. The primary

objective of those reported sensors is to integrate them with RFID system and operate them

wirelessly and without battery. The earlier reported sensors can work nicely wirelessly and

without battery. However, the sensing elements are fundamentally analog in nature. The

sensors convert physical parameters to electrically realizable signal such as voltage, current

or impedance. As the output voltage, current, and impedance from the sensors are analog

in nature, interfacing those sensors with conventional RFID system becomes requires use of

digital interface, which is a challenge in nature due to requirement of low power ADCs.

23

(a)

(b)

Figure 2.6: Wireless and batteryless sensing system: (a) a pH sensor with sensor output
embedded in phase information [3]; (b) a food volatile sensor with sensor output embedded
in the resonance frequency shift [4].

In earlier reported wireless and batteryless sensing systems, the analog sensors are usually

integrated to directly impact the phase or coupling power level of the interrogating signal.

For example, in [3, 50] the sensor output is converted into phase and integrated with the

interrogating signal. The sensor signal is extracted at the interrogator from the relative phase

change. However, the system needs initial calibration for the phase change due to distance

traversed by interrogating signal, which propagates through the medium between sensor tag

and interrogator. The scenario becomes specially challenging when the sensor tag is not static

and the relative distance of the tag from the interrogator is continuously changing. Among

other analog sensing methods, the resonance frequency shift method is also very popular

[4, 51]. In resonance frequency method, coupling of two coils are used, where one coil is with

the sensing element and the other with the interrogator. The sensing element is usually

capacitively coupled with the inductive sensor coil, forming a LC tank. The capacitive

loading of the sensor is detected from the shift in resonance frequency. Coupled coil based

sensing has become very popular as they require minimum components. However, there are

two major challenges with this kind of sensing: 1) a comparatively wide frequency sweep is

required to ascertain the resonance frequency peak, requiring a wider system bandwidth; and

24

2) the coupling co-efficient starts reducing drastically when the relative distance between the

sensor and interrogator increases, hence the minimum detection limit is a function of relative

distance. Also, the relative orientation of the coupled coils greatly affects the coupling co-

efficient. Examples for both type of wireless and batteryless sensor system is shown in Fig.

2.6.

(a)

(b)

Figure 2.7: Example of sensor integrated with conventional RFID IC: (a) the sensing element
is integrated on top of the antenna [5]; (b) the sensor tag antenna and in the inset sensing
element integrated in between antenna and RFID IC [6].

In order to integrate the wireless and batteryless sensors with RFID, another alternative

approach became popular, where the sensing element is directly embedded on the antenna

of conventional RFID tag or in between the antenna and RFID IC [5,6]. In [5], two example

sensors are demonstrated, where the sensing element is placed on the antenna and the sensing

element changes impedance based on external stimulus. As the antenna impedance changes,

the power received at the interrogator changes and from the change, the sensor output is

extracted. Similarly, in [6], the sensor is placed in between the tag and antenna. Based

on the impedance change, the reflected power level from the tag is changed, which would

provide the sensor state. Those kind of RFID integrated sensors are simple in architecture,

but the tag reflected power level will also be effected by the environment. To mitigate that

effect, a redundant tag can be used to provide the reference sensor level. However, other

25

complexities such as coupling between multiple tags, area requirement, special packaging

will arise. Examples for both type of conventional RFID IC integrated sensor system are

shown in Fig. 2.7.

A more practical approach is demonstrated in [55], where the analog output of the sensor

is converted into digital bits and integrated with the RFID tag ID bits before sending the

ID sequence back to the reader.

In this method, the sensor output is not susceptible to

environment factors. However, an extra analog to digital converter (ADC) is required to

convert analog signal into digital bits, which usually consumes extra energy unless selectively

chosen. However, this kind of sensing introduces quantization error, which reduces the tag

sensor sensitivity.

2.3 Low Power Operation and response Time

Passive RFIDs predominately operate on harvested energy and RF to DC conversion is

used due to ubiquitous operation requirement independent of light, temperature, or wind.

Hence, the power consumption at the tag is limited by the energy harvesting. Naturally,

the range of the tag is limited by the power consumption, which increases with addition of

functionality such as digital sensing, security, and so on. Hence, a popular idea is to charge

a comparatively large storage capacitor for longer time and use the stored energy within a

short interval of time. This technique helps in reduction of average power consumption but

at the cost of longer start-up time due to charging of larger capacitance. Hence, a tradeoff

should be maintained between the following two factors while designing an energy harvesting

unit: (a) maximum discharge energy, and (b) start-up time.

The power consumption at the tag for digital operation are primarily (a) static, and (b)

26

Figure 2.8: Schematic block diagram of the RFID tag during energy harvest.

dynamic consumption. The static power consumption remains constant at a fixed voltage

but the dynamic power consumption strongly depends upon the switching frequency.

In

digital circuits, the dynamic power consumption usually contains the larger share of total

consumption and is denoted as in (2.5).

Pdynamic ∝ αf V 2

DD

(2.5)

From (2.5), it is evident that the power consumption increases in proportionate to the

frequency of operation f and in square law to the voltage of operation VDD. In (2.5), α is

termed as the switching probability between the ’0’ and ’1’ state. For low power operation,

the operating frequency and VDD level should be kept as small as possible. VDD level is

determined by the process technology using which the monolithic IC is fabricated. With

introduction of smaller process nodes such as 10nm, 7nm, or 5nm, the power consumption

can be substantially reduced. On the other hand, the frequency of operation is primarily a

design parameter and can be reduced by efficient design architecture.

Apart from power consumption, the energy harvesting should be efficient enough to

supply the necessary power for consumption. The harvested energy at the tag Pharvest is

expressed as in (2.5), where the reader transmitted power is Pt, antenna gain is Gt, distance

between tag and antenna is d, tag antenna gain is Gr, reflection co-efficient between tag IC

and tag antenna is Γ, energy harvester efficiency is η assuming d > (2D2/λ), where D is the

27

Figure 2.9: Analysis on energy harvest at the tag vs. distance with varying consumption.

largest aperture of the reader antenna, and Gt >> Gr. The scenario is shown in Fig. 2.8

and expressed in (2.6).

Pharvest =

PtGtGrλ2

(4πd)2

(1 − Γ2)η

(2.6)

The maximum range of the tag dmax is defined as the distance between the tag and

reader, for which the following condition is fulfilled as expressed in (2.7), where Pdynamic is

the dynamic power consumption at the tag and Pstatic is the static power consumption.

Pharvest ≥ Pdynamic + Pstatic

(2.7)

A simple analysis with Pt = 30 dBm, Gt = 6 dBi, Gr = 0 dBi, Γ = 0, and η = 30%, is

shown in Fig. 2.9 about the read range of the tag, considering different power consumption at

the tag. The maximum read range decreases from 9 m to 6 m, when the power consumption

at the tag increases from 10 uW to 20 uW. Hence, the average power consumption at the

tag should be as low as possible for maximum read range. A detailed analysis on the energy

harvesting will be performed to show the efficiency and response time dependency on the

circuit components.

28

2.4 Prolonged RF transmission from reader

Continuous RF radiation from the reader is another challenge in conventional RFID.

To keep the RFID tag functional, the reader should be always illuminating and continuous

RF radiation may not be always desired. Recently, there has been multiple studies on the

adverse effects of prolonged RF radiation on human body [56–58], with few studies reporting

significant chromosomal DNA damage. Most of the studies included cellular, bluetooth,

wearable device frequency bands, starting from 700 MHz upto 5 GHz. Among different

applications of UHF RFID, few desired ones are sensor integrated RFID within human

body, or integrated sensors as wearable devices or in vicinity of human body. Continuous

RF radiation can be detrimental in those cases if the radiation time is long and of high

power. As the conventional RFID tags can backscatter only when RF carrier is present,

continuous RF power should always reach at the tag for desired modulation. Hence, it is

fundamentally impossible to extract sensor information from conventional sensor integrated

RFID without radiating RF power from the reader. The scenario is described pictorially

in Fig. 2.10, where the desired scenario with reduced RF radiation transmit time from the

reader is shown in Fig. 2.10(b).

As a preventive measure, alternate ways of reduced RF transmit time was explored

in this work. The proposed dual mode RFID will be capable of backscattering as well

as synthesizing its own carrier to communicate with the reader. The dual mode RFID

has significant advantage in reduction of effective transmit time while monitoring the tag

integrated seamlessly.

29

(a)

(b)

Figure 2.10: (a) RF radiation on human body, and (b) Desired effective RF radiation time
reduction from the reader.

2.5 Summary

Different challenges in conventional RFID are outlined in this chapter. The solutions for

the four major problems as outlined at the beginning of the chapter, are proposed individually

in the subsequent chapters from Chapter 3 to Chapter 6. In Chapter 7, the possibility of a

single integrated hybrid RFID is proposed, capable of mitigating all the four major problems.

30

Chapter 3

Harmonic RFID

In chapter 2, the challenge due to clutter issue in conventional RFID was discussed

in detail. To overcome clutter issue, different digital encoding schemes were explored in

the literature. In [59], an optimum maximum-likelihood sequence detector (MLSD) scheme

was implemented which showed a 3 dB improved system performance compared to the

conventional ML sequence detector. However, there is a fundamental limit to the system

performance of passive tags under the same frequency operation [60, 61]. The SCR can be

increased dramatically by implementing the RFID communication channel in two different

frequency bands, one for uplink and the other for downlink. In this dual frequency operation,

the weak signal from the RFID tag can easily be filtered and amplified at the interrogator

to obtain a clutter-free signal. Harmonic RFID tags operate in two frequency bands by

transmitting ID data at the harmonic of the received signal frequency and thus it has emerged

as a potential solution for long-range RFID tags in a cluttered environment [1].

3.1 Background

Clutter and localization issues can be substantially mitigated by employing harmonic

RFID solution. In harmonic RFID, the tag and reader operates at two different frequencies:

a) reader to tag downlink at fundamental frequency, and b) tag to reader uplink at harmonic

frequency. In harmonic RFID, the tag generates harmonic of the reader transmitted signal

31

and uses the harmonic as carrier for modulation. Harmonic RFID has advantage over clutter

as the clutter appears at fundamental frequency, whereas the tag return signal is at harmonic

frequency. Hence, the clutter cannot obscure the desired signal unlike the conventional RFID

system.

Similarly, the localization is also improved using harmonic RFID. In [7], it was exper-

imentally shown that as the phase information is contained within the second harmonic,

interferences and phase errors caused by direct reflections of the interrogating signal are

greatly reduced. Also, the phase information is better preserved in harmonic operation and

immune to background interference, as demonstrated in [3]. Simultaneously, as the readers

transmit only fundamental frequency, reader to reader interference will be greatly reduced.

Additionally, as the harmonic frequency reduces the operating wavelength, the localization

accuracy is significantly enhanced [62].

(a)

(b)

Figure 3.1: (a) Harmonic phase based localization in presence of a scatterer [7] (b) Harmonic
phase based sensing with better immunity to background clutter [3].

CMOS based frequency multipliers are quite popular as they can readily be integrated

with other digital processing CMOS circuits, albeit their high power consumption [63]. Other

choices are the use of zero bias Schottky diodes [39] or varactor diodes, largely used in the

design of nonlinear transmission line (NLTL) [64]. However, Schottky diode based harmonic

32

generators are not suitable for ID modulation because of self-biasing and large current re-

quirements during the on stage. On the other hand, the potential of NLTL as a harmonic

RFID was shown in [65] and a developed tag circuit was reported in [1]. However, the har-

monic RFID tag in [1] used a microcontroller based architecture, which required relatively
high power during operation (∼ 100 W) and was not a cost-effective solution. Due to high

power consumption, the range of a microcontroller based RFID tag would be limited. In

some architectures, the power is harvested over a long period of time and the tag sends the

ID within a very short time period after activation [66].

In this case, a complex system

architecture and signal latency would be an issue. To overcome these problems, an ultra-low

power harmonic RFID tag is proposed in this work that operates in a continuous mode.

The wireless communication link of RFID tags can be categorized broadly into two groups:

1) mutual coupling of the magnetic field through coupled coils [67] and, 2) using antennas

for far-field communications [39]. For the first type, the tag is mutually coupled to an

interrogator within one wavelength distance usually in LF and HF bands. Whereas, far field

antenna based tags mostly operate in UHF band as high as 2.5 GHz [39]. In the spectrum

of license-free RF bands, there are very few bands available, which meet the criteria of

fundamental and harmonic bands. Among the license-free bands, available bands are ISM

6.78 MHz and ISM 13.56 MHz duplex, ISM 13.56 MHz and ISM 27.12 MHz duplex or ISM

434 MHz and SRD 868 MHz duplex [68]. At low frequency, RF communication is performed

using magnetically coupled coils, which are inefficient in power transfer [69]. Hence, relatively

high-frequency band 434 MHz and 868 MHz duplex were chosen for harmonic tag design in

this work. The proposed harmonic RFID system consists of an interrogator and an harmonic

RFID tag. The interrogator transmits a signal at fundamental frequency (f0). When the

signal is received by the harmonic tag, it generates and transmits back the second harmonic

33

Figure 3.2: Fundamental and harmonic antenna setup in between the tag and reader.

(2f0) of the fundamental frequency at a specific sequence, which is decided by the ID or

IP address of the tag. Upon receiving the transmitted signal from the tag, the interrogator

demodulates the sequence at 2f0 and the specific ID of the tag is obtained.

This chapter is organized as followed:

• Design of NLTL, a low power harmonic generator is discussed.

• A first generation harmonic RFID tag with two antennas is shown.

• A second generation single antenna based miniaturized harmonic RFID is shown.

• Lastly, efficacy of harmonic RFID in presence of clutter is illustrated.

3.2 Nonlinear Transmission Line: NLTL

In this section, it will be shown how the harmonic generation efficiency can be enhanced

by using specific operating condition and multiple number of varactor diode as harmonic gen-

erator elements. The generated harmonic signals from multiple sections are added in-phase

to generate very low conversion loss harmonic output. One such structure is called nonlinear

transmission line (NLTL). The NLTL can be a continuous co-planar waveguide (CPW) line

34

on a semiconductor substrate with discrete metal semiconductor junction diodes [8, 70], or

distributed configuration using metal-semiconductor junction on GaAs substrate [71], or dis-

crete inductor and varactor diode structure fabricated monolithically [9], or discrete inductor

and varactor diode structure fabricated on a PCB using discrete packaged components [72].

As the maximum bandwidth is dictated by the NLTL cut-off frequency, a monolithic fabri-

cation would provide the most operational bandwidth as high as 20-30 GHz [8]. Examples

of differently fabricated NLTLs are shown in Fig. 3.3, for (a) discrete diodes on CPW line,

and (b) discrete diodes and inductors.

(a)

(b)

Figure 3.3: (a) Example of NLTL: (a) Discrete diode with CPW line [8] (b) Discrete inductors
and diodes shown at the top [9].

3.2.1 Design principle

The design principle of NLTL is described in detail in [2, 8, 72] with periodic structure of

multiple sections of NLTL. There are different methods of designing NLTL circuits, among

which the prevalent one is to approximate an equivalent LC network similar to a transmission

line but with nonlinear capacitance due to diode instead of constant capacitance.

Any diode operating in nonlinear region acts as a harmonic generator. The voltage

dependent capacitance of a diode is expressed as followed in Equation (3.1). In (3.1), Cj0,

35

Vj and m are diode characteristics parameters, where Vj is the forward junction voltage and

m depends on the doping profile of the semiconductor. The voltage V denotes the applied

voltage across the diode, which consists of a DC term Vdc and an AC term Vac. The DC

term denotes the fixed voltage, at which the diode is biased constantly. The time varying

charge q(t) across the diode can be represented by time varying AC voltage and the DC bias

as shown in Equation (3.2).

C =

Cj0

(cid:0)1 + V
(cid:20)

Vj

q(t) = Cj0

1 +

(cid:1)m

where, V = Vdc + Vac

(cid:21)−m(cid:0)Vdc + Vac(t)(cid:1)

Vdc + Vac(t)

Vj

(3.1)

(3.2)

If Equation (3.2) is expanded, q(t) is expressed as a combination of different higher order

AC voltage terms. In (3.3), a0, a1, and a2 are the co-efficients of corresponding DC, first

order AC, and second order AC (or harmonic) terms of the time varying voltage. The other

higher order modes are neglected, as primary focus is generation of second order harmonic.

q(t) = a0 + a1Vac(t) + a2V 2

ac(t) + a3V 3

ac(t) + ...

(3.3)

If the diode is terminated across a load R, the output voltage can be obtained in terms

of all harmonics or higher order modes of applied time varying input voltage. If the input

voltage is a single sinusoid term, all the harmonics will appear at the output, as shown below.

Vout(t) =i(t) ∗ R

ac(t) ∗(cid:2)a1 + 2a2Vac(t) + 3a3V 2

ac(t) + ...(cid:3)

=V (cid:48)

(3.4)

36

Figure 3.4: Lumped element representation of the NLTL with a single cell NLTL at the
bottom.

The amplitude of the output harmonic power will primarily depend on the diode pa-

rameters such as m, and Vj. Additionally, the constant DC bias Vdc will also effect the

harmonic generation. Physically, any diode or nonlinear device generates strong harmonics

at its maximum non-linear region. The maximum non-linearity region can be obtained of a

varactor diode by measuring capacitance with change in bias voltage. Curvature co-efficient

γ, which is a pre-factor of non-linearity can be extracted from the C-V curve of a varactor

diode.

Once, conditions for maximum harmonic generation from a single varactor diode is ob-

tained, the efficiency can be further enhanced by combining multiple varactor diodes, and

the design topology is termed as nonlinear transmission line (NLTL). Due to periodicity, one

popular way to model the lumped element based NLTL sections is use of ABCD matrix.

The lumped element based NLTL is shown in Fig. 3.4 with a single cell. The ABCD matrix

of a single cell NLTL is first formulated. Based on the single cell model, the multi-cell NLTL

can be extended with cascaded multiple single cell. For a single cell NLTL as shown in Fig.

3.4, the ABCD matrix is expressed as given in Equation (3.5). After simplification, the form

of the ABCD matrix becomes as given in Equation (3.6).

37

0

1

=

single

C D

A B
lossless
1 jωL/2
 .
 1
lossless
A B
1 − ω2LC
A B
lossy


C D

single

jωC

=

=

k

2

C D

single

G + jωC
k = 1 + RG − ω2LC

jωC 1

 .

0

1

0

1 jωL/2



(cid:1)k
(cid:19)

k

)

2

L(1 − ω2LC
1 − ω2LC

4

jω(cid:0)R + jωL
(cid:18) LG

2

(3.5)

(3.6)

(3.7)

where,

+ jω

+ RC

2

2

Now, as the NLTL system should be symmetric at a fixed C, the ABCD matrix given

in (3.6) should be symmetric in nature. However, the expression in (3.6) is in simplified

form ignoring the loss in series inductors and shunt capacitors. By incorporating the series

resistance R and parallel conductance G, the expression in (3.6) can be modified as in (3.7).

With the simplified symmetric condition of ABCD matrix being imposed, the characteristics

impedance of the NLTL as ZN LT L can be expressed in terms of design parameters as in

(3.8). The solution for L can be readily derived by solving (3.8) with the known capacitance

of the diode.

L
C

− ω2L2
4

= Z2

N LT L

(3.8)

For a long network of NLTL with multi-stage elements of m, the total composed ABCD

matrix of the m-stage network becomes as given in Equation (3.9). Now, with any order of

NLTL, the final composed network will be symmetric in nature too, as each of the elemental

matrix is symmetric in nature. The loss effect increases exponentially With more number

38

of stages from the derived ABCD matrix for lossy case. Hence, an optimum stage number

exist, beyond which the harmonic generation does not improve with addition of more stage

number. A theoretical analysis was performed in [73] to describe the harmonic generation

by a NLTL.

A B

C D



A B

C D

=

m

single

mstage

(3.9)

The Telegrapher’s equations for NLTL was solved in [73] to express the output at har-

monic frequency in terms of nonlinearity factor and fundamental frequency input. As men-

tioned earlier, number of NLTL stages is important as it decides the maximum harmonic

output power availability. Primarily two major events happen in NLTL: (a) the output

harmonic power is added in-phase as the input propagates through the NLTL, and (b) at-

tenuation of both harmonic and input power as the physical NLTL is lossy in nature. The

in-phase output power addition is used extensively in optical regime as parametric mix-

ing [74, 75]. In parametric mixing, a high power pump signal and a low power idler signal is

provided as input and the difference term is obtained as mixing output. When the idler is

exactly at twice frequency of pump frequency, a harmonic generator is obtained. However, in

microwave harmonic generators, the idler is not required to be provided separately as there is

always small amount of harmonic will be present with the pump signal. The optimum num-

ber of stages or the NLTL length would be based on this two contrasting factors, until the

harmonic generation is more than attenuation. As a good practice, the NLTL attenuation

should be kept as low as possible for maximum harmonic output power.

39

3.2.2 Simulation and measurement

Prior to the design, all the required circuit simulations were performed using ADS from

Keysight. For the simulation, varactor diode model recommended by Skyworks was used.

As lumped components were used for the design, lumped passive components were used in

simulation. After modeling, the simulation result was later verified with the measurement

result. The measurements were performed using a Vector Network Analyzer (VNA) for

reflection and transmission co-efficients of the NLTL circuit. The results are shown below

for different number of stages.

(a)

(b)

Figure 3.5: (a) S11 and (b) S21 for different number of NLTL stages.

40

(a)

(b)

Figure 3.6: (a) S11 and (b) S21 for three number of NLTL stages at different DC bias.

(a)

(b)

Figure 3.7: (a) S11 and (b) S21 for four number of NLTL stages at different DC bias.

The measurement results show the transmission line behavior of the NLTL. As the number

of stages are increased, the roll-off of the S21 increases as shown Fig. 3.7(b) and 3.6(b). The

S11 behavior of different NLTL stages shows that most of the power is passed through the

NLTL upto 1 GHz. When the bias voltage is changed, the effective capacitance of the

41

varactor diode changes, and it will change the transmission characteristics of the NLTL.

3.3 Harmonic RFID: Gen I

The first generation harmonic RFID developed at MSU is presented in this section. The

harmonic RFID system in general includes RFID reader and tag. The primary features of

this harmonic RFID system are:

• Gen I harmonic RFID tag has 8-bit ID integrated with it.

• It has two antennas, one for transmission and other for reception at the tag and reader.

• The tag has a RF power splitter to provide RF energy for harmonic generation and

energy harvesting.

Different parts of the Gen I harmonic RFID system are discussed in detail subsequently.

The three parts discussed here are: (a) antennas used for both reader and tag, (b) harmonic

tag circuit, and (c) harmonic reader circuit.

3.3.1 Antennas

The antenna provides the wireless link for power and data transfer in between the har-

monic RFID tag and interrogator. Three different antennas designs are needed for the

complete system design as illustrated in Fig. 3.2. The antennas associated with the tag

should be small for a miniaturized tag design. On the other hand, the reader antenna should

have a high realized gain for efficient long range tag interrogation. Additionally, the size

constraint is not stringent on the reader antenna as compared to the tag. Hence, two dif-

ferent type of antennas were designed at the fundamental frequency (f0) for the reader and

42

Figure 3.8: Antenna structures for (a) Meandered dipole (434 MHz) for tag side, (b) Me-
andered dipole (868 MHz) for tag and interrogator sides and (c) Yagi-Uda (434 MHz) for
interrogator side.

Table 3.1: The dimension of different antenna designs

Yagi-Uda (f0)

Reflector

Director

Parameter Dimension(mm) Parameter Dimension(mm)

a

282

b

134

Parameter Dimension(mm) Parameter Dimension(mm)

Meandered Dipole (f0)

a
b
c
d

6.25
55
31
6.75

e
f
g
h

48
17
3
5

Parameter Dimension(mm) Parameter Dimension(mm)

Meandered Dipole (2f0)

a
b

23
10

c

8

tag. The transmitting antenna of the reader is a high gain two elements Yagi-Uda antenna

and the receiving tag antenna is a small meandered dipole antenna operating at (f0). Apart

from f0, the harmonic system also operates at 2f0 as shown in Fig. 3.2. Hence, a common

meandered dipole at harmonic (2f0) frequency was designed and implemented for the tag

transmitter and the interrogator receiving end.

43

Figure 3.9: Effect of T-match on input impedance shown on Smith chart from 100 MHz to
600 MHz.

Figure 3.10: Radiation pattern of the Yagi-Uda with and without the back reflector.

44

(a)

(b)

(c)

Figure 3.11: Simulation and measurement results for (a) 434 MHz meandered dipole, (b)
434 MHz Yagi-Uda and, (c) 868 MHz meandered dipole antenna.

3.3.1.1 Meandered dipole antenna

Different meandered dipole or loop antenna designs have been proposed to achieve com-

pact size in the UHF frequency band [76, 77]. Meandered dipole antennas have become

popular for RFID applications due to their compact size and easier to impedance match to

high impedance RFID ICs [76]. Here, a λg/2 meandered dipole antenna was designed and

fabricated on RO4350B board from Rogers Corporation. It has a dielectric constant (
r) of

3.6, where λg corresponds to the effective wavelength at (f0). By meandering, the effective

dipole length of the antenna is maintained while reducing the board size. However, there

is a tradeoff between antenna effective size and gain. In [78], genetic algorithm was used

to optimize in between gain and area occupied by the meandered dipole. In this work, an

optimized design topology from [78, 79] was adopted for the meandered dipole with reason-

able gain while maintaining small size. The meandered dipole antenna as shown in Fig.

3.8(a) has a gain of 0.98 dBi with an effective board area of 48 cm2 at (f0) after simulation.

Additionally, the input radiation resistance of a meandered dipole reduces from the standard

dipole as described in [80]. Conventionally, a T-match is used to match the low impedance

antenna to a high impedance system [76]. The antenna design parameters are shown in

45

Table 1.1. The effect of T-match on input impedance matching of the meandered dipole is

shown in Fig. 3.9 with respect to a 50 Ω measurement system.

Another meandered dipole at (2f0) was designed following the same design principles.

This antenna showed a gain of 2.1 dBi with an effective board area of 18.7 cm2 at (2f0) from

the simulation. The design parameters of the antennas are provided in Table 3.1 as shown

in Fig. 3.8(b). This way two meandered dipole antennas were designed at fundamental (f0)

434 MHz frequency and harmonic (2f0) 868 MHz frequency.

3.3.1.2 Yagi-Uda antenna

A two element Yagi-Uda antenna was designed for the interrogator to achieve high gain

during transmission. As the fundamental operating frequency is at 434 MHz, a planar multi-

element Yagi-Uda implementation would occupy large space on the board. Hence, the design

was restricted only to two elements having one driving element and one reflector element.

The design parameters of the antenna elements is shown in Fig. 3.8(c) are provided in

Table 3.1. Simulated antenna gain patterns in the E-plane are shown in Fig. 3.10 with and

without the back reflector. From the simulation result, it can be verified that the forward

gain improves by 4 dB and an improved front to back ratio (FBR) of 10 dB compared to a

simple dipole design.

The antennas were fabricated on a 1.52 mm thick RO4350B board using a milling ma-

chine. The reflection coefficients (S11) were measured using a network analyzer. Fig. 3.11

shows the simulation and measured results of the antennas. The conductor line width is 1.5

mm for all the antennas unless it is mentioned otherwise.

46

Figure 3.12: Harmonic tag architecture with different components.

3.3.2 Harmonic tag design

The architecture of the harmonic tag circuit along with antennas is shown in Fig. 3.12. It

consists of the following components: 1) Power splitter (PS), 2) Harmonic doubler (HD), 3)

Energy harvesting unit (EHU) and 4) Digital modulation unit (DMU). The PS, at the input,

divides the received RF signal into two parts: (a) for energy harvesting in the EHU and (b)

for harmonic generation from the fundamental carrier signal (f0) in HD. One advantage of the

designed harmonic tag is that it can operate in continuous mode, which means that it harvests

energy and transmits ID information simultaneously. This is in contrast to a conventional

tag, which first harvests and then transmits the ID. Here, once the EHU harvests the required

energy, it activates the DMU, which triggers the HD ‘off’ and ‘on’ sequentially. A simple

amplitude shift keying (ASK) based modulation was adopted for triggering. Based on the

‘off’ and ‘on’ sequence, the HD subsequently generates the harmonic signal. The interrogator

identifies the ID by reading only the harmonic signal.

3.3.2.1 HD using NLTL

Use of NLTL as an efficient HD has been reported previously [64]. The equivalent discrete

component realization of a NLTL is shown in Fig. 3.13.

It consists of a periodic series

inductance and shunt varactor diode combination. The capacitance of a varactor diode

shows non-linear behavior as a function of DC bias. The harmonics of the input f0 signal

47

Figure 3.13: A three stage nonlinear transmission line (NLTL) with matched termination at
434 MHz.

are generated at the varactor due to its inherent non-linearity. The varactor should be biased

at the strongest nonlinear region of its C-V characteristics for maximum harmonic power

generation as described in [65]. Once this bias condition was determined, the inductance

value was calculated for impedance matching to a 50 Ω load at 434 MHz (f0). A three-stage

NLTL was designed using L=10 nH and Skyworks SMV1405 varactor diodes. The varactor

diode shows maximum non-linearity close to its threshold voltage, where the depletion width

is minimum. The DC sweep of the three-stage NLTL shows a maximum harmonic generation

at 0.6 V forward bias as shown in Fig. 3.14. However, increasing the bias voltage further did

not help as the varactor started operating in the conduction mode. As shown in Fig. 3.14,

the NLTL produces a maximum harmonic power of -25 dBm at a 0.6 V bias. In comparison,

at 0 V it produces -51 dBm at an input power of -10 dBm. Hence, an analog DC voltage 0.6

V was used for digital state ‘1’ and 0 V for the digital state ‘0’ to achieve an RF isolation of

26 dB between the ‘on’ and ‘off’ states. The modulation signal from DMU was fed through

an high impedance inductor, 0.1 mH, to isolate the RF path from the low-frequency digital

signal as shown in Fig. 3.13.

48

Figure 3.14: Measured second harmonic output power of a three stage NLTL as function of
DC bias for input power of -10 dBm at 434 MHz.

During wireless operation, one end of the NLTL was terminated with an antenna operat-

ing at f0 for RF reception and another end with an antenna operating at 2f0 for harmonic

transmission. However, the 2f0 operating antenna does not show 50 Ω input impedance at

f0 as shown in Fig. 3.8(b). Hence, the NLTL was terminated with the 2f0 antenna through

a matched termination circuit as shown in Fig. 3.13. The other choice was to choose the

inductors of the NLTL for impedance matching with the 2f0 antenna. However, this would

complicate the performance evaluation of the NLTL, which would be easier when matched to

the standard instrument impedance of 50 Ω. The matched termination uses a high frequency

50 Ω resistor for impedance matching at f0. Additionally, the matched circuit includes a

series resonator at 2f0 to prevent the drainage of harmonic power into the resistor. The re-

flection coefficient (S11) of the matched termination is shown in Fig. 3.15 with and without

the 2f0 operating antenna. A good matching was observed at f0 irrespective of the presence

of the (2f0) operating antenna. Whereas at 2f0, all the input power was reflected back in

49

Figure 3.15: Measured reflection coefficient of the matched termination with and without
2f0 antenna.

the absence of antenna and radiated in the presence of antenna. Hence, the harmonic power

was transmitted instead of being absorbed by the resistor. The design parameters for the

matched termination are provided as C1=6.8 pF, L1=5 nH, C2=8.2 pF, L2=9 nH and R=50

Ω.

3.3.2.2 Energy Harvesting Unit (EHU)

The EHU provides the required DC power to drive the DMU and HD. Among the many

energy harvesting methods that can be used, here an RF energy harvesting circuit (RF

rectifier) was adopted. Diode-based RF rectifiers are usually used due to good conversion

efficiency from RF to DC power. Schottky diode HSMS-2850 was chosen due to its higher

conversion efficiency at low RF power [81]. As the RFID tag was intended to use at very

low received power (< -5 dBm), the DC voltage level generated by a single rectifier was not

sufficient to drive other modules. Hence, a voltage multiplier topology was adopted to boost

50

Figure 3.16: 4-stage charge pump circuit with matching network and DC voltage regulator.

up the rectified voltage. A conventional Dickson charge pump configuration was adopted as

shown in Fig. 3.16 [82]. The 8 diodes were connected along a long chain of 4-stage in this

configuration. A matching network was designed using CM = 10 pF and LM = 26 nH to

maximize power conversion at f0. However, the DC voltage at the charge pump output was

not regulated. Hence, a 1.8 V Linear Drop Output (LDO) voltage regulator (TPS79718Q)

by Texas Instruments was used to supply regulated DC voltage within the circuit. Two

capacitors CDC = 10 F were used at the input and output of the voltage regulator to

maintain a stable DC supply and remove high-frequency noises. The measurement results

of regulated and unregulated DC voltage are shown in Fig. 3.17. The EHU was capable of

generating regulated 1.8 V at the minimum input power of -12 dBm.

The input impedance of the EHU is shown in Fig. 3.18, which shows good matching at

434 MHz. The designed EHU had a measured quality factor of 8 with the matching network.

3.3.2.3 Digital Modulation Unit (DMU)

The DMU provides the ‘on’ or ‘off’ ASK modulation signal to the HD according to the

ID. A simple design from [83] was adopted in this work for the modulation scheme. The

design uses a parallel to serial 8-bit shift register to load up the stored digital data and

51

Figure 3.17: The unregulated and regulated DC voltage generated by the EHU.

Figure 3.18: Input impedance of the EHU with marker at 434 MHz.

52

transmit it. When the LD pin of the shift register (SN74HC165PWRG3) was activated, it

loaded the 8-bit ID in parallel and then transmitted serially upon deactivation of the LD

pin. The timing diagram of the serial output pin due to clock and LD input signal is shown

in Fig. 3.19. The LD signal should be deactivated for 8 clock periods as the shift register

sends 8 bit of data signal serially during that period. The in-phase LD signal was derived

from the clock signal using four D flip-flops. The block diagram of the LD signal generation

from the clock along with the serial ID is shown in Fig. 3.20. The low power D flip-flops

(SN74AUP1G80DCKR) generated the LD signal at frequency fclk/8, where fclk was the

frequency of the clock signal. An ultra-low power oscillator by Micro Crystal was used to

generate the clock signal at 32.768 kHz. Each component of the DMU was designed to

work at least at 1.8 V. However, the complete DMU can work properly without bit loss at

comparably lower DC voltage of 1.5 V. Implementation of the DMU using this architecture

would be cost and power effective and easy to implement compared to a microcontroller,

where programming is required. The number of ID bits is scalable by using multiple shift

registers and synchronizing the clock timing. Serial output from the shift register was fed

to a voltage divider to convert the CMOS voltage levels into NLTL realizable voltage levels,

which was 0.6 V for high state and 0 V for low state.

3.3.2.4 Power Splitter (PS)

In contrast to conventional RFIDs, the tag was designed to operate in a continuous

mode, which means the tag would send the ID data continuously during interrogation. In

this mode, the EHU and HD need to operate simultaneously. Hence, a PS was implemented

to divide the received power into two parts: 1) The first part would be harvested in EHU

and power up the DMU 2) the second part would provide the fundamental input signal

53

Figure 3.19: The different stages of signal generated by the DMU. Description from top to
bottom: 1) Serial output of the shift register, which is the ID, 2) LD signal, and 3) Clock
signal.

Figure 3.20: The D flip-flop setup to generate the LD signal from the Clock signal.

54

Figure 3.21: DC power consumption and harvest by HD, DMU, and EHU independently.

to the HD. Apart from the DMU, the HD also consumed DC power at logic ‘high’ state.

However, the power requirement by the DMU and the HD unit was different. To determine

the DC power needed by these units, the DC power consumption of each unit was measured

independently at varying RF input power, as shown in Fig. 3.21. The generated DC power in

EHU was calculated from the efficiency of the Schottky diode as described in [84]. The DMU

consumed a constant DC power of 32.5 W irrespective of the input RF power. However,

the DC power consumption by the HD was dependent on the input RF power. The total DC

power consumption was the addition of HD and DMU power consumption. The tag can only

operate properly when the harvested DC power is more than the total DC power consumed.

From Fig. 3.21, it is evident that the majority of the DC power was consumed by the DMU

compared to the HD at low RF power. Hence, more RF power should be channeled into the

EHU compared to the HD to minimize the RF activation power of the tag.

For comparison, two studies were performed, one with equal and the other with unequal

55

Figure 3.22: Total power consumption and harvest for equal and unequal power splitting.

Figure 3.23: Power splitter design with unequal power division.

56

Figure 3.24: The tag circuit photograph with different components as compared to a penny.

power splitting. In the first case, the PS divided RF signal equally to the EHU and HD.

Whereas in the second case, the EHU received 6 dB more power compared to the HD. Both

the cases were plotted in Fig. 3.22 and it can be noted that the unequal power splitting

activated the tag at 3 dB less power compared to equal power splitting. However, due to

unequal splitting, the HD received 4 dB less RF input signal compared to equal splitting,

which means the generated output harmonic power would be lower for unequal splitting,

which was compensated by increasing the LNA gain in the interrogator.

The PS was designed as shown in Fig. 3.23 for unequal division. Load impedance of

50 Ω at (f0) was considered at all input and output ports of the PS design. After the

design and measurement validation of each component, the complete tag was assembled on

a printed circuit board (PCB) using surface mount discrete components. A photograph of

the completed tag circuit is shown in Fig. 3.24. Antennas were attached with the tag circuit

using SMA connectors to form a complete harmonic RFID tag.

57

Table 3.2: Comparison with other work reported in literature.

Ref

Technology

Architecture

Microcontroller

Harmonic RFID

Microcontroller

Single frequency RFID

0.13 m CMOS
Schottky diode

Single frequency RFID

EPC Gen2

Harmonic RFID

ID

Yes

Yes

No

Yes

Yes
Yes

Frequency

Min. Input

Read Range

Band

915 MHz-
1.8 GHz
915 MHz

915 MHz
1.2 GHz-
2.4 GHz
900 MHz

900 MHz
434 MHz-
868 MHz

Power
0 dBm

0 dBm

-14 dBm

NA

-18.5 dBm

@ Iload = 1.5 A

-22 dBm
-6 dBm

NA

4.5 m @ 30 dBm

transmitted

7 m

40 cm @ 8 dBm

transmitted

4.5 m @ 4 W EIRP

transmitted

NA

1.8 m @ 18 dBm

transmitted and 7.2 m
@ 30 dBm projected

[1]

[85]

[86]
[87]

[88]

[89]

0.35 m CMOS

Single frequency RFID

NA

Single frequency RFID

This work

Discrete component

Harmonic RFID

3.3.3 Reader design

An interrogator was developed as shown in Fig. 3.25 to read out the ID from the harmonic

RFID tag. A stable frequency synthesizer was used to generate the 434 MHz (f0) signal.

The (f0) signal was amplified at 18 dBm using a power amplifier (PA). As the receiver

works at the second harmonic (2f0) frequency, it was important to mitigate the transmission

of harmonics, which were generated from the PA. Hence, a low-pass filter with a cut-off

frequency of 500 MHz was used at the output of the PA and before antenna element. When

a harmonic tag is activated within the illuminated area it sends back the modulated (2f0)

signal to the interrogator. Once the interrogator receives the harmonic (2f0) signal from a

tag, the signal is passed through a high-pass filter to eliminate the clutter at (f0). It should be

noted that the reduction of coupling between the receiving and the transmitting antenna at

both (f0) and (2f0) enhances the SCR, antenna performance, and the maximum detectable

range. Hence, in addition to the filters and narrowband antennas, the transmitting and

receiving antennas were aligned in a cross-polarization position in both tag and interrogator

to further minimize coupling. The received RF signal at (2f0) was then amplified using an

LNA with 34 dB gain before feeding it to an envelope detector. As the returned harmonic

power has a weak signal, a sensitive detector diode was used for the envelope detection. In

the subsequent stage, an RC low pass filter was used to obtain only the low-frequency ID

58

signal to remove the high-frequency RF signal. Once the time domain demodulated sampled

data is acquired from a digitizer, sample averaging is performed followed by pulse shaping

to obtain the digital ID bits.

3.3.4 Wireless measurement

Figure 3.25: Integrated interrogator system.

After assembly of the board, the harmonic tag was programmed with an ID sequence of

‘01100101’ and kept at a distance of 1.8 meters from the interrogator for the measurements.

Different processing stages of the demodulated received signal are shown in Fig. 3.26. The

SNR of the signal improved once it was averaged over 100 points. The pulses were shaped to

logic levels after comparing them to the zero crossing voltage value, which was dynamically

assigned from the average of ‘high’ and ‘low’ level of voltage magnitudes.

59

Figure 3.26: Demodulated received signal (a) after digitization, (b) after averaging, and (c)
after pulse shaping and bit recovery.

3.4 Harmonic RFID: Gen II

Although the Gen I tag satisfied all the functional conditions, it had few disadvantages,

which are prevalent in contemporary harmonic RFID tags. The reported harmonic tags have

a major disadvantage of using multiple antennas or inefficient energy harvesting mechanism.

Due to operation at two separate carrier frequencies for uplink and downlink communication,

miniaturized compact antenna design is a challenge for harmonic RFID. The total antenna

occupied area increases dramatically for multiple antennas [38, 41]. The harmonic RFID tag

was improved in Gen II version. The primary features of Gen II harmonic RFID system are:

• Gen II harmonic RFID tag has programmable and expandable memory for ID gener-

ation.

• Number of antenna at the tag was reduced to one. The reader still uses two antennas

like in Gen I.

• This tag does not require extra RF power splitter. Hence, the number of functional

components is reduced.

60

Apart from antenna design, harmonic RFID tag circuit configuration also needs to be

improved for enhanced harvested power. In RFID operation, the modulation is performed by

periodically changing the antenna impedance, shorting for one state and matched termination

for the other state. When the antenna is shorted, no RF energy enters in the tag and hence

the tag requires a latency time for sufficient energy harvesting before starting modulation. In

other harmonic RFID tag configuration, reflection mode harmonic back-scattering was used

similar to conventional RFID, where the entire RF energy is directed towards the harmonic

generator at one of the modulation states [1]. In both cases, the energy harvester does not

harvest for one of the modulation states. The clock speed for modulation can be increased or

additional energy hungry operation such as sensing or computation can be included within

the tag ciruit by harvesting energy during both the modulation states.

In this work, a compact single antenna based harmonic tag is proposed, which can harvest

RF energy during entire modulation period. As the development was primarily done for

tag circuit, development of Gen II tag is only discussed here. The harmonic tag has two

components 1) antenna and 2) tag circuit. The block diagram and the fabricated tag design

is shown in Fig. 3.27. First, the dual frequency compact antenna design procedure and

response is shown. In following, detailed design of NLTL integrated energy harvester as a

harmonic RFID tag is discussed.

3.4.1 Antenna

The design procedure of the dual frequency antenna is described step by step in this sec-

tion. One primary objective of the antenna design is miniaturization. Among different tech-

niques, meandering is a popular choice for miniaturized antenna from design perspective [78].

However, a dipole antenna usually starts to show as a narrow-band behavior with increase

61

(a)

(b)

Figure 3.27: (a) The block diagram and (b) fabricated part of the single antenna harmonic
tag.

in meandering [79]. Now, if the responses at both fundamental and harmonic frequency are

narrow-band, with a frequency shift due to fabrication tolerance, the new narrow-band har-

monic resonance may not be exactly twice of the new narrow-band fundamental resonance.

Hence, making one of the two resonance frequency wide-band will minimize the issues from

fabrication tolerance. The harmonic frequency was chosen for the wider band resonance for a

compact tag size. First, a simple bow-tie was designed at the harmonic resonance frequency.

In the next step, meandering was performed at the side edges of the bow tie until a second

resonance is observed at the fundamental frequency. Finally, the design was fine tuned to

obtain good matching at both fundamental and harmonic resonances. However, it should be

noted that although the meandering will create a second resonance, the radiation will not be

efficient at that frequency. As meandering is performed to miniaturize the antenna at low

frequency, the bandwidth at low frequency is very narrow.

The antenna design starts with a simple bow-tie as shown in Fig. 3.30(a). A planar

feed is added to connect the circuit with the antenna. First the bow-tie is optimized for

a broad-band response at the harmonic frequency of 900 MHz. At this stage, the antenna

has bandwidth of 110 MHz with center frequency at 900 MHz from the simulation result.

62

Figure 3.28: Dual frequency designed antenna. All the dimensions are in mm.

Figure 3.29: The reflection co-efficient of the dual frequency antenna.

63

Figure 3.30: The step by step procedure for dual frequency single antenna design.

In the consecutive step, slits are cut at the ends of the bow tie to provide effective line

length for the fundamental resonance frequency as shown in Fig. 3.30(b). As shown in

reflection co-efficient response, the harmonic resonance frequency remains almost unaltered

for structure in Fig. 3.30(b) due to negligible edge current at harmonic frequency. At

fundamental frequency, the new resonance frequency is obtained at 445 MHz. At this stage,

the antenna has bandwidth of 5 MHz with center frequency at 445 MHz and bandwidth

of 70 MHz with center at 935 MHz. The shift of harmonic resonance frequency center is

due to the slits. At the final stage, the antenna was fine-tuned as shown in Fig. 3.30(c) to

obtain desired fundamental and harmonic frequency response with center at 445 MHz and

890 MHz respectively. The simulation was performed using HFSS by ANSYS to validate the

final design before fabrication.

In Fig. 3.29, the simulation and measured results are shown for the designed antenna as

in Fig. 3.28. Simulation was performed using Ansys HFSS and the antenna was fabricated on

64

top of Rogers RO4350 board using standard lithography process as outlined in APPENDIX

B. Due to fabrication tolerance, the fundamental resonance frequency has shifted by 11

MHz. As 70 MHz bandwidth was considered at the designed harmonic resonance, with

the fundamental resonance frequency shift, the antenna still resonates at the harmonic of

the new fundamental resonance. the post-fabrication final working pair of fundamental and

harmonic frequency pair are at 434 MHz and 868 MHz respectively.

3.4.2 Tag circuit

The tag circuit primarily has three components: 1) Energy harvester, 2) Nonlinear Trans-

mission Line (NLTL), and digital circuit. NLTL can be realized by periodically arranged

inductor and varactor diodes with detailed design procedure as demonstrated in [41]. The

NLTL generates multiple harmonics when interrogated by a single frequency RF signal un-

der activation due to the nonlinear varactor diodes. In Fig. 3.31 the NLTL response with

multiple stages and different bias conditions is shown.

When measured using a VNA, the NLTL with different number of stages show a broad-
band response from 100 MHz to 1 GHz. The broadband response remains unaltered (S11 ≤

-10 dB) upto 900 MHz for a two stage NLTL under forward bias condition from 0 V to 0.6

V as shown in Figs. 3.31(a) and 3.31(c). At 0 V bias, the transmission loss in 2-stage NLTL

is primarily due to lossy inductors. When the NLTL is driven at forward bias condition, due

to sub-threshold conduction of the varactor diodes, the NLTL starts behaving as a lossy line

and the transmission loss increases with increase in bias as shown in Fig. 3.31(d). Addition-

ally, the loss also increases with higher number of NLTL stages at higher applied bias. The

purpose of applying DC bias is to generate strong harmonics from the NLTL. In Fig. 3.32(a),

the conversion loss for different number of NLTL stages is shown. The maximum harmonic

65

(a)

(b)

(c)

(d)

Figure 3.31: Measured results for (a) S11 and (b) S21 for different number of NLTL stages
at no bias condition; (c) S11 and (d) S21 for two stage NLTL at different bias.

66

(a)

(b)

Figure 3.32: Measured results for (a) conversion loss and (b) DC power consumption for
different number of NLTL stages.

is generated close to 0.6 V and the the conversion loss increases with reduction or further

increase in bias voltage. Due to forward biasing, the NLTL consumes very small amount

of DC power and the power consumption for different number of NLTL stages is shown in

Fig. 3.32(b). Based on the power consumption, conversion loss, component requirement,

and transmission characteristics, 2-stage NLTL was chosen for the tag circuit.

However, the broadband behavior of the NLTL changes when it is terminated with the

energy harvester in place of a 50 Ω load. Impedance matching is required for the energy

harvester at the fundamental and harmonic frequency regions for the desired matching of the

NLTL at both operating frequencies. This deign approach helps in realizing two objectives:

1) replacement of the 50 Ω load, and 2) energy harvest while modulation. When the 2-stage

NLTL is at ’off’ state or under 0 V bias condition, the measured S21 is -1.2 dB and at

’on’ state or 0.55 V bias the S21 is -1.21 dB at 434 MHz. Hence, the energy harvester is

receiving almost same amount of input power at both the modulation states. A 10 stage

Schottky diode based charge pump configuration was used for energy harvesting as described

in detail in [41]. The reflection co-efficient of the energy harvester only and energy harvester

67

Figure 3.33: Measured reflection co-efficient of the energy harvester with and without NLTL.

coupled with NLTL is shown in Fig. 3.33.

In both cases, good matching is observed at

434 MHz, at which the energy harvesting is performed. The DC power harvested by energy

harvesting unit is consumed by the digital circuit for generating the ID. A low power voltage

regulator was used in between the energy harvesting unit and digital circuit. The major part

of the digital circuit is microcontroller (MC), which was programmed for the bit sequence to

generate the modulation. The voltage level from the MC output was level shifted by series

diodes to the NLTL realizable voltage level.

3.4.3 Wireless measurement

Once the tag was designed and fabricated, wireless measurement was performed to find

the maximum operable range of the harmonic RFID tag. Reader circuit as described in [41]

was used to retrieve the information or ID from the harmonic tag. The reader transmitted

30 dBm of power at 434 MHz in continuous mode. A separate antenna was used to receive

the harmonic power at 864 MHz at the reader. The output power is visualized directly in an

68

Figure 3.34: Measured harmonic voltage at different distance and modulation frequency.

Oscilloscope after the amplification of the harmonic power received from the tag. The power

consumption in the MC is directly related to the frequency of the operating internal clock.

To show that the tag is capable of harvesting during both the modulation cycles, the MC was

programmed for periodic bit stream of ‘1’ and ‘0’ continuously. In the first experiment, the

internal clock was set at 32 kHz, which corresponds to 1 kHz of data modulation frequency.

At this clock, the current consumption of the MC was 17 µA at 3 V bias. With this setup,

the tag could operate at 10 m from the reader. The demodulated harmonic periodic signal

is shown in Fig. 3.34. In the next part of experiment, the MC was programmed at 125 kHz

internal clock, which corresponds to 4 kHz of data modulation. The sequence of bit-stream

‘1’ and ‘0’ was kept same as earlier. At this clock rate, the power consumption went up

significantly higher at 109 µA at 3 V bias. Due to increased power consumption, the tag

could perform the modulation as expected upto 4 m from the reader. When the distance

was increased to 7 m, the bit error rate increased significantly within MC due to significant

increase in power consumption. The demodulated harmonic signal is shown in Fig. 3.34 at

69

(a)

(b)

Figure 3.35: Efficiency calculation block diagram of harmonic RFID for (a) harmonic effi-
ciency and (b) energy harvester of charge pump.

4 kHz data modulation for both at 4 m and 7 m. The tag is deactivated completely once

the distance in increased further at 4 kHz of data modulation. As expected, the read range

decreases with higher clock rate operation as the power consumption dominates over the

power harvest.

3.5 Harmonic RFID Tag Efficiency

The harmonic RFID tag has multiple components and the overall efficiency analysis of

harmonic generation is also required. The efficiency of NLTL and different antennas has

been shown earlier separately. The cumulative efficiency of the harmonic RFID tag is shown

in Fig. 3.35 with individual efficiency of antenna, energy harvester charge pump (CP),

and harmonic generator. The individual efficiency of antenna at fundamental frequency

is given as ηAN T F , at harmonic frequency as ηAN T H , CP efficiency as ηCP , and NLTL

efficiency as ηN LT L. The cumulative efficiency of the harmonic generation system will be
ηAN T F × ηN LT L × ηAN T H . Also, the harvesting efficiency should be enough in parallel, so

that the computation part of the tag can generate the modulation signal. The cumulative
efficiency of the antenna and CP system will be ηAN T F × ηCP . A harmonic RFID tag has

70

a minimum sensitivity of Psen, below which the tag is not activated. If the received power

of harmonic RFID is Pin, then the cumulative efficiency of the harmonic RFID tag will be
ηAN T F × ηN LT L × ηAN T H , provided that (ηAN T F × ηCP × Pin > Psen).

3.6 Harmonic RFID Performance Under Clutter

After development and testing of the harmonic tag, efficacy of Gen I harmonic tag per-

formance over single frequency commercial tag was tested. First, performance of commercial

tag under clutter is measured. M6E Nano UHF RFID reader (-80 dbm RS) with circularly

polarized antenna (8 dBi gain) was used to monitor the single frequency tag backscattered

signal at 926 MHz. The reader internally generates 10 dBm of RF power, which was fed to

the antenna through two circulators as shown in Fig. 3.36. Transmit power of 10 dBm was

chosen for small area anechoic chamber requirement. To emulate the scenario of 30 dBm

RFID transmit power, extra 20 dB attenuation was added to the tag emitted return signal.

The 20 dB attenuation represents the extra 20 dB path loss. The complete setup as shown

in Fig. 3.37 enables test experiment of 30 dBm transmit power within a small chamber.

First, the dynamic range (DR) was quantized in terms of correct tag ID detection counts

per second. With reduction in transmit power, the DR would decrease with fixed reader

sensitivity and hence the detection rate would drop. Impinj Monza R6 UHF RFID tag was

selected as commercial single frequency RFID tag for its high sensitivity (-22 dBm). The

detection rate was measured with a=52 cm without any presence of reflectors as shown in

Fig. 3.38. As the transmit power was decreased from 10 dBm to 5 dBm, the detection rate

dropped from 15.67 counts/s to 2 counts/s. Once the DR was quantized, the effect of a

reflector was measured in terms of detection rate.

71

Figure 3.36: Measurement setup for performance comparison.

Figure 3.37: Measurement setup with commercial RFID tag at the center.

Figure 3.38: Correlation in between transmitted power and tag detection count.

72

(a)

(b)

(c)

Figure 3.39: Power profile for 10 dBm transmitted power at 926 MHz with (a) no tag or
reflector, (b) only tag is present, and (c) tag in presence of reflector.

The transmitted carrier signal was fixed at 10 dBm and 926 MHz for the measurements

as shown in Fig. 3.39 and the return signal was measured using a spectrum analyzer at

bandwidth of 1 KHz. When the tag is not present within the reader antenna’s field of view,

only the reader query modulation signal is visible as shown in Fig. 3.39(a). When the tag

was kept at a=52 cm, the perturbation in power spectrum was clearly visible as shown in

Fig. 3.39(b). The detection rate was found to be 15.5 count/s without any reflector. When

a reflector of size (15 cm X 25 cm) was kept at b=10 cm and c=44 cm, the tag was still

activated as shown in Fig. 3.39(c) but the detection rate dropped to no detection due to

increased clutter level. The clutter from the reflector obstructed the RFID tag signal by

dropping the DR level. However, it should be noted that the actual tag response signal is

buried under the reader query signal as both of them operate within 500 kHz channel using

time multiplexing. With the query and tag modulation signal superimposed on top of 926

MHz single tone, It is not possible to experimentally demonstrate only the single tone phase

noise behavior as explained in Fig. 2.3 with the commercial RFID system. However, it

can be noticed that the peak amplitude of the return signal is 4 dB higher in Fig. 3.39(c)

compared to Fig. 3.39(b). The increase in peak power is due to the reflected signal from

73

reflector analogous to Fig. 2.3.

Next, the clutter effect was measured on the designed harmonic tag. The tag and a

square reflector (30 cm X 50 cm metal foil) was placed at a=104 cm, b=20 cm and c=88 cm

using similar setup as shown in Fig. 3.36. The distances a, b and c were scaled according

to the wavelength of operation. The RFID reader was replaced by a 434 MHz source and

the reflected power from tag was monitored using an antenna resonating at 868 MHz at the

source side. During the measurement, the transmitter power was set at 18 dBm at 434 MHz

after power amplification. When no reflector was present, the read count was measured to

be 2 count/ms. In presence of the reflector, the read count barely changed to 1.98 count/ms.

The read counts were averaged with 10 measurements. When only the harmonic power was

measured, it barely changed (<0.5 dB, within measurement error) in presence or absence of

any reflector. This experiment clearly shows that an RFID operating at single frequency will

suffer a decrease in DR due to increase in background clutter as compared to a harmonic

RFID.

3.7 Summary

The design and implementation of a complete ultra-low power Gen I harmonic RFID tag

with integrated antennas is demonstrated. The harmonic tag can operate at a minimum

received signal strength of -6 dBm. A simple interrogator is also designed to demodulate the

ID from the harmonic tag. A read range of 1.8 m was achieved with the transmitted power

of 18 dBm. The read range can be extrapolated to 7.2 m based on one-way path loss with 30

dBm power transmitted while achieving good SCR of the return signal. As long as the tag

can receive the minimum activation power (-6 dBm), it can generate the harmonic and the

74

weak harmonic signal can be amplified at the interrogator to compensate the return path

loss. In comparison to tag designs reported in open literature, this tag provides superior

performance with respect to read range while working in a cluttered environment. The read

range can be further increased by using high gain antennas, high transmitted power, low

power tag circuitry and a sensitive detector. The harmonic RFID tag system demonstrated

here holds significant potential for a range of tagging applications.

A miniaturized Gen II single antenna based harmonic RFID tag is demonstrated next.

With the use of a simplified design architecture, minimum number of components was used

for the harmonic RFID. The design can be easily adopted by replacing the antenna modula-

tion switch in conventional RFID with the harmonic generator NLTL. Another advantage of

the proposed harmonic tag is continuous harvesting, which is not possible for existing single

frequency RFID antennas. At 1 kHz data modulation by the harmonic tag, a read range of

10 m was achieved. However, the read range decreased when the data modulation rate was

increased due to more power consumption. The circuit can be miniaturized by fabricating

the complete circuit using a monolithic process. Harmonic RFID tags are useful for under-

ground or biological applications, where the background clutter phase noise can dominate

over tag return signal.

Gen I harmonic RFID tag was also demonstrated for its improved performance and

was compared with a single frequency commercial state-of-the-art RFID tag in presence of

strong clutter source. From the results, it was clear that even the conventional tag was

activated in presence of reflector, the detection rate can drop due to poor SCR. On the

other hand, the clutter signal had almost no effect on the harmonic tag. Hence, conventional

single frequency tags would not be suitable for long range applications in presence of strong

background clutter. The harmonic RFID tag would be useful for RFID applications in case

75

of self-jamming and also in cluttered environment such as underground object tagging, long

range tag detection in an industrial set up with strong reflectors such a metal in the vicinity

of the tag.

76

Chapter 4

RFID With Integrated Sensor

The various drawbacks of sensor integration with conventional RFID system are provided

in Chapter 2. The most accepted platform for RFID integrated sensor system is wireless

identification and sensing platform (WISP) [55]. Diverse range of sensors were integrated

with the proposed RFID platform, light, temperature, camera, human activity, and neural

signal to name a few [90–93]. In the proposed works, the analog output from the sensors

is digitized and integrated with the ID bits of the RFID. A high level schematic of analog

sensor integration with digital interface of ADC and microcontroller is shown in Fig. 4.1.

In those architectures, primarily there are two major concerns: 1) power consumption at

the tag, and 2) impact of quantization error on sensitivity. As the passive RFID tags are

operated on harvested energy, it is very important to minimize the power consumption at the

tag to maximize the read range of the sensor integrated RFID tags. Hence, the read range

of sensor integrated RFID will be lower than the conventional RFID due to addition of ADC

functionality. The power consumption of the ADC should be carefully monitored to keep the

read range at a desired distance. Also, many times an impedance buffer is required between

the sensor output and ADC input. This scenario particularly arises when an ADC within

a microcontroller is used. The substantial internal microcontroller leakage path in between

sensor output and ADC input leads to requirement of impedance buffer, which consumes

further DC power. Another important scenario is quantization error, which arises due to

77

Figure 4.1: High level picture of analog sensor integration with ADC and microconroller.

analog to digital conversion. The quantization error leads to reduction of sensitivity of the

sensor. To address this, either more conversion bits are to be used or the analog reference

voltage level difference used for conversion should be reduced.

Among many applications of sensor integrated passive RFIDs, one major field is biosens-

ing. With recent advancements in healthcare, now it has become possible to monitor many

physical and chemical vitals of human body such as blood pressure, heart rate, glucose, pH,

uric acid, specific cations such as sodium and potassium in sweat [94–97]. Wireless configura-

tion of the proposed sensors are using either low energy bluetooth [97], or near field coupling

based communication methods [95]. In this work, batteryless and wireless communication

method of electrochemical pH sensor is proposed for biological applications.

4.1 Background

Batteryless wireless pH sensors received significant attention recently due to their longer

shelf life, reduced cost, wireless measurement, and less hazardous as a body implant. Over

the years, the need for wireless pH sensing has increased in healthcare due to ease of diagnosis

of different potential medical conditions such as gingivitis, periodontitis or gastroesophageal

reflux disease (GERD) by wirelessly monitoring pH of salivary or gastroesophageal fluid

[98–101]. Additionally, continuous wireless pH monitoring has a huge potential to track

78

the quality of meat and fish based perishable products in cold food supply chain [102, 103].

In literature, multiple wireless methods have been proposed to track the pH [96, 101–105].

Coupled coil based resonance frequency shift method is quite popular among the wireless

methods due to simple setup and least number of circuit components required. However,

as the resonance frequency strongly depends on the mutual coupling in between the sensor

and reader coils, the measurement accuracy depends on the orientation of the coils and

hence requires an initial calibration [105]. The measurement setup also requires experienced

personnel to interpret the sensor data accurately using instruments such as Vector Network

Analyzer (VNA) or Impedance Analyzer (IA). As an alternative, digital modulation method

is proposed in this work for pH measurement, which is less dependent on antenna coupling

and is compatible with commercial UHF RFID technology. In digital modulation technique,

the pH electrode voltage is converted into digital bits, which are transmitted wirelessly

[96, 101].

In digital systems, identification (ID) can be integrated easily along with the

sensors, which is an integral component to distinguish among similar objects. In Fig. 4.2, a

scenario is shown where more than one identical pH sensor tags are present. The handheld

interrogator sends a RF signal, and the sensor tags reply with different IDs. Without the

physical ID, it will be very difficult for a machine to separate the data from identical sensors.

However, digital modulation requires a DC power supply module to drive the modulation

circuit, which makes the sensor design and reader setup complicated.

In this batteryless

pH sensor, the battery is replaced with an energy harvesting circuit to make the pH sensor

compatible as a body implant, because the hazardous chemicals can be harmful in case of

battery leakage [106].

Apart from different sensing methods, different types of pH electrodes are also well re-

searched. pH is quantified as the H+ ion concentration within a fluid environment and is

79

Figure 4.2: Interaction of a handheld reader with identical pH sensors with different IDs.

electrically measured using a combination of reference electrode and a pH sensitive elec-

trode. Double junction glass electrodes are commonly used due to high linearity and stabil-

ity over a wide pH range (0-14). However, the glass electrode fabrication is not compatible

with standard photolithography, making it difficult to integrate the sensor with standard

CMOS process. Hence, a combination of different type of metal oxide electrode with sil-

ver/silver chloride (Ag/AgCl) reference electrode was developed to fabricate pH microelec-

trodes along with solid state readout circuit.

In literature, different type of pH sensitive

electrodes such as iridium oxide, antimony oxide, mixed metal oxide, carbon nanotube

are reported [104, 105, 107]. Among these, iridium is a costly transition metal and car-

bon nanotube electrode requires complicated fabrication process. Hence, a cost effective

antimony/antimony oxide (Sb/Sb2O3) electrode was chosen for the pH sensor design.

In this work, we first present the design, fabrication and characterization of Ag/AgCl

and Sb/Sb2O3 based combination pH electrodes along with the sensor readout electronic

circuit. In next section, the design and fabrication of pH electrodes and sensor tag circuit is

80

discussed. Next, the measurement results are shown to demonstrate a fully working wireless

ID integrated pH sensor and finally a potential application of this work is demonstrated.

4.2 Design and Fabrication

The measurement system consists of two main parts: 1) a pH sensor electrode pair with

an electronic readout circuit, and 2) an RF interrogator to wake up the digital sensor and

extract the ID and sensor information wirelessly. First, the working principle of the complete

system is presented. Then, the pH electrode fabrication process is described followed by the

design of the readout circuit.

4.2.1 RF interrogator

The RF interrogator is required to activate the sensor wirelessly and extract the sensor

data. The RF source generates a continuous RF signal at 900 MHz, which is split equally in

two branches using a power splitter. One signal branch is power amplified and transmitted

through an antenna resonating at 900 MHz. A meandered dipole antenna from [41] is used
for this purpose, which has reported gain of 2.1 dBi in simulation using Ansys HFSS R(cid:13).

Once the sensor replies back with the modulated RF signal, the reply signal is mixed with

the other signal branch in a RF mixer to demodulate the sensor signal. The RF circulator

enables the transmission of continuous RF signal and reception of the modulated signal using

a single antenna. A baseband amplifier with 26 dB gain and a cut-off frequency of 150 kHz

was used to further amplify the low power demodulated signal with high frequency spurious

signal rejection before feeding it to the scope. The block diagram of the complete setup is

shown in Fig. 4.3.

81

Figure 4.3: The block diagram of the measurement setup with an interrogator and the RFID
sensor.

4.2.2 Readout circuit

The electronic readout circuit at the sensor consists of two major parts: 1) an RF front

end with antenna for energy harvesting, and 2) a digital back end to process the analog

sensor signal into digital bit sequence and to generate the ID. The RF communication works

at 900 MHz UHF RFID frequency. When the sensor tag receives RF power, it first harvests

the RF energy and converts it into DC for biasing in order to drive the sensor and the digital

circuit. HSMS 2852 low barrier Schottky diode is used to harvest energy from the incoming

RF signal. Six diodes are used in a Dickson charge pump configuration to boost up the DC

voltage at desired level. The details of the charge pump rectifier circuit design is provided

in our earlier work [41]. The generated DC voltage is rectified using a low power voltage

regulator (TPS79730) at 3 V to drive the digital circuit. The schematic of the electronic

readout circuit is shown in Fig. 4.4. A low power compact microcontroller (PIC12LF1822) is

used to generate the digital ID sequence. An ADC is required to convert the analog voltage

generated by the electrode pair into a digital bit stream. The use of internal ADC within

the microcontroller was avoided as the DC current consumption increases by 335 µA upon

82

Figure 4.4: The sensor voltage readout digital circuit with ID generation.

the ADC activation. Hence, an external ultra-low power 8 bit SAR ADC (ADS7040) is used.

Apart from ID, the microcontroller also generates the control signals for the ADC. Finally,

a low power ‘OR’ gate (SN74AUP1G32) is used to combine the ID and the ADC bits into a

single channel. Once the digital circuit generates the single channel digital bit sequence, a

high frequency switch (BF1105R) is used to modulate the antenna impedance according to

the digital bit sequence [108]. A detailed analysis is shown in Chapter 1 about the choice of

two impedance states for good modulation depth at the reader. Based on the analysis, 50

Ω and 0 Ω were chosen across the tag antenna load for digital ’0’ and ’1’ modulation depth.

The modulation is captured at the interrogator and the digital bit is demodulated for pH

measurement.

4.2.3 pH sensor fabrication

The pH electrode pair consists of one Ag/AgCl electrode and another Sb/Sb2O3 electrode,

fabricated on top of a 0.38 mm thick flexible PTFE Rogers RT/duroid 5880 board. PTFE

83

based substrate is chosen as it has a very low moisture absorption (0.02%) capacity. Hence,

the effect of potential change due to the solution in contact from the backside of the electrode

is negligible. At the initial step, the copper metal on top of the substrate was patterned

and etched to create a ‘T’ shaped backbone interconnect for the electrodes. Between the

two electrodes, Sb/Sb2O3 electrode was fabricated first since it requires a high temperature

oxidation process. S1813 negative photoresist from Shipley was patterned to create an inverse

mask for a subsequent lift-off process. A 6 nm thick titanium (Ti) layer followed by a 2.5

µm thick Sb layer was sputter deposited on top of the substrate. The thin Ti layer helps in

better adhesion of Sb layer to the substrate. After deposition, lift-off process was followed

to remove the unwanted Sb layer. The selectively deposited Sb metal was oxidized using a

Nordon March RIE 1701 at 99% oxygen concentration followed by annealing. The annealing

process was performed by ramping the temperature from 100 oC to 300 oC in 3 equal steps

with 10 minutes time duration at each step. Rapid and high temperature heating was

avoided to prevent forming of cracks in the thin continuous Sb film due to thermal stress.

The process of oxidation followed by annealing was performed twice for obtaining a thicker

oxide layer. As shown in Fig. 4.9(b), the antimony electrode surface becomes uneven during

heat treatment. Hence, a thicker oxide will reduce the possibility of leaving an oxide layer

defect.

The Ag/AgCl electrode was fabricated next on the other half of exposed copper backbone.

Ag was deposited in similar way by sputter depositing a 2.5 µm thick Ag layer on the inverse

photoresist mask followed by a lift-off process. A thin layer of AgCl was formed on top of

the Ag electrode using an electrochemical chlorination. In the chlorination process, 2 V was

applied at the Ag electrode with respect to a Pt cathode of 0.5 mm diameter for 50 s in a

0.1 M potassium chloride (KCl) solution. The Ag/AgCl electrode was then heated at 90 oC

84

Figure 4.5: Fabrication procedure of pH combination electrodes a) Cu plated RT/duroid
5880 substrate; b) patterning of Cu layer to create ‘T’ shaped Cu backbone interconnect; c)
sputtering followed by lift-off of Sb metal; d) oxidation of Sb layer; e) sputtering followed by
lift-off of Ag metal; f) chlorination of Ag layer.

for 20 minutes to allow a uniform diffusion of chloride ions. Following the AgCl deposition,

the electrode was dip-coated with a protective Nafion layer by dipping in 5% Nafion solution

of Methanol (Liquion-1105-MeOH) from Ion Power for three times to create a thick layer.

After dip coating, the electrode pair was dried at 110 oC for 1 hour in a vacuum chamber to

evaporate away methanol [109]. Nafion was chosen for its ion exchange selectivity as it allows
the H+ cations to pass through but prevents Cl− anions to migrate from the electrode. The

complete fabrication process of the electrodes is described in Fig. 4.5.

The contamination of thin Ag or Sb layer can happen by interdiffusion of Cu atoms from

the underlying layer [110]. Additionally, as the Ag layer or Sb layer is very thin, irregularity

in the deposition can create small pockets, through which the electrolyte can directly interact

with Cu layer, and thus can change the desired potential. Hence, a continuous Cu backbone

layer was avoided beneath the electrodes. Detailed elemental analysis of the pH electrodes

is demonstrated later.

85

Figure 4.6: Voltage response of the fabricated electrode in different pH solutions.

4.3 Results

4.3.1 Electrode performance

The pH electrode pair performance was characterized in terms of 1) stability, 2) repeata-

bility, 3) sensitivity, and 4) response time. The output voltage was measured from the

electrode pair using a digital multimeter with an input impedance of 2 MΩ. The voltage

reading from the multimeter was acquired using LabView. All the buffer solutions used for

pH measurement were prepared using proper mixture of Potassium diHydrogen Phosphate

(KH2PO4) in water with 0.1 M Sodium Hydroxide (NaOH) solution. The pH of all the

buffer solutions were measured using a commercial pH meter (Checker plus HI98100) from

Hannah Instruments.

First, the linearity and sensitivity of the electrode pair were measured. The pH of the

buffer solution was varied from 4.3 to 10.7 and the output voltage was measured as shown

86

Figure 4.7: Stability of the voltage response of the electrode in different pH solutions.

Figure 4.8: Repeatability test of the electrode in different pH solutions.

87

in Fig. 4.6. During measurement, the Ag/AgCl and Sb/Sb2O3 electrodes were connected

respectively to the positive and negative terminals of the multimeter. The R-square value

shows that the voltage response as a function of pH is fairly linear. Also, the electrode pair

shows good sensitivity of 46 mV/pH, which is consistent to earlier reported result on a similar

thick electrode [104]. The electrode pair was washed in DI water in between measurements

to avoid contamination from different buffer solution mixing. The measurement was taken

five minutes after dipping the electrode pair in the target solutions to provide enough time

for voltage stabilization.

Once the sensitivity of the electrode pair was measured, next the voltage stability was

monitored. The sensor electrode is dipped into different buffer solutions of pH 4.3, 7, and 8.6

and the voltage stability was measured for more than 5 minutes as shown in Fig. 4.7. The

maximum voltage deviation is noticed to be 3.8 mV among all the three pH buffer solutions

over time.

Next, the repeatability of the sensor electrode was measured in three different pH so-

lutions. The sensor electrode was dipped into each pH solution and output voltage was

recorded for 100 seconds at an interval of 1 second. As shown in Fig. 4.8, the electrode was

first dipped into pH 4.3 solution, then it was transferred to pH 7 solution and finally to pH

8.6 solution. The same procedure was repeated in reverse order to observe the repeatability.

The maximum voltage response time is 10 s including the electrode transfer time from one

solution to another. The presented thin film electrode helps in reducing the response time

by 20 s compared to a thick film based electrode presented in [104]. As observed in Fig. 4.8,

the settling time in pH 7 solution is smaller compared to other two pH solutions as the pH

7 solution is more stable compared to the others. However, it should be noted that while

performing the repeatability test, the sensor electrode was not washed in DI water before

88

transferring from one pH solution to another. As the test pH solution volume was low (∼10

mL) contamination of pH solution can happen during electrode transfer. Among the three

test solutions, pH 4.3 was the weakest and hence the drift in voltage is more noticeable for

pH 4.3.

4.3.2 Elemental analysis of sensor electrodes

Elemental analysis of the electrodes was performed to verify the material composition

of the electrodes. The samples were examined in a JEOL-JSM-7500F Scanning Electron

Microscope (SEM) and Energy Dispersive X-ray Spectroscopy (EDX) was performed using

an Oxford Instruments AZtec system, using a 20 mm2 Silicon Drift Detector crystal (6610LV)

and an ultrathin window. The samples were coated with 10 nm thick osmium for SEM

imaging and examined across cross-sectional cut.

Figure 4.9: Locations for EDX elemental characterization: (a) reference electrode with high-
lighted black box for Ag/AgCl layer and red box for Nafion layer, (b) pH sensitive electrode
with highlighted black box for Sb/Sb2O3 layer.

In Fig. 4.9, the locations for elemental analysis is highlighted for both reference and pH

sensitive electrode. Two different sites for reference electrode were chosen as it consists of one

Ag/AgCl layer and another Nafion layer. The EDX report as in Fig. 4.10 of the highlighted

black region from Fig. 4.9(a) shows presence of both Ag and Cl atoms with lower weight

89

Figure 4.10: Elemental characterization for Ag/AgCl layer.

Figure 4.11: Elemental characterization for Nafion layer.

Figure 4.12: Elemental characterization for Sb/Sb2O3 layer.

90

percentage of Cl atoms. This shows that Cl atoms are primarily accumulated at the surface.

Fig. 4.11 shows the presence of C, F and O atoms, and the composition closely follows with

earlier reported Nafion composition [111]. A trace amount of Na or K atoms were found,

which can attach at the electrode surface during chlorination process. The presence of Sb

and O atoms in Fig. 4.12 shows the oxidation of Sb layer. However, a comparatively small

weight percentage of O atoms means a thin layer of oxide formation. The presence of Ti is

due to the adhesion layer and C and F atoms are from the duroid substrate.

4.3.3 Multiple sensor electrode response

Due to fabrication tolerance, the voltage response of different pH sensors can be different.

To study the phenomenon, voltage response of three different pH sensors was measured and

is shown in Fig. 4.13. It can be noted that the sensitivity of all the sensor electrodes is

close and similar. However, there is a specific offset among the different electrodes, which

can arise due to fabrication process variation. Hence, one-time initial calibration is required

with standard pH solutions before proceeding with measurement. In this paper, all the test

results reported are based on Device 1.

4.3.4 Readout circuit

The digital signals for ID, ADC output, and union of both are shown in Fig. 4.14. The

ID of the readout circuit was set to ‘101101001’ and the DC bias point was set to 1 V at

the input of ADC. The dynamic range of the input of ADC was set to 1.8 V, which means

the 8 bit digital output would be ‘00000000’ for 0 V bias and ‘11111111’ for 1.8 V bias.

For 1 V DC bias point at the ADC input, after conversion, the output of the ADC was

91

Figure 4.13: Voltage response of the three different fabricated electrode in different pH
solutions.

‘10010000’, which corresponds to an analog 1.01 V. The microcontroller was programmed at

low voltage programming (LVP) mode in MPLAB X IDE for generation of ADC clock, ADC

control signal, and ID. According to the datasheet, a single bit of ADC constitutes twice

the bit period of the clock generated by the microcontroller [112]. As the ID was generated

internally from the microcontroller, the bit period of the ID was same as the ADC clock,

which is 362 µs. Additionally, the ID bit period was half of the ADC output bit period

as in Fig. 4.14. The MSB of the ID and the LSB of the ADC output are separated by

4.5 ms of idle time. The known time difference helps to resolve the sensor data accurately

from the composite bit-stream of ID and ADC output together. An ’OR’ gate was used to

combine both the ID and ADC digital outputs. Additionally, there is another idle time of 15

ms in between two digital bit stream of combined ID and sensor data. The readout circuit

harvests the required RF energy during these idle time. At this current setup, the sensor tag

sends data to the interrogator at 0.595 kbps speed including the idle time. The minimum

voltage for activation of the readout circuit is 2 V. When the bias voltage was set at 2.2 V,

92

Figure 4.14: From top to bottom: a) The ID bit ‘101101001’ b) 8 bits ADC output ‘10010000’
of analog bias 1 V, and c) combined ID and ADC output bits. The ’idle time’ in between
the bit streams is also highlighted.

the readout circuit consumed a total of 33 µW of DC power. With the current setup, the

resolution of the sensor is 6.5 bits/pH according to the slope of the linear fit (46 mV/pH) as

shown in Fig. 4.6.

4.3.5 Wireless measurement

An antenna resonating at 900 MHz from our earlier work was connected to the tag and

interrogator for wireless measurement [41]. The sensor tag was kept at a distance of 45 cm

from the interrogator, which was set at 15 dBm of transmit power at 900 MHz. The sensor

tag with antenna and pH electrode pair is shown in Fig. 4.15. The complete tag was tested

in different pH solutions and the digital bit sequence was recorded in the reader as shown in

Fig. 4.16. The ID was kept at ‘101101001’, which can be observed identical for all the pH

93

Figure 4.15: The complete sensor tag with zoomed in view of the electrode pair in the inset.
The scale is shown in inches. In the inset, the left electrode is Nafion coated Ag/AgCl and
the right one is Sb/Sb2O3 electrode.

digital bits. The digital response is ‘111010’ for pH 4.3 solution, ‘1001100’ for pH 7 solution,

and ‘1010110’ for pH 8.6 solution. After converting the bits into analog voltage, the voltage

readings become 408 mV, 534 mV, and 605 mV for respective pH solutions of 4.3, 7 and 8.6.

Using the linear fit of the data points from Fig. 4.6, the corresponding voltages are 413.6

mV, 537.8 mV, and 611.4 mV, respectively. The digitization error of 7 mV from the ADC

of the readout circuit can limit the sensor tag’s minimum sensitivity to 0.15 pH.

4.4 Application: Wearable Mouthguard

One biological application of pH electrode as an wireless salivary pH monitoring device is

demonstrated in [113]. The importance of salivary pH monitoring in order to determine any

imminent disease was already described in the introduction of this chapter. The building

block components for a salivary wearable pH mouthguard are the same as a RFID based

pH sensor, which is already described. In this section, the integration of different parts for

94

Figure 4.16: Digital bit response of the tag at different pH solutions in wireless setup. From
top to bottom: a) bit response of ‘111010’ at pH 4.3, b) bit response of ‘1001100’ at pH 7,
and c) bit response of ‘1010110’ at pH 8.6.

95

Figure 4.17: Antenna dimensions for (a) inside human mouth, and (b) in air medium.

a complete mouthguard fabrication is described. Apart from the digital circuits, and pH

sensors, the transponder mouthguard includes antenna to wirelessly receive RF power and

transmit back the sensor signals.

4.4.1 Antenna design

When the mouthguard is to be placed within human mouth, the antenna needs to be

changed due to loading of the antenna by the human body. A human head model was

considered as described in [114] to design the antenna. Due to dielectric loading, the effective

length of the antenna reduces resulting in a compact and small size antenna. The antenna

dimensions for inside human mouth and in air medium are given as in Fig. 4.17. The

simulation result for those antennas is shown in Fig. 4.18. From the reflection co-efficient,

it can be observed that the in air antenna is resonating at a fixed frequency. However, the

simulation result of the in-mouth antenna shows apart from the resonance frequency, the

reflection co-efficient is broad-band matched, due to loading effect by the human body.

96

Figure 4.18: Simulated reflection co-efficient for in-air and in-mouth antenna designs.

4.4.2 Mouthguard design

The complete transponder is shown in Fig. 4.19 from different side views in Fig. 4.19(b)

and Fig. 4.19(c) and the top view of the bare commercial mouthguard in Fig. 4.19(a). The

transponder consists of two circuit boards a) the digital part is visible at the right side and

b) the RF part is visible at the left side of mouthguard in Fig. 4.19. The dipole antenna is

placed at the bottom of the mouthguard along the two arms and the electrodes are mounted

at the center. HSMS 2852 Schottky diode was used to make a six stage RF to DC converter.

The complete sensor was integrated with pH sensor electrodes and transponder with con-

nected antenna resonating at 930 MHz. Each part of the whole sensor was integrated on a

mouthguard and the back-scattered response from the tag was measured using an Oscillo-

scope at different pH solutions. When buffer pH solution of 5.3 was used, a digital sensor

sequence of ’00010100’ was detected. Digital sensor sequence of ’00111000’ was received in

buffer pH solution of 9.5. The antenna was designed to use in air medium. The antenna

design can be further optimized for use within human mouth as proposed earlier. During

measurement, the tag was kept at 8 cm from the reader, which was transmitting 13 dBm of

97

power at 930 MHz.

Figure 4.19: The sensor tag integrated on mouthguard (a) bare mouthguard, (b) digital part
of the transponder at the right side and (c) RF circuit of the transponder at the left side of
the mouthguard.

Figure 4.20: Detected signal in Oscilloscope in buffer pH solution of 5.3.

Figure 4.21: Detected signal in Oscilloscope in buffer pH solution of 9.5.

98

Table 4.1: Performance comparison with other works.

Application

Architecture

pH Electrode

Sensitivity

Energy Source

Numerous

Coupled coil analog

Commercial

(PH105 by ExStik)

57.9 mV/pH

Not

required

Coupled coil digital

AgCl/IrOx

49.7 mV/pH

RF harvested

Food supply

chain

Biomedical

Digital

Biomedical

NFC Digital

floating-electrode

ISFET

AgCl/IrOx

48 mV/pH

On-board

battery

68.7 mV/pH

RF harvested

Biomedical

Digital

ISFET

NA

Numerous

Dual frequency analog

Commercial

glass electrode

55 mV/pH

On-board

battery

Not

required

Ref

[102]

[103]

[96]

[115]

[116]

[3]

This work

Numerous

RFID digital

AgCl/Sb2O3

46 mV/pH

RF harvested

4.5 Summary

Read range and
additional info

18 cm

@5 MHz

16 cm @1.34 MHz

and 100 Vpp

NA

3.5 cm

@13.56 MHz

Data rate
@9.6 kbps

18 cm @2 and 4 GHz

and 15 dBm

45 cm @900 MHz

and 15 dBm

data rate @.595 kbps

In Table 4.1, the proposed sensor is compared with other wireless pH sensors reported

earlier. In comparison to other works, the proposed sensor tag achieved a good read range

with moderate sensitivity and compatible with RFID technology without any requirement of

external power supply. As mentioned in Section III.C, the digital resolution of the sensor tag

is limited due to the quantization error of the ADC. The resolution can be further increased

by dynamically setting the low and high analog conversion voltage instead of setting them

at 0 V and 1.8 V respectively. For example, changing the low and high conversion voltage

from 0 V and 1.8 V to 0.4 V and 0.7 V respectively would change the resolution from 6.5

bits/pH to 39 bits/pH. However, the microcontroller has to perform additional computation

for dynamic voltage allocation, which will add up the DC power consumption. Also, varying

temperature is another challenge to obtain the correct reading from the sensor electrodes.

The voltage response changes for the same pH solution at different temperatures. However,

an additional temperature sensor can be added along with the pH sensor and the digitized

temperature information can be sent in serial along with the pH and ID information. Based

on the temperature information, the reader can correct the pH value.

The design and implementation of a batteryless and completely digital pH sensor is

99

demonstrated in this paper. The fabrication process of a thin film reference Ag/AgCl elec-

trode and a pH sensitive Sb/Sb2O3 electrode pair is presented. The characterization of the

electrode pair showed good linearity, good stability and fast response. As ID is integrated

with the sensor tag, it makes easier to identify individual sensor tags, which is difficult for

analog wireless sensors. The sensor tag harvests the required energy from the incident RF

field and hence no on-board power supply is required. The passive nature of the sensor tag

would extend the operating lifetime while reducing the footprint of the complete circuit.

The sensor tag can be further miniaturized by custom fabricating the discrete digital circuit

components into a single CMOS chip and miniaturized antenna. A read range of 45 cm

was achieved using a transmitted power of 15 dBm from the RF interrogator in the wireless

setup. The read range can be further increased by transmitting higher power or using higher

gain antenna. As the size constraint is not so stringent on the reader antenna, with different

configurations (such as Yagi-Uda) the antenna gain can be further increased by 4 to 5 dB.

100

Chapter 5

Energy Harvester Modeling

Energy harvester is a necessary component for passive RFID tags as it replaces the battery

and supplies the necessary power for seamless operation of the tag. Batteries usually become

bulky in order to achieve the required power density. Also, in structural health monitoring

(SHM) applications, sensors are integrated within structures, where probing or future access

is not possible for battery replacement. Hence, energy harvester is widely used in low power

ultra high frequency (UHF) wireless and batteryless systems such as RFID, wireless sensors

to provide required DC power for continuous operation in numerous applications, ranging

from retail to hospital supply chain and real-time sensing [117–119]. One key parameter for

energy harvester is RF to DC conversion efficiency, which provides a measure of how much

actual power is available at the RFID tag. Apart from efficiency, another key parameter

is response time. The response time can be critical for few applications with real-time

constraints, such as those involved in control loops [120]. The response of a batteryless

sensor tag depends on the time taken by the energy harvester to harvest enough voltage

and power. Influence of different circuit components on those two parameters are examined

closely in this chapter: a) efficiency, and b) response time.

101

5.1 Background

Among different energy harvester circuitss, charge pump (CP) configuration is widely

used for RFID and wireless sensors at UHF frequency. Dickson CP was originally proposed

as a DC-DC on-chip voltage multiplier. As there is no initial power available to kick-start

the clock switching circuit for very low power RF energy harvesting based applications, the

Dickson CP circuit was modified for RF energy harvesting while simultaneously boosting

the rectified voltage and now is widely used in UHF batteryless systems [119, 121, 122]. The

research on CP based energy harvester has been primarily focused on efficiency analysis and

improvement [123,124], circuit component optimization for different harvester configurations

[122,125], adaptively controlling the number of stages [126] and different material studies for

harvesting element performance [127]. In [128], an equivalent model of the original Dickson

CP was proposed, which can be used for transient analysis. Later, [129] extended the diode

based design principle as in [128] for CMOS based CP. However, the model proposed for

original Dickson CP cannot be used for modified Dickson CP due to following reasons:

• One of the primary assumptions for original Dickson CP model was that the cycle time

of the clock is sufficiently large and hence the charging time constant was negligible

compared to the cycle time. However, the assumption is valid when the input is DC

but not applicable in modified Dickson CP, where the input is RF.

• A matching network is required in between antenna and CP for maximum power trans-

fer in UHF energy harvester as shown in Fig. 5.1. Earlier proposed models on Dickson

CP did not require to include matching network in the model.

Diode based modified Dickson CP circuit is quite popular in UHF batteryless systems

due to choice of high efficiency diodes and shorter prototyping time [119, 121]. In this work,

102

(a)

(b)

Figure 5.1: (a) The block diagram of a wireless batteryless system with a multi-stage modified
Dickson charge pump; (b) Measurement setup with a fabricated 5-stage charge pump.

(a)

(b)

(c)

Figure 5.2: (a) A single stage CP, and it’s equivalent circuit for (b) negative half cycle, and
(c) positive half cycle.

an analytical model is proposed to estimate the transient response of a diode based multi-

stage modified Dickson CP taking consideration of the matching network at any arbitrary

load condition. The analysis and measurements were primarily focused on 900 MHz ISM

band due to large number of earlier reported works in this band.

5.2 Modified Dickson Charge Pump

In this part, the working principle of a single stage modified Dickson CP was first analyzed

and the key design parameters were identified from the analysis. The circuit components

and their values were later chosen based on the analysis. A single stage CP with pump

capacitance CP , load capacitance Cload, and diodes D1, D2 is shown in Fig. 5.2(a). When

103

an ac source with amplitude Vac is applied at the input, during first negative half cycle,

the diode D1 is forward biased and CP is charged through diode resistance RD as in Fig.

5.2(b). Due to charge accumulation, a voltage VCP

(t) appears across CP . In next positive

half cycle, the diode D2 will be forward biased and Cload gets charged. In this half cycle,

the total charging voltage for Cload appears as (Vac + VCP

(t)) as shown in Fig. 5.2(c). By

periodically charging in this fashion, the steady state DC voltage of a single stage CP will be

2 ∗ (Vac − VD) as maximum voltage Vac is possible for VCP

(t), where VD is forward voltage

drop of the diodes. In similar way, for a n stage CP, the steady state DC voltage will be
n ∗ (Vac − VD).

In high frequency CP design, the choice of pump capacitor CP and diode is important.

If CP is kept too small, it will not store enough charge to operate as a voltage source in

subsequent half cycle. Simultaneously, very high value of CP will not be practically useful

as the cut-off frequency of a capacitor keeps reducing for higher capacitance value. The

effect of CP on Vss is shown in Fig. 5.3 for a single stage CP. In this case, a square wave

is fed with operating frequency of 1 GHz and amplitude Vac = 0.5 V, and Cload was kept

at 200 pF. As shown in Fig. 5.3(a), for CP = 1 pF, Vss saturates at much lower voltage
than (2 ∗ Vac). With higher CP = 20 pF, Vss saturates close to (2 ∗ Vac). Hence, maximum

CP is chosen within the capacitor’s practical cut-off limit. To have the maximum voltage

multiplication from a CP, very low barrier Schottky diode is preferred due to very small VD

and fast response.

104

Figure 5.3: Effect of pump capacitor CP on steady state voltage.

(a)

(b)

Figure 5.4: (a) A Schottky diode model with pad parasitics. (b) ’L-match’ as two-element
matching network for n-stage CP.

5.3 Effect of Matching Network

After identifying the design elements, the effect of matching network on the CP is an-

alyzed next. Before using a standard matching configuration, it is important to determine

the nature of the input impedance of a single stage CP.

The circuit model of a Schottky diode with parasitic components is shown in Fig. 5.4(a).

Rj and Cj are the junction resistance and junction capacitance, Lpad, Cpad are pad induc-

tance and capacitance, and RD is the diode series resistance. The diode input impedance

ZD with the circuit components is expressed in (5.1), for sub-GHz frequency by assuming

(ωCpadRD << 1). The expression for ZD in (5.1) is further simplified in (5.2). Depending

105

on the operating condition of a diode, Ref f or Cef f can dominate in expression of ZD.

ZD = jωLpad + RD + Rj||

1

jω(Cj + Cpad)

ZD = jωLpad + Ref f +

1

jωCef f

(5.1)

(5.2)

The diode impedance ZD can now be used to compute the input impedance of the CP.

For a single stage CP, the effective input impedance ZL is combination of ZD1, ZD2, and CP .

At sub-GHz frequency, the diode and pad capacitance dominates over the pad inductance.

Hence, ZD is resistive and capacitive in nature and ZL will be of similar nature as expressed

in (5.3).

(cid:18)
+ ZD1||

ZL =

1

jωCP

ZD2 +

1

jωCload

(cid:19)

(5.3)

The extracted input impedance of a single stage diode pump with CP = 20 pF, Cload =
200 pF, and HSMS-2852 diode was (3.8 − 105j) Ω at 900 MHz and -10 dBm input power.

The capacitive nature of the CP input reactance is preserved even when multiple stages are

used, as shown later.

With the knowledge of ZL, the matching networks for single and multi-stage CP are

designed. In general, two-element based impedance matching is widely used for single fre-
quency matching. Considering the CP input impedance as ZL = RL − jXL, two-element

based ‘L-match’ as shown in Fig. 5.4(b) is used for (RL < 50). After matching, the input

impedance Zin of the combined system should be close to 50 Ω for impedance matching to

a 50 Ω antenna.

If Vin is considered as the peak amplitude of the time varying induced

106

(a)

(b)

Figure 5.5: A typical matching trajectory (dotted line) of single stage CP. For case (a), the
matching network uses series inductor and shunt inductor (X1 > 0, X2 > 0). For case (b),
the matching network uses series inductor and shunt capacitor (X1 > 0, X2 < 0).

voltage across the antenna, and VL as the peak voltage induced at the input of the CP as in

Fig. 5.4(b), then Vin can be calculated as(cid:112)2 ∗ (Pin ∗ (1 − Γ2) ∗ |Zin|), where Γ is reflection

co-efficient and Pin is antenna received power. For perfect matching between antenna and

the CP, Γ is 0. VL is important to compute, as it decides the final output of the CP. The

voltages VL and Vin are related according to (5.4).

(cid:12)(cid:12)(cid:12)(cid:12)

(cid:12)(cid:12)(cid:12)(cid:12)

(5.4)

VinP
VLP

=

RL − jXL

RL − jXL + R1 + jX1

5.3.1 Single-stage formulation

Based on the single-stage input impedance, the two possible impedance matching trajec-

tory are shown in Fig. 5.5 by applying the proposed matching network. During matching,

(R1 + jX1) is applied in both cases to relocate ZL on the 50 Ω admittance circle. Next,
(R2 ± jX2) is applied to finally relocate ZL at 50 Ω impedance point. R1 and R2 represent

the resistive loss of the series and shunt matching elements respectively.

107

(a)

(b)

Figure 5.6: (a) n stage charge pump, and (b) it’s simplified impedance model.

With application of series (R1 + jX1), the admittance of combined ZL and (R1 + jX1)
will be 1/((RL + R1)− j(XL− X1)). By imposing the constant unity normalized admittance

condition at the impedance tracing for both the cases as shown in Fig. 5.5(a) and Fig.

5.5(b), the following relation as (5.5) is obtained. In (5.5), Rnorm represents the reference

resistance used for normalization, which is 50 Ω. The steady state DC voltage Vss1 for a

single-stage CP can be expressed in terms Vin, diode voltage drop (VD), CP impedance
(RL − jXL) as in (5.6) by substituting (5.5) in (5.4), which can be subsequently simplified
for (XL >> RL), and zero bias diode (VD ≈ 0).

(RL + R1)/((RL + R1)2 + (XL − X1)2) = 1/Rnorm

Vss1 = 2VL − VD
2Vin

≈

(cid:114)

(cid:19)

(cid:18) XL√

RL

(cid:1)

R1
RL

(cid:0)1 +

Rnorm

(5.5)

(5.6)

108

5.3.2 Multi-stage formulation

A simplified formulation of steady state DC voltage Vssn for n-stage CP can be obtained

with further approximations. By assuming pump capacitance CP is large enough to be

considered as RF short at 900 MHz, which will simplify the total input impedance as ZL/n,

as shown in Fig. 5.6. By using the equivalent impedance model, Vssn can be obtained as

expressed in (5.7), using approximate form of (5.6), provided the same Rnorm was used while

matching. In (5.7), R1n denotes the loss resistance of the series matching element.

(cid:114)

Vssn ≈

Rnorm

2n ∗ Vin

(cid:0)1 +

(cid:1)

nR1n
RL

(cid:19)

(cid:18) XL/n(cid:112)RL/n

≈ √

nVss1

(5.7)

The input impedance of the n-stage CP reduces as (RL/n − jXL/n). As a result, the

series matching element (R1n + jX1n) needs to be reduced. If series inductors of identical Q-
factor are used, a further assumption can be made as nR1n ≈ R1 to maintain the matching
condition. However, this assumption is valid as long as (X1n − XL/n) remains almost
identical to (X1 − XL) during matching. With the approximations, a further simplistic form

of Vssn for n-stage is expressed in terms of only Vss1 and n.

5.4 Transient Model for Arbitrary Load

The 10%-90% rise time (Trn) is obtained from transient response of a n-stage CP system.

The capacitive load Cload gets charged primarily through the diode video resistance, referred
as pump resistance RP = n0kT /(q ∗ (Is + Ib)), where n0 is ideality factor, k is Boltzmann

constant, Is and Ib are reverse saturation and bias current respectively [130]. With RP ,

Cload, and Vss1, the equivalent model of a single-stage CP is shown in Fig. 5.7(a). RP of a

109

(a)

(b)

Figure 5.7: Equivalent transient model for (a) single stage CP, and (b) n stage CP.

(a)

(b)

Figure 5.8: (a) Extracted pump resistance (RP ) of a single stage CP at different input power.
(b) Model vs. discrete circuit based single stage CP simulation time comparison.

single-stage CP is extracted at different input power level as shown in Fig. 5.8(a). Due to

self-biasing of a Schottky diode, bias current increases at higher input power and hence RP

decreases.

Using equivalent circuit from a single-stage CP, the equivalent model of a n-stage CP

is extended as shown in Fig. 5.7(b). A further simplification was performed under the

condition Cload >> nCP as Cload is usually in nF range or higher whereas CP is in pF

range.

In addition to Cload, when a load resistance Rload is also present, Rload forms a

voltage divider with internal nRP .

In (5.8), vn(t) of a n-stage CP is expressed in terms

of Vssn, RP at any load condition. Finally, Trn is calculated from time dependent vn(t)

expression.

It should be noted that increase in stage number will reduce the input RF

110

impedance whereas increase the output DC impedance.

vn(t) = Vssn

1
n∗RP
Rload

1 +

(cid:18)

−
1 − e

(n∗RP ||Rload)∗Cload

t

(cid:19)

(5.8)

5.5 Measurement Results

Use of model shows a substantial reduction in actual time taken by a simulator compared

to discrete component based transient circuit simulation with RF input as in Fig. 5.8(b).

For example, a single stage CP with the discrete diodes, CP , and matching network takes

actual time of 1059 s for transient simulation duration of 2 ms. This happens due to a very

small timestep reqirement to resolve the full bandwidth of the RF input for convergence. On

the other hand, as Fig. 5.7(a) based single CP model uses a DC step input, the simulation

time reduces to 1.6 s with same convergence criteria. For comparison, both the transient

simulations were performed using Keysight ADS at MSU DECS server (2 Intel Xeon Gold

6150 CPU and 50 GB RAM).

One to five stage CP was assembled on a FR-4 PCB with HSMS-2852 diode and CP = 20

pF to verify the developed model with setup as shown in Fig. 5.1(b). A single stage CP
shows good matching (S11 ≤ −10 dB) around 900 MHz at diiferent input power as shown

in Fig. 5.9(a). Also, CPs with different stages show good matching around 900 MHz at -10

dBm input power as in Fig. 5.9(b).

The transient model was further compared with measurements for different stages as in

Fig. 5.10, and different Cload and Rload as in Fig. 5.11(a). The voltage normalization in

Fig. 5.10 and 5.11(a) was performed with respect to Vssn at no load resistance. Additionally,

measured efficiency as shown in 5.11(b), demonstrates the maximum efficiency when Rload

111

(a)

(b)

Figure 5.9: Measurement results of reflection co-efficient (S11) (a) single stage CP at different
input power, and (b) multiple stages CP at -10 dBm input power.

Figure 5.10: Voltage normalized transient response of CPs with different stages with Cload =
100 nF and no external Rload.

112

(a)

(b)

(c)

Figure 5.11: (a) Voltage normalized transient response of a 2-stage CP at different load
conditions. (b) Measured efficiency of CPs with different stages. (c) Scaling factor of Trn
and Vssn compared to a single stage CP.

Table 5.1: Comparison with relevant literature.

Ref.

[122]
[125]
[129]
[128]
[131]
[132]

[This work]

900 MHz

Frequency

Application

Technology

No. of stages

Impedance
matching
in model

Transient
analysis

using model

950 MHz
915 MHz

DC
DC

5.8 GHz

2.45 GHz, 5.8 GHz

RFID energy harvesting
RFID energy harvesting

DC-DC voltage boost
DC-DC voltage boost

180 nm CMOS
90 nm CMOS
65 nm CMOS
Diode based

RF-DC energy harvesting
RF-DC energy harvesting
RF-DC Energy harvesting

BAT15-03W diode
MA4E1317 diode
HSMS-2850 diode

2-3 (measured)
17 (measured)

8-16 (simulated)
8-16 (simulated)

Single diode
Single diode

1-5 (measured)

Yes
No
No
No
No
No
Yes

No
No
Yes
Yes
No ∗
No ∗
Yes

*Modeled in steady state time domain

is close to nRP for respective CP. As the total loss associated with diode resistance increases

with n, peak efficiency drops with increase in stage number. All the measurements were

performed at -10 dBm input power and within (926 MHz - 935 MHz) band depending on

the CP’s best S11 response.

Finally, the theoretical and measured scaling factor for Trn/Tr1 and Vssn/Vss1 are com-

pared in Fig. 5.11(c) for different number of stages at Cload = 100 nF and no external Rload.

For a single stage CP, Vss1 was measured as 1.14 V and Tr1 as 1.8 ms at 926 MHz and

-10 dBm input power. As derived earlier, (Trn/Tr1) varies proportionate to n according to

(5.8), while (Vssn/Vss1) varies proportionate to

√
n according to (5.7). However, it should

be noted that the scaling factor for different n is valid when the CPs with similar matching

113

configuration are measured at identical power level.

5.6 Summary

In this chapter, the modified Dickson CP circuit is analyzed for different design parame-

ters and a model is proposed to capture the transient response. The model takes account of

impedance matching network, which is necessary to match a CP to an antenna in wireless

systems. Additionally, the model offers a great reduction in simulation time especially when

the CP rise time starts exceeding 1 ms. The model supported by measurement result showed

that the rise time increases proportionally with increasing the number of stages, which is also

necessary to boost the steady state output voltage. The model will be helpful in estimating

charging rise time for wireless energy harvesting applications.

114

Chapter 6

Effective Transmit Time Reduction of

RFID Reader

Prolonged RF radiation reduction on human body can be detrimental and there has been

multiple related studies, which were mentioned in chapter 2. RFID has become a successful

choice for integrated sensing due to small feature size and batteryless architecture. However,

RFID tags require continuous RF illumination from the reader to transfer data from the tag.

This scenario may trigger for undesired prolonged transmission from the RFID reader. To

mitigate the problem, a dual mode RFID tag with active transmission is proposed in this

chapter, where the RFID tag will fetch information from the sensor locally. In this chapter,

a dual mode RFID system is proposed for intra-vehicle wireless sensing application, where

there is a risk of prolonged RF radiation exposure to human bodies.

6.1 Background

Sleep monitoring of patients during acute illness or clinical condition is very important.

The hybrid RFID based wireless and batteryless sensing system can be used to monitor a

patient’s movement during sleep. In this case, a pressure monitoring sensor can be integrated

with the hybrid RFID system. Also, with an increase in intelligent transport system, more

sensors need to be integrated within a transport system for various applications. Currently,

115

Figure 6.1: Dual mode RFID tags for patients’ sleep monitoring system.

majority of the sensors are interfaced with the transport control system using wires, which

results in long redundant wires routed all over a transport system’s interior. This leads

to added cost, weight, and wiring complexity. The size of the wire harness can be signif-

icantly reduced if the connectivity of the sensors is made wireless. Hence, there has been

a drive recently to make the sensor connectivity wireless [133–135]. However, with wireless

connectivity of the sensors to the transport control system, power supply becomes an issue

for those sensors [136]. Usually battery is used for supplying power to the wireless sensors.

However, using batteries will limit the life-time of the sensors based on power consumption.

Additionally, many of those sensors are embedded within the body or parts, which may not

be accessible to change batteries. Hence, a wireless and batteryless system is required for

the sensors to be installed within a modern transport system to remove many redundant

wire connections and will be operational over the life-time of the sensors without requiring

service such as battery change.

RFID based sensors have been proposed earlier for different low power sensing applica-

116

Figure 6.2: Different modes of communication in between the reader and tag.

tions [46, 118, 137]. The central controller within the tag periodically activates the sensor

and transmits the data using backscatter modulation. The harvested power in the RFID is

used for powering up the tag circuit and the sensor. Use of the controller with ADC en-

ables the RFID based sensor tag to interrogate a broad range of analog and digital sensors.

In [118,137], analog electrochemical sensors pH and ammonia were used. The analog output

voltage was converted into digital data and sent from the tag to the interrogator. A similar

technique can be used for patients’ sleep monitoring in clinical conditions. In Fig. 6.1, an

example scenario of the dual mode wireless sensors is shown, where rooftop reader antenna

is used to monitor the embedded sensors. The sensors for sleep monitoring are usually piezo-

electric or piezoresistive in nature, which generates voltage or changes impedance in case of

applied pressure is changed. By detecting the pressure, the movements of a patient is quan-

tified, which will provide the data for quality sleeping. Using the same concept, multiple

sensing modalities can also be incorporated into a single tag.

117

6.2 Finite State Machine and Event Driven Modes

The design of the system can be conceptualized in terms of a finite state machine (FSM).

FSM is a mathematical model used to develop a logical process where a finite number of

operational conditions or states exist and at a time the machine can only be at a single state.

Transition from one stage to another occurs based on single or multiple events. The FSM for

the tag and reader communication is represented using the variables (Σ, M, M0, δ), where:

• Σ is the set of finite input events which initiate the state transitions.

• M is the finite set of states as {M0, M1}, where in M0 the tags communicate with the

reader using backscattering, and in M1 the reader waits for the tag transmitted status

change signal.

• M0 is the initial state.

• δ is defined as the state transition function, where δ : Σ × M → M .

6.2.1 M0 state: Tag backscatter communication

In the initial state M0, the tags transfer the sensor status using backscatter communi-

cation. In backscatter communication, the tag modulates its antenna impedance to transfer

data to the reader. In this state, the reader supplies the RF carrier and the tags use the

carrier to perform the modulation as well as storing the redundant energy. As shown in Fig.
6.2, the reader sends queries sequentially as {Q1, Q2, ..., QN} for N tags and receives the
consecutive replies {R1, R2, ..., RN}. The sequential and individual query of tag helps in

avoiding tag reply message collision at the reader. The state M0 stays in the same state as

118

long as event driven mode m00 prevails. When the mode m01 is triggered, the state changes

from M0 to M1.

6.2.2 M1 state: Tag active transmit

In state M1, the reader does not transmit any RF carrier. Instead of the reader directly

monitoring the tag sensor status as in M0 state, the reader delegates the sensor monitoring

to the tag locally in M1 state. A tag uses the stored energy from M0 state for sensor

monitoring. In case of sensor status change of ith tag Ti, the tag dispatches alarm signal. In

this state, the reader waits to listen for any alarm. M1 state is maintained as long as m11

mode is active until when m10 mode is trigerred and the state changes to M0.

6.2.3 Events and driven modes

The state transition takes place due to the driven modes and each of the modes is
dependent of single or multiple events. The event set Σ : {e1, e2, e3} is triggered due to

voltage level VTi

or alarm signal status ATi

of Tag Ti. The event signal e1 is set as 0 as long

as ATi

status is inactive for all the tags. As soon as ATi

is triggered from atleast one tag,

e1 status will change to 1.

e1 =

1 [∃ Ti | (ATi

== 1)]

== 0)∀ Ti]

0 [(ATi
where, Ti ∈ {T1, T2, ..., TN}

(6.1)

The input event e2 is set as 0 as long as VTi

of atleast one tag is below the predicted

voltage V P eak

ch

by the reader while the tags are charging. As soon as VTi

of all the tags are

119

higher or equal to maximum voltage V P eak

ch

, e2 status is changed to 1. Similarly, e3 is set

at 0 as long as VTi

of all tags is above the minimum voltage V P eak

dis

by the reader while the

of atleast one tag drops below V P eak

dis

, e3 is set as 1. The

tags are discharging. When VTi
parameters V P eak

and V P eak

ch

dis

< V P eak

ch

)]

≥ V P eak

)∀ Ti]

ch

1 [(VTi
where, Ti ∈ {T1, T2, ..., TN}

are described later.

0 [∃ Ti | (VTi
1 [∃ Ti | (VTi

e2 =

e3 =

< V P eak

dis

)]

≥ V P eak

)∀ Ti]

dis

0 [(VTi
where, Ti ∈ {T1, T2, ..., TN}

(6.2)

(6.3)

Different modes m00, m01, m10, m11 are executed based on the event combinations as

expressed in (6.4). The modes are primarily triggered by e2 and e3 events. It should be

noted that e2 and e3 are two mutually exclusive events, where e2 can only occur in M0 state

and e3 can only occur in M1 state. Generally speaking, the machine remains in M0 state

until all the tags are fully charged. The machine then stays in M0 state until atleast one

of the tags is discharged or ATi

is dispatched. In the instance of two tags dispatching ATi
simultaneously where (i (cid:54)= j), the reader will be unable to decipher as there will

and ATj

be information collision. To avoid this, the machine changes state to M0 where the reader

inquire each tag individually and sequentially.

120

Figure 6.3: Different sub-modules of the tag circuit.

m00 = 1, if(cid:2)(e2 == 0)(cid:3)
m01 = 1, if(cid:2)(e2 == 1)&&(e1 == 0)(cid:3)
m10 = 1, if(cid:2)(e3 == 1)||(e1 == 1)(cid:3)
m11 = 1, if(cid:2)(e3 == 0)(cid:3)

(6.4)

6.3 System Architecture

As described earlier, the system includes multiple tags and a reader transceiver to obtain

information from the tag both in backscatter and active-transmit mode.

In this section,

different parts of the tag circuit is described first, then the reader part is described.

6.3.1 Tag design

The complete tag circuit with the sub-modules is shown in Fig. 6.3. The tag contains:

1) Energy Harvester to harvest DC power from the RF power; 2) Reader Code Demodulator

to decode specific instruction from the reader and act accordingly; 3) Oscillator to generate

carrier frequency at 900 MHz; 4) Charge Control Module to store redundant charge and act

121

Figure 6.4: Energy harvester with 3-stage Dickson charge pump configuration.

as a power supply to the tag when the reader is not transmitting; 5) Micro Controller to

perform the task activities; and finally 6) Sensor to monitor the intended parameter. Each

of the sub-modules are described in detail as followed:

6.3.1.1 Energy harvester

The energy harvester converts the RF energy into DC for biasing, in order to supply DC

bias to drive the peripheral circuits. Dickson charge pump configuration was used as the

energy harvester as it converts DC voltage, and simultaneously boosts up the DC voltage at

desired level [118]. From our previous analysis in Chapter 5, it was shown that more number

of diodes help in boosting the DC voltage but simultaneously degrading the efficiency and

charging rise time. Six diode based charge pump configuration was found to be optimum

for the energy harvester. Very low barrier Schottky diode HSMS 2852 was used to harvest

energy due to good efficiency. The generated DC voltage is regulated using a low power

voltage regulator (TPS79733) at 3.3 V to drive the peripheral circuit. The schematic of the

energy harvesting circuit is shown in Fig. 6.4, which takes EH in as input RF and provides

three output: 1) DC 3.3 V, and 2) RCD in1, 3) RCD in2 for the reader code demodulator.

122

Figure 6.5: Reader code demodulator to decode reader instruction at the tag.

6.3.1.2 Reader code demodulator

The reader code demodulator decodes specific instruction from the reader and generates

interrupt. To avoid tag reply collision, it is necessary that a single tag will reply at a time. In

conventional RFID system, the tag reply collision is avoided by activating random counter in

the tag. Randomizing the tag reply in this way minimizes the tag collision but do not fully

mitigate it. By activating one tag at a time based, the system can avoid potential tag reply

collision. Two low pass filters (LPF) are used in between the energy harvester and reader

code demodulator to pass the low frequency reader demodulation signal and stop the high

frequency RF signal as shown in Fig. 6.5. The schematic of the reader code demodulator

is shown in Fig. 6.5. RCD in2 input provides reference level and RCD in1 carries the

low frequency modulation. Both of the inputs are compared and interrupt IN T is issued

whenever RCD in1 is lower than RCD in2. In addition to providing IN T , the reader code

demodulator sends the reader instruction signal through the same channel.

6.3.1.3 Oscillator

A standard oscillator in cross-couple configuration is used to generate the 900 MHz carrier

as shown in Fig. 6.6. Two resistors of 10 kΩ are used to set the output voltage swing of

123

Figure 6.6: Oscillator circuit to generate 900 MHz RF carrier.

the RF output signal RF out at low power consumption. LC tank was realized at 900

MHz using lumped 5.6 nH inductor and 2.4 pF of capacitance. BFT25A BJT with 5 GHz

cut-off frequency was used to provide sufficient negative resistance for the tank to sustain

oscillation. A common emitter buffer stage was used to drive the 50 Ω antenna system using

the oscillator output. The base of the BJT in common emitter configuration is biased using a

RF choke of 4.7 uH. Two 12 pF capacitors were used to isolate the DC biasing of cross-couple

and common emitter configurations. The oscillator is triggered by micro-controller port

signal, whose voltage level varies depending on the supply level of the micro-controller. The

oscillator activation signal OSC in was fed into a low power voltage regulator (TPS79718)

to keep the voltage level fixed at 1.8 V and hence, the power consumption due to oscillator

was fixed irrespective of the micro-controller supply voltage.

6.3.1.4 Charge control module

The charge control module uses a 2.2 mF capacitor to store energy in the tag locally and

use that energy to power the tag circuit when energy is unavailable from the energy harvester

as shown in Fig. 6.7(a). The charge control module uses two switches: a) current controlled

switch, and b) bi-directional switch to charge the capacitor. At the initial stage of charging,

124

(a)

(b)

Figure 6.7: Charge control module with (a) switches and a 2.2 mF storage capacitor, and
(b) voltage comparator.

the capacitor acts as a short circuit and hence charges rapidly resulting in dropping the DC

voltage level of the energy harvester. When the voltage level drops from required level, the

tag stops operating as desired. Hence, the charge flow in capacitor is controlled using a

current controlled switch at the initial stage of charging. Once a threshold minimum voltage

is reached at which the tag can perform, the charging is done through a bi-directional switch.

As the charge flow is controlled in current controlled switch, it is slower, whereas charging

through bi-directional switch is very fast. Control pins CCSW and BDSW are used to

control the two switches. A very low barrier ( 0.1 V) and ultra low reverse leakage Schottky

diode was used in between the energy harvester and rest of the peripheral to minimize the

leakage power in reader code demodulator and energy harvester when the tag is operating

on stored capacitor charge. M C V DD is used to supply bias to the micro-controller and

C V DD denotes the voltage level of the capacitor.

A simple zener diode based reference is used for voltage comparator as shown in Fig.

6.7(b). The zener diode provides 2 V reference for activating CCSW from BDSW switch.

To restrict power consumption, the voltage comparator is activated using V C act. C V DD

voltage level is compared with 2 V voltage level, and the comparator output is provided

using V C out as digital signal.

125

(a)

(b)

Figure 6.8: (a) Input and output ports connections of micro controller, and (b) sensor circuit
configuration.

6.3.1.5 Micro controller

PIC16LF1823 with operating voltage at 1.8 V was used as the micro controller to perform

the logical operations. The internal clock frequency was set at very low frequency of 31 kHz

for low power consumption. The controller uses IN T as external interrupt, which triggers

different operational mode other than its usual operational mode. Different input and output

ports of the controller are summarized in Fig. 6.8(a).

An example sensor was used as a test case. A medium power consuming sensor was

used to show the efficacy of the proposed RFID system. A hall effect based sensor is used

as an example sensor. The sensor with biasing components is shown in Fig. 6.8(b), where

the sensor generates output depending on presence or absence of magnetic field. The sensor

circuit is activated using SEN act port and the sensor status is read from SEN out port.

6.3.1.6 Antenna and front end

The tag requires an antenna and a RF front end consisting of RF switches to receive

RF power and send oscillator trigger signals efficiently in M0 and M1 states respectively. A

standard planar inverted-F antenna (PIFA) was designed to miniaturize the antenna struc-

ture and also to reduce human loading effect. Two different boards were used as shown in

126

Figure 6.9: Antenna dimensions with front and side views.

Figure 6.10: Simulated and measured PIFA antenna reflection co-efficient comparison.

front and side views in Fig. 6.9, where the PIFA is in top board and the bottom board is

only back metal reflector. The back reflector was used to reduce the effect of any material

from the backside of the antenna. The simulation and measurement results for reflection

co-efficient (S11) comparison of PIFA antenna is shown in Fig. 6.10.

6.3.2 Reader design

The reader circuit is shown in Fig. 6.11 with the RF front end, and the controller part.

The continuous frequency at 900 MHz generated by the source is modulated or controlled

using a RF switch. A splitter is used to bifurcate the RF signal in two parts: a) one part

127

Figure 6.11: The reader to interrogate the tags in two modes.

is amplified using power amplifier (PA) and transmitted using an antenna; b) the other

part is used in the mixer to downconvert the tag replied signal in M0 state or backscatter

communication mode. The circulator separates the transmit and received RF signal in the

reader. After circulator, the tag return signal is fed into a SPDT switch, which directs the

signal in two alternative paths. In M0 state the signal is fed into the mixer, otherwise fed into

a detector diode after passing through a bandpass filter at 900 MHz in tag active transmit

mode or M1 state. As the tag return signal is weak in both the modes, a baseband amplifier

(BA) or a low noise amplifier (LNA) is used to amplify the tag signal. The two modes are

controlled by interchanging the RF and SPDT switch. The RF switch is also used to send
reader code targeted for specific tag reply. A high gain antenna (∼8 dBi) was used at the

reader side to communicate with the tags.

128

Figure 6.12: Charging of the capacitor using controlled current.

6.4 Charging and Discharging Operation

The tag has primarily two operational states: 1) backscatter communication, and 2) ac-

tive transmission, as mentioned earlier. From charge holding point of view, the tag capacitor

gets charged during the M0 state and discharged during M1 state. Now, for the reader to

initiate and stop RF transmission, it is important to model the charging and discharging

time at the tag.

6.4.1 Charging time

In the backscatter mode, the objective is to get the storage capacitor charged up. The

charge control module uses two stages while charging the capacitor: a) controlled current

charging, and b) bi-directional charging. To continue the tag microcontroller functional

during charging of the storage capacitor, during the initial stage of charging, controlled

current charging was used until the threshold voltage level VT H is reached.

6.4.1.1 Controlled current charging

The equivalent circuit of the current controlled charging is shown as in Fig. 6.12, where

the transistor emitter current is modeled as a voltage controlled current source Ich. The

energy harvester is modeled as equivalent source with VP mp and pump resistance RP as

129

derived in Chapter 5. The impedance of micro controller is represented as an equivalent load

resistance of RL from its voltage current relationship. The micro controller and capacitor

voltage level is shown as VM C and Vcap. The transistor current Ich is controlled by the voltage

level of the switch signal CCSW , maintained by the micro controller and is denoted as VCN ,

which can have maximum value of VM C . In between energy harvester and the current control

circuit, the voltage regulator and diode will regulate the VM C level, and the maximum level
of VM C is defined as Vreg − VD, where Vreg is the regulated output and VD is forward diode

voltage drop. By imposing KCL on the circuit as shown in Fig. 6.12, expression for VM C

level is obtained as in (6.5). It is important to note that the micro controller requires atleast
Vlvl supply voltage to remain operational. By using the bound (VM C ≥ Vlvl), the maximum
bound on Ich is expressed as (Ich ≤ VP mp/RP − Vlvl/(RP||RL)).

(cid:18)
(RP||RL)

(cid:18) VP mp

(cid:19)
(cid:19)
, (Vreg − VD)

(6.5)

VM C = min

− Ich

RP

When VCN is applied at the transistor base, the emitter current is obtained as in (6.6),

where β is the forward current gain, RB is base resistance, and VBE,on is base emitter on-
voltage. VCN can attain maximum value of VM C , and hence α ≤ 1. Higher value of α is

desired to maximize Ich and hence shorter charging time for the storage capacitor. However,

increasing Ich will risk at dropping the VM C level bellow Vlvl and hence, shutting down the

whole system. Hence, Ich should be optimally controlled by using the VM C level during the

initial stage. At the later stage of charging, once the capacitor starts accumulating charge,

the voltage Vcap starts increasing and Ich starts dropping. As a remedy, the value of α

should be kept lower during initial charging stage and higher during later charging stage.

Value of α can be controlled using micro-controller digital to analog converter (DAC). The

130

time dependent relation between Ich and Vcap is expressed as in (6.7).

Ich =

β + 1
RB

(VCN − VBE,on − Vcap)

where, VCN = αVM C

Ich = C

dVcap

dt

(6.6)

(6.7)

Using VP mp and RP values at different power level, atleast -1 dBm input power is required

to keep Vlvl = 2 V for parameters at α = 1, RL = 160 kΩ, VBE,on = 0.7 V, RB = 1.2 MΩ,

β = 80, and C = 2.2 mF. At -1 dBm input power, VP mp = 4.7 V and RP = 18 kΩ are

obtained for a three stage energy harvester from Chapter 5 analysis. In order to reduce the

operating power of the tag while maintaining Vlvl = 2 V, α should be less than 1 while

keeping other parameters to be same. For example, when input power is -3 dBm, VP mp

and RP result as 3.8 V and 20 kΩ, and by solving (6.5) and (6.6) with initial condition as

Vcap = 0 V, α = 0.86 is required as the current controlled charging kicks in. α can only

become 1 once Vcap reaches 0.26 V, which results in controlled current charging for atleast

9.2 s with α = 0.86. The measured result of both conditions at -1 dBm and -3 dBm input

power is shown in Fig. 6.13. When α = 1 starting from the begining for -3 dBm input

power, the VM C level falls below Vlvl during the initial 8 s.

6.4.1.2 Bi-directional charging

Once, Vcap reaches the VT H level, the charging circuit switches on the bi-directional

switch and turns off the current controlled switch for fast charging. The Vcap can be obtained

by equating the power balance expression as in (6.8), where η is power conversion efficiency

from RF to DC, and PRF is input RF power. By imposing initial condition, time varying

131

Figure 6.13: VM C and Vcap voltage level for 3 cases:
input power is -1 dBm and α = 1,
input power is -3 dBm and α = 1, input power is -3 dBm and α = 0.86 for initial 9.2 s and
later α = 1.

expression for Vcap is obtained as in (6.9).

ηPRF = CVcap

dVcap

dt

(cid:114)

2ηPRF

C

t

Vcap(t) = VT H +

(6.8)

(6.9)

Now with understanding of the storage capacitor charging behavior, the charging voltage

function V P
ch(τch, VT H ) can be defined as the voltage level of the storage capacitor with peak
value of (Vreg−VD), provided it reaches VT H after a time τch. Now, as the reader constantly

monitors the voltage level of a tag, τch will be known to the reader for a specific tag and
the time required to reach from VT H to (Vreg − VD) can be calculated based on (6.9). Then

the total charging time Tch for a specific tag will be as given in (6.10). Once the reader

transmits RF power for Tch duration, the reader sends end-of-transmit (EOT) signal to all

the tags. Upon receiving EOT, all the tags enter into active transmit mode.

132

Tch = τch +

C(Vreg − VD − VT H )2

2ηPRF

(6.10)

6.4.2 Discharging time

After all the tags are charged up, the reader stops transmitting and the storage capacitor

starts slow discharging. To maximize the discharge time, two primary objectives are followed:

a) leakage current within the tag components is as minimum as possible, and b) after knowing

the sensor status, the micro controller quickly enters into sleep mode to conserve power.

There are primarily four leakage current paths from the storage capacitor: 1) bi-directional

switch internal leakage Isw, 2) transistor emitter-base leakage Itran, 3) voltage comparator

positive input internal leakage Ivc, and 4) diode reverse leakage Id. The micro-controller

currents in sleep mode is considered as Isl and in active mode as Iac. Now with sensor current

consumption as Isen and the fractional active time as γ, the total current consumption Itot

can be expressed as (6.11). The discharging time τdis for a specific tag is given as in (6.12) for
the voltage to drop from (Vreg − VD) to VT H and the function is defined as V P
dis(τdis, VDD),
where VDD = Vreg − VD. The current consumption of the individual branches is provided
in Table 6.1. V P

dis(τdis, VDD) takes either τdis or VDD as input and provides the other

parameter as output.

Itot = Isw + Itran + Ivc + Id+γ ∗ (Iac + Isen)
+ (1 − γ) ∗ Isl

τdis =

C ∗ (Vreg − VD − VT H )

Itot

(6.11)

(6.12)

133

Table 6.1: Current consumption branches.

Current branch

Value

Isw
Itran
Ivc
Id
Iac
Isl
Isen
Iosc

50 nA @ 3.2 V
40 nA @ 3.2 V
40 nA @ 3.2 V
60 nA @ 3.2 V
10.2 uA @ 3.2 V

1 uA @ 3.2 V
10 uA @ 3.2 V
400 uA @ 1.8 V

Although the current consumption in active mode is much higher compared to the sleep

mode current consumption, Itot can be reduced by keeping γ much lower. By employing the

condition that the tag inquires the sensor once in every 250 ms, γ = 0.002 can be achieved.

When the oscillator is activated upon the sensor change event, the discharging of the stor-

age capacitor is fast and the active mode discharging function is given as V P

disosc(Vf , Vinit),

which takes Vinit as initial voltage level of the capacitor before the oscillator starts and Vf

as final voltage level after the tag stops transmitting. Vf is calculated based on the current

consumption by the oscillator as Iosc, oscillator on time ∆t, and Vinit as in (6.13). The on

time ∆t is fixed at 1.5 ms, which is sufficient for the reader to detect tag transmitted RF

signal.

Vf = Vinit − Iosc ∗ ∆t

C

(6.13)

6.5 Measurement Results

Once, the working principle of the dual mode RFID is decided, both wired and wireless

measurements were performed for validation. The complete fabricated wireless tag circuit

and the fabricated PIFA antenna are shown in Fig. 6.14. The PIFA antenna has SMA outlet

134

(a)

(b)

Figure 6.14: (a) Dual mode RFID tag circuit with SMA, and (b) front and back side of the
PIFA antenna with SMA outlet, compatible to be embedded within seat.

from the backside, which was connected with the SMA outlet of the tag circuit using a SMA

connector during wireless measurement.

The measured results for storage capacitor and micro-controller voltage level are shown

in Fig. 6.15, due to charging switch control signal change. As mentioned earlier, there are

two modes of charging: (a) controlled current charging, controlled by CCSW , and (b) bi-

directional switch charging, controlled by BDSW . When the CCSW is active and BDSW

is inactive, the storage capacitor charging is slower, which ramps up when the BDSW switch

becomes active as shown in Fig. 6.15. Once the BDSW switch is active, the micro-controller

and storage capacitor voltage levels get shorted and hence, the M C V DD drops to same

level as C V DD.

To avoid tag collision, RCD module was integrated with the tag, which activates each

and every tag sequentially based on the code sent from the reader. The scenario is shown

pictorially in Fig. 6.16, where the tag only replies when it receives a specific code sequence.

The RCD module is activated only when the M C V DD reaches the Vlvl voltage, which

depends upon the energy harvester time to harvest required voltage.

When the tag is measured in wireless setup, the information received at the reader from

a dual mode tag is shown in Fig. 6.17. The tag returned raw signal is shown at the top, and

135

Figure 6.15: Storage capacitor and micro-controller voltage level at different switching sig-
nals.

Figure 6.16: Activation of a dual mode tag due to specific code sent from the reader.

136

Figure 6.17: The information received at the reader sent by a dual mode RFID tag.

the processed signal is shown at the bottom. In the received bit-stream, the 32 bit ID, 1 bit

sensor information, and 1 bit voltage status information are sequentially sent. The voltage

status signal goes from ’0’ to ’1’ when the C V DD level crosses the VT H voltage level. The

sensor status shows the presence or absence of a magnet as bit ’0’ and ’1’ respectively.

Finally, the sensor monitoring during the active transmit state was measured. The dual

mode tag takes 2.5 minutes to charge up completely during initial stage and only 0.5 minutes

in successive charging events. While discharging, if there is no sensor status change, the tag

can monitor the sensor locally for 30 minutes. Each tag interrogates the sensor locally once

in every 250 us. Hence, at worst case scenario, the reader knows of a sensor event change

within 1 second.

6.6 Summary

In this work, the working principle of a dual mode RFID sensor tag is described and

experimentally validated. The RFID tag monitors the sensor locally and in case of an event

trigger, the tag notifies the the reader. Any analog or digital sensor can be integrated with

the proposed dual mode RFID system. The sensor integrated dual mode tag showed a

137

significant reduction in transmit time from the reader. The transmit time reduction strongly

depends upon how many times the tag sends the alert signal to the reader. This scenario is

beneficial for a sensor, which doesn’t change its state frequently. The proposed dual mode

RFID tag has strong potential for a sensor networks such as hospital system or modern

transport system, where reduced RF transmission around the patients or passengers is a

desired goal.

138

Chapter 7

Conclusion

Four primary challenges of a conventional RFID based batteryless system are identified:

(a) clutter effect, (b) sensor integration, (c) response time, and (d) RF radiation. Many

possible scenarios for how the challenges arise are discussed with various applications. Next,

solutions are proposed for each of the challenges as (a) harmonic RFID for enhanced opera-

tion under cutter effect, (b) low power digital interface sensing platform for electrochemical

sensor electrode integration for biochemical applications, (c) component level circuit analysis

to model the harvesting efficiency and response time, and finally (d) dual mode RFID to

effectively reduce the RF radiation.

A miniaturized single antenna based harmonic RFID tag was developed and demonstrated

with prototype. This harmonic RFID tag only requires the harmonic generator as an extra

component in addition to other components used in conventional RFID. The harmonic RFID

will provide a better SNR compared to conventional RFID in presence of clutter.

A pH sensor integrated RFID tag was fabricated and demonstrated for biochemical ap-

plications. The proposed sensing system illustrates a low power digital interface for analog

sensors. The architecture is scalable and multiple sensors can be integrated with a single

RFID.

A dual mode RFID tag was also developed and shown to reduce the effective on time of

RFID reader. The dual mode RFID tag stores the redundant harvested energy and can use

139

the energy to monitor a sensor locally. The RFID reader will not be required to transmit

RF power during this local monitoring. In dual mode tag, the tag can synthesize its own

carrier and use it for communication while the reader does not provide any carrier.

Harmonic RFID, sensor integrated RFID, and dual mode RFID has inter-compatibility

and one can be merged with other as per requirement. Hence, a hybrid RFID can be formed

by integrating all the three RFID structures. In many applications, combination of the four

primary challenges can arise and a hybrid RFID will be capable in mitigating the challenges

simultaneously.

7.1 Future Work

In this thesis work, the primary work was development of different functional elements

of hybrid RFID system and the reported prototypes were fabricated on a Printed Circuit

Board (PCB) with discrete components. The final circuit board was quite big due to use

of individually packaged components. The proposed future work primarily focuses on three

aspects: (a) miniaturization, (b) security, and (c) packaging.

7.1.1 Miniaturization

The complete circuit of a hybrid RFID tag can be miniaturized greatly by using mono-

lithic fabrication of different modules on a single substrate. There are primarily three parts

for a hybrid RFID tag: (a) RF part, (b) digital part, and (c) antenna. In the current designs,

the RF part consumes the primary real estate area. Using monolithic fabrication of CMOS

process, a hybrid RFID tag can be miniaturized to a comparable size of conventional RFID

tag circuit.

140

Apart from the tag circuit area, another area consuming part is antenna. While the

tag circuit area can be miniaturized using monolithic fabrication without compromising the

circuit performance, the antenna miniaturization is related to the tag performance. As the

antenna gets smaller, the radiation efficiency reduces, impacting on the overall tag perfor-

mance. There are primarily two ways for miniaturizing the tag antenna: a) use of material

with high relative permittivity, or permeability, or both; and b) efficient design. Use of high

dielectric constant substrate or magnetic material can reduce the effective wavelength and

hence the overall antenna dimension will be smaller. However, the substrate loss increases at

high frequency with increase in relative permittivity or permeability of a material. Again, a

tradeoff should be maintained, as the antenna miniaturization taking advantage of material

property will impact on the antenna performance. On the other hand, using optimized design

with miniaturized size and considerable gain is very much desired. Use of an optimization

tool such as genetic algorithm (GA) or particle swarm optimization (PSO) for antenna op-

timization will be helpful for optimizing with multiple objectives such as gain maximization

and area minimization. Also, antennas are usually designed as a two-dimensional planar

structure. Taking advantage of all the three dimensions can help in minimizing the antenna

structure without sacrificing much gain. However, fabrication in bulk and installing the

RFIDs with 3D antenna will be a challenge.

7.1.1.1 Electrically small antenna

In attempt to continue the above miniaturization challenges specifically for antenna,

it is worthwhile to mention the requirement of small antenna design. A small antenna is

defined as electrically small antenna with maximum dimension < λ/3, where λ is the effective

wavelength. A significant amount of research has been done previously on small antenna and

141

it would be worthwhile to mention the development and integration of small antennas with

RFID tags [138–140]. With antenna dimension becoming smaller, the fundamental challenges

appear as reduced bandwidth, reduced gain, and reduced coupling with reader antenna, all

of which can effect on the RFID system performance [140–142]. Hence, apart from making

the tag antenna smaller, a broader level system analysis should also be performed keeping

those three factors in mind.

7.1.2 Security

The only authentication of a RFID tag is the identification bits, which can be cloned and

can be a big security breach. In next five years, there is an estimated 19.21% compound

annual growth rate of RFID devices in the electronic payment market. For at least this

reason, there is a need to procure data securely from RFID sensors. A tamper-proof secured

RFID tag can be implemented using multiple harmonics monitoring technique [143]. An

extra layer of authentication can be added to existing RFID tags by monitoring multiple

harmonics. As there will always be tolerance in the components used to design the NLTL,

there will always be different mismatch in the impedance of a single NLTL stage resulting in a

different amplitude level and phase of the received harmonic levels. The different higher order

harmonics would be a unique signature or fingerprint of the specific hybrid RFID tag under

identical excitation modulation signals. The relative power level and the phase difference of

the higher order harmonics is stored for a specific RFID during initial activation and always

verified later against the information. A wide band antenna capable of transmitting multiple

harmonics is necessary for both the RFID tag and the reader.

142

7.1.3 Packaging

Bulk scale packaging technology is important for hybrid RFID tags as that can effectively

reduce the final cost of the hybrid RFID tag. The hybrid RFID tag will have primarily three

components, which need to be integrated at the final stage of packaging: 1) antenna, 2)

RFID IC, and 3) sensing elements. Single frequency resonating antenna is required for a

hybrid RFID tag except harmonic operation. Hence, options for antenna can be kept at

final stage of integration. Like conventional RFID, the hybrid RFID IC can be fabricated

separately and then integrated with the desired antenna at the final stage of packaging.

The electrochemical material requirement is different for different parameter sensing, hence

the sensors need to be fabricated separately. At the final stage of packaging, the desired

sensor heads can be integrated with the hybrid RFID IC along with the antenna. Different

bulk printing techniques such as pad printing or inkjet printing can be used for the final

stage integration. Pad printing is a fast printing technique for thin-film substrate based

integration, whereas inkjet printing is more desired for fine resolution. Combination of those

two printing techniques can be used for a complete hybrid RFID fabrication. The antenna

can be manufactured using pad printing, whereas inkjet printing can be used for interconnect

printing between antenna, IC, and sensors together.

143

APPENDICES

144

APPENDIX A

Modulation Depth Formulation of

RFID Tag Impedance

A simplified representation of coupled reader and tag antenna with different dimensions

is shown in Fig. A.1. The center of the two antennas are at z = 0, and separated by r0. As

shown in Fig. A.1, both the antennas are cylindrical center-fed dipoles. The two antenna

network can be represented as a four-terminal impedance network as shown in Fig. A.2.

The impedance network with different impedance parameters is shown in Equation (A.1).

For a reciprocal network, we can write Z12 = Z21. In Equation (A.1), V1 and V2 represents

the voltage across the antenna terminals and I1(0) and I2(0) represents the current flowing

into the feeding terminals at location z = 0. Now, if the tag antenna is terminated with a

load impedance ZL, the terminal voltage of the tag will be according to Equation (A.2).

V1 = Z11I1(0) + Z12I2(0)

V2 = Z21I1(0) + Z22I2(0)

V2 = −ZLI2(0)

(cid:20)
Z11 −

V1 =

(cid:21)

I1(0)

Z2
12

Z22 + ZL

145

(A.1)

(A.2)

(A.3)

Figure A.1: Two cylindrical dipole antennas with different dimensions.

Figure A.2: Four-terminal impedance network for two antenna system.

146

Using reciprocity and substituting Equation (A.2) in Equation (A.1), we can obtain a

relationship between V1 and I1(0), which is given in Equation (A.3). Now in Equation (A.3),

there are two parts in the impedance expression. If the input impedance of reader antenna is

considered as Zin,1 in free space, then the back-scattered voltage V s

r induced by tag antenna

at the reader antenna is given in Equation (A.4).

V1 = V s

r + Zin,1I1(0)

(cid:20)(cid:0)Z11 − Zin,1

(cid:1) −

V s
r =

(cid:21)

I1(0)

Z2
12

Z22 + ZL

(A.4)

(A.5)

Packlington’s EFIE for a single antenna

Figure A.3: A single center-fed dipole antenna.

Before deriving the system of equations for two antenna system, first the Electric Field

Integral Equation (EFIE) is formed for a single antenna. To begin with, a few approximations

are considered. First, the antennas are considered as center-fed cylindrical dipole.

’Thin-

wire’ approximation is considered for the antenna, which means dipole length is much higher

147

than dipole diameter. ’Slice-gap’ terminal generator model is used at the feed point of the

antenna. And finally, the antenna is considered made of a perfect conductor. The schematic

of the antenna is provided in Fig. A.3. On a perfect conductor surface, the total tangential

electric field is zero. Hence, we can write

ˆt · (cid:126)E = 0
or,
ˆt · ( (cid:126)Ei + (cid:126)Es) = 0

(A.6)

The electric filed (cid:126)E represents the total electric field. (cid:126)Ei represents the impressed electric

field, which excites the system and (cid:126)Es represents the scattered electric field due to induced
In the following notations, (cid:126)r(cid:48) denotes the source vector and (cid:126)r

current (cid:126)J excited by (cid:126)Ei.

denotes the vector at evaluation point, on the surface of conductor. We can write (cid:126)Es((cid:126)r) due
to (cid:126)J((cid:126)r(cid:48)) as below

(cid:90)

S

(cid:2)∇(cid:48) · (cid:126)J((cid:126)r(cid:48))∇ + β2 (cid:126)J((cid:126)r(cid:48))(cid:3)e−jkR
surface. In Equation (A.7), η is given as(cid:112)µ/
 and β is provided as ω

(cid:126)Es((cid:126)r) = −jη
β

dS(cid:48)

4πR

(cid:126)J((cid:126)r(cid:48)) denotes the surface current on surface S and ˆt is an unit vector tangential to the

(A.7)

√

µ
, where µ is the

permeability and 
 is the permittivity of the medium through which (cid:126)Es propagates. Substi-

tuting Equation (A.7) in Equation (A.6), the following expression can be shown. With the

’slice-gap’ model, the terminal voltage at the antenna input can be substituted in Equation

(A.8) to form Equation (A.9). In Equation (A.9), δ(z) is dirac delta function.

(cid:90)

S

jη
4πβ

(cid:2)∇(cid:48) · (cid:126)J((cid:126)r(cid:48)) ˆt · ∇ + β2ˆt · (cid:126)J((cid:126)r(cid:48))(cid:3)e−jkR

R

dS(cid:48) = ˆt · (cid:126)Ei((cid:126)r)

(A.8)

148

(cid:90)

S

jη
4πβ

(cid:2)∇(cid:48) · (cid:126)J((cid:126)r(cid:48)) ˆt · ∇ + β2ˆt · (cid:126)J((cid:126)r(cid:48))(cid:3)e−jkR

R

dS(cid:48) = V δ(z − 0)

(A.9)

For a thin wire antenna as shown in Fig. A.3, the current distribution is primarily a

function along the ˆz distribution and is a weak function of radial distribution. By consid-
ering the current distribution (cid:126)J((cid:126)r(cid:48)) = f (φ(cid:48))I(z(cid:48)) ˆz(cid:48) along the wire length, and few vector

operations, the EFIE can be modified as shown in Equation (A.10). In Equation (A.10),
f (φ(cid:48)) represents the circular current distribution function and Γ is the axial path along which

the line integral is performed.

(cid:90)

(cid:20)∂I(z(cid:48))

dz(cid:48)

(cid:21)(cid:73)
+ β2I(z(cid:48))

C

∂
dz

f (φ(cid:48))

e−jkR
R

dφ(cid:48)dz(cid:48) = V δ(z)

(A.10)

jη
4πβ

Γ

After proper substitution of vector operators, and finally substituting the boundary con-
ditions of current at the dipole ends, which is I(l) = I(l(cid:48)) = 0, Equation (A.10) takes form
of famous Packlington’s form of EFIE as shown in Equation (A.11), where g(z|z(cid:48)) = e−jkr
and r =(cid:112)((z − z(cid:48))2 + a2).

4πr

(cid:20) ∂2

(cid:90) l

−l

∂z2 + β2

(cid:21)

jη
4πβ

I(z(cid:48))g(z|z(cid:48)) dz(cid:48) = V δ(z)

(A.11)

Coupling of two antennas

Once the EFIE is formulated for a single antenna, it can be extended to a system of

two or more antennas. For back-scattering analysis, only two antennas are considered. The

formulation for two antenna system as shown in Fig. A.1 becomes

(cid:90) l1

−l1

V1δ(z) =

(cid:90) l2

−l2

I1(z(cid:48))G11(z|z(cid:48)) dz(cid:48) +

149

I2(z(cid:48))G12(z|z(cid:48)) dz(cid:48)

(A.12)

(cid:90) l1

−l1

V2δ(z) =

I1(z(cid:48))G12(z|z(cid:48)) dz(cid:48) +

where,

G11(z − z(cid:48)) =

jβη
4π

1
β2

∂2
∂z2

(cid:90) l2

−l2

(cid:18)

1 +

(A.13)

I2(z(cid:48))G22(z|z(cid:48)) dz(cid:48)
(cid:19) e
(cid:113)

(cid:113)
((z−z(cid:48))2+a2
−jk
1)
((z − z(cid:48))2 + a2
1)

The other kernels G12 and G22 are obtained by replacing a1 with r0 and a2 in Equations

(A.12), (A.13), where r0 and a2 are distance between the two antennas and radius of the

tag-side antenna as shown in Fig. A.1.

Variational formulation

Equations (A.12), (A.13) are the founding equation set for the variational formulation. In

the following, steps from [33] is followed to establish the variation relation of the impedance.

Once, Equation (A.1) is substituted in Equations (A.12) and (A.13), the equation pair

becomes as shown in Equation (A.14) and (A.15).

(cid:2)Z11I1(0) + Z12I2(0)(cid:3)δ(z) =

(cid:2)Z12I1(0) + Z22I2(0)(cid:3)δ(z) =

(cid:90) l1

−l1

(cid:90) l1

−l1

I1(z(cid:48))G11(z|z(cid:48)) dz(cid:48) +

I1(z(cid:48))G12(z|z(cid:48)) dz(cid:48) +

(cid:90) l2

−l2

(cid:90) l2

−l2

I2(z(cid:48))G12(z|z(cid:48)) dz(cid:48)

(A.14)

I2(z(cid:48))G22(z|z(cid:48)) dz(cid:48)

(A.15)

Let, I(cid:48)

1(0) and I(cid:48)

2(0) be another pair of feeding current, then I(cid:48)

1(0) and I(cid:48)

2(0) pair will

also satisfy the Equations (A.14) and (A.15). If Equation (A.14) is multiplied by I(cid:48)
Equation (A.15) by I(cid:48)

1(z) and
2(z) followed by integration of both of them individually from −l1 to

150

l1 and −l2 to l2. After integration, followed by summing up, one obtains Equation (A.16).

For Equation (A.16), at the left hand side, the integration results (cid:82) l1
and(cid:82) l2

I1(z)δ(z)dz = I1(0)

I2(z)δ(z)dz = I2(0) are used.

l1

l2

2(0)Z12I2(0)

1(0)Z12I2(0) + I(cid:48)
1(0)Z11I1(0) + I(cid:48)
I(cid:48)
l1(cid:90)
l1(cid:90)
l1(cid:90)
l2(cid:90)

=
−l1

2(0)Z21I1(0) + I(cid:48)
l1(cid:90)
l2(cid:90)

1(z) dz(cid:48) dz +

I1(z(cid:48))G11(z|z(cid:48))I(cid:48)

−l1

−l1

I1(z(cid:48))G12(z|z(cid:48))I(cid:48)

2(z) dz(cid:48) dz +

l2(cid:90)
l2(cid:90)

−l2

+
−l2

−l1

I2(z(cid:48))G12(z|z(cid:48))I(cid:48)

1(z) dz(cid:48) dz

I2(z(cid:48))G22(z|z(cid:48))I(cid:48)

2(z) dz(cid:48) dz

(A.16)

−l2

−l2

Once, we obtain Equation (A.16), one can substitute I(cid:48)

1(0) = I1(0) and I(cid:48)

2(0) = I2(0)

1(0), I(cid:48)

as the current pair I1(0), I2(0) and I(cid:48)
current, the distribution currents become also same, reducing to I(cid:48)
2(z) =
I2(z). Also, by exchanging z(cid:48) and z variables of the third integrand in Equation (A.16) and
due to (G12(z|z(cid:48)) = G12(z(cid:48)|z)), one can obtain Equation (A.17).

2(0) excite same system. With same excitation

1(z) = I1(z) and I(cid:48)

Z11I2

1 (0) + 2Z12I1(0)I2(0) + Z22I2

2 (0)

I1(z(cid:48))G11(z|z(cid:48))I1(z) dz(cid:48) dz + 2

I2(z(cid:48))G12(z|z(cid:48))I1(z) dz(cid:48) dz

(A.17)

l1(cid:90)

l2(cid:90)

−l1

−l2

l1(cid:90)
l2(cid:90)

l1(cid:90)
l2(cid:90)

−l1

=
−l1

+
−l2

−l2

I2(z(cid:48))G22(z|z(cid:48))I2(z) dz(cid:48) dz

Using Equation (A.17), one can relate the variational relation in between current and

151

impedance. If a small variation of current at the antenna input terminal is considered about

its true value, in Equation (A.17), one can replace I1(0) with I1(0) + δI1(0), I2(0) with

I2(0) + δI2(0). Due to the excitation current change, if there is any change in impedance,

which one can consider as Z11 + δZ11, Z12 + δZ12, and Z22 + δZ22. A subsequent change in
current distribution is also expected, which is considered as I1(z(cid:48))+δI1(z(cid:48)) and I2(z(cid:48))+δI2(z(cid:48))
in place of I1(z(cid:48)) and I2(z(cid:48)) correspondingly.

In this analysis, only first order variation

is considered, which means any second order variations are neglected. The second order

variations such as (δIi(0))2, (δZij)2, δZijδZji, δIi(0)δZij, and(cid:82)(cid:82) δIi(z(cid:48))Gij(z|z(cid:48))δIj(z)dz(cid:48)dz

are considered as 0, where i = 1, 2, and j = 1, 2. After substitution and simplification,

Equation (A.17) can be expressed as in Equation (A.18).

l1(cid:90)

δZ11I2

1 (0) + 2δZ12I1(0)I2(0) + δZ22I2

2 (0)

+2δI1(0)(cid:2)Z11I1(0) + Z12I2(0)(cid:3) + 2δI2(0)(cid:2)Z12I1(0) + Z22I2(0)(cid:3)
(cid:34) l1(cid:90)
l1(cid:90)
l2(cid:90)

δI1(z(cid:48))G11(z|z(cid:48))I1(z) dz(cid:48) dz +
l2(cid:90)

−l1
δI1(z(cid:48))G12(z|z(cid:48))I2(z) dz(cid:48) dz +

δI2(z(cid:48))G12(z|z(cid:48))I1(z) dz(cid:48) dz
(cid:35)

−l2
δI2(z(cid:48))G22(z|z(cid:48))I2(z) dz(cid:48) dz

l1(cid:90)
l2(cid:90)

l2(cid:90)

−l1

=2

−l1

+
−l2

−l1

−l2

−l2

(A.18)

By multiplying Equations (A.14) and (A.15) with δI1(z) and δI2(z) consecutively, inte-
gration can be performed on them individually from −l1 to l1 and −l2 to l2. After addition

of the two integrations, one can obtain Equation (A.19). After substitution of Equation

(A.19) in Equation (A.18), one can obtain Equation (A.20).

152

δI1(0)(cid:2)Z11I1(0) + Z12I2(0)(cid:3) + δI2(0)(cid:2)Z12I1(0) + Z22I2(0)(cid:3)

l1(cid:90)
l2(cid:90)

l2(cid:90)
l2(cid:90)

−l1

−l2

−l2

−l2

δI2(z(cid:48))G12(z|z(cid:48))I1(z) dz(cid:48) dz

(A.19)

δI2(z(cid:48))G22(z|z(cid:48))I2(z) dz(cid:48) dz

l1(cid:90)
l2(cid:90)

l1(cid:90)
l1(cid:90)

−l1

=
−l1

+
−l2

−l1

δI1(z(cid:48))G11(z|z(cid:48))I1(z) dz(cid:48) dz +

δI1(z(cid:48))G12(z|z(cid:48))I2(z) dz(cid:48) dz +

δZ11I2

1 (0) + 2δZ12I1(0)I2(0) + δZ22I2

2 (0) = 0

(A.20)

The only solution to Equation (A.20) is δZ11 = δZ12 = δZ22 = 0. The solution is

very important from the variational method. This shows if approximate currents differ from

the true currents by first order errors, the impedance parameters differ from true values by

second order error. Hence, an approximate current would provide a somewhat more accurate

impedance result.

Current approximation in linear dipole

The current distribution in the dipole antennas can now be approximated to obtain the

impedance parameters. Using the analytical current distribution at antenna 1 and antenna

2, the impedance parameter can be obtained. A reasonable current distribution over linear

antennas can be assumed as linear combination of a constant term, a sine term, and a cosine
term as shown in Equation (A.21). Using the conditions as I(z) = I(−z), I(l) = I(−l) = 0,

and I(0) as the feeding terminal current, Equation (A.21) reduces to Equation (A.22).

I(z) = a + b cosβz + c sinβz

(A.21)

153

I(z) = I(0)

cosβz − cosβl

1 − cosβl

+ k(cid:2)sinβ(l − |z|) + sinβ|z| − sinβl(cid:3)

(A.22)

I(z) = I0(z) + kI1(z)

where

k =

c

1 − cosβl

However, the above approximation fails to represent the current distribution when l ≥ λ.

Since both sine and cosine terms are periodic functions, their values start repeating as

the argument βz changes by 2π. A better approximation can be formed by introducing

another term cos βz

2 and the current distribution takes form of three term King Wu current

approximation. In this analysis, as dipole length upto l = λ is considered for both the reader

and tag antennas, current distribution as in Equation (A.21) is sufficient for this analysis.

Modulation depth formulation

Now in scenario of A.2, if both the antennas are fixed at their positions and the load

impedance of antenna 2 or tag antenna is changed, then the back-scattered voltage V s

r at

antenna 1 terminal would be changed according to ZL. By changing the impedance ZL

sequentially, the voltage V s

r will be modulated and that is how information from RFID tag

can be obtained at the RFID reader. For example, if V s

r (1) is induced due to ZL1 and V s

r (2)

due to ZL2, the tag would be more easily detectable with higher difference between V s

r (2). The quantity(cid:0)V s

r (1)−V s

and V s
to find load impedance values for set {ZL1, ZL2} to maximize the modulation depth.

r (2)(cid:1) is termed as the modulation depth. It is inquisitive
(cid:20)

r (1)

(cid:21)

I1(0)

(A.23)

r (2) − V s
V s

r (1) = Z2
12

1

−

1

Z22 + ZL2

Z22 + ZL1

Now, the current distribution from Equation (A.22) can be used in Equation (A.16)

154

to find expressions for different impedance parameters. From Equation (A.23), only the

impedance parameters Z22 and Z12 need to be found. According to [33], for Z22, considering
1(0) = I1(0) = 0 and I(cid:48)
I(cid:48)
choosing I1(0) = I(cid:48)

2(0) = I2(0) in Equation (A.17) provides Equation (A.24). And

2(0) = 0 in Equation (A.16) would provide Z12 as provided in Equation

(A.25).

Z22I2

2 (0) =

l1(cid:90)
l2(cid:90)

−l1

I1(z(cid:48))G11(z|z(cid:48))I1(z) dz(cid:48) dz + 2

l1(cid:90)

l2(cid:90)

−l1

−l2

I2(z(cid:48))G22(z|z(cid:48))I2(z) dz(cid:48) dz

l1(cid:90)
l2(cid:90)

−l1

−l2

I2(z(cid:48))G12(z|z(cid:48))I1(z) dz(cid:48) dz

(A.24)

+
−l2

I(cid:48)
1(0)Z12I2(0) =

l1(cid:90)
l2(cid:90)

−l1

l1(cid:90)
l1(cid:90)

−l1

−l1

+
−l2

I1(z(cid:48))G11(z|z(cid:48))I(cid:48)

1(z) dz(cid:48) dz +

I1(z(cid:48))G12(z|z(cid:48))I(cid:48)

2(z) dz(cid:48) dz +

l1(cid:90)
l2(cid:90)

−l1

−l2

l2(cid:90)
l2(cid:90)

−l2

−l2

I2(z(cid:48))G12(z|z(cid:48))I(cid:48)

1(z) dz(cid:48) dz

I2(z(cid:48))G22(z|z(cid:48))I(cid:48)

2(z) dz(cid:48) dz

(A.25)

Approximation for short dipole at reader side

The set of equations becomes simplified with approximation of short dipole at the reader

side. With short dipole approximation only for reader antenna, the impedance parameters

in Equations (A.24) and (A.25) can be simplified for the coupled antenna system. For
small dipole with (r0 >> l2) reasonable approximations as G12(z|z(cid:48)) = G12(z(cid:48)) for I1(z) =
I1(0) and G12(z|z(cid:48)) = G12(z) for I1(z(cid:48)) = I1(0) can be considered. With approximation

I1(0) = 0, Equation (A.24) reduces as in Equation (A.26). For Equation (A.25), with earlier

155

approximations as I1(0) = 0 and I(cid:48)

2(0) = 0, it reduces to Equation (A.27). As the reader

antenna considered is a short dipole, the parameter ∆l is the length of the dipole.

l2(cid:90)

l2(cid:90)

−l2

−l2

Z22I2

2 (0) =

I2(z(cid:48))G22(z|z(cid:48))I2(z) dz(cid:48) dz

(A.26)

I(cid:48)
1(0)Z12I2(0) =I(cid:48)

1(0)∆l

l2(cid:90)

−l2

G12(z(cid:48))I2(z(cid:48)) dz(cid:48) +

l2(cid:90)

l2(cid:90)

−l2

−l2

I2(z(cid:48))G22(z|z(cid:48))I(cid:48)

2(z) dz(cid:48) dz

(A.27)

Now, current distribution from Equation (A.22) can be substituted for the tag antenna

in Equations (A.26) and (A.27) and the following form as in Equations (A.28) and (A.29) is
obtained. To obtain Equation (A.29), I(cid:48)

2(0) = 0 is considered as earlier.

l2(cid:90)

l2(cid:90)

−l2

−l2

Z22I2

2 (0) =

(I0(z(cid:48)) + k22I1(z(cid:48)))G22(z|z(cid:48))(I0(z) + k22I1(z)) dz(cid:48) dz

(A.28)

l2(cid:90)

−l2

1(0)∆l

1(0)Z12I2(0) = I(cid:48)
I(cid:48)
l2(cid:90)

l2(cid:90)

G12(z(cid:48))(I0(z(cid:48)) + k12I1(z(cid:48))) dz(cid:48)

(A.29)

12I1(z(cid:48))G22(z|z(cid:48))(I0(z) + k12I1(z)) dz(cid:48) dz
k(cid:48)

+
−l2

−l2

The constants k22, k12, and k(cid:48)

12 are adjustable in the trial functions and as the impedance

expressions are stationary, all the constants can be determined by setting ∂Z/∂k = 0. After

finding the constants, the constants are substituted in Equations (A.28) and (A.29) to find

expressions in terms of the trial functions and the geometry of antenna.

156

l2(cid:90)

l2(cid:90)

−l2

−l2

Z22I2

2 (0) =

I0(z(cid:48))G22(z|z(cid:48))I0(z) dz(cid:48) dz −

(cid:34) l2(cid:82)
l2(cid:82)

−l2

−l2

l2(cid:82)
l2(cid:82)

−l2

−l2

(cid:35)2

I0(z(cid:48))G22(z|z(cid:48))I1(z) dz(cid:48) dz

I1(z(cid:48))G22(z|z(cid:48))I1(z) dz(cid:48) dz

(A.30)

1(0)∆l

G12(z(cid:48))I0(z(cid:48)) dz(cid:48)

1(0)Z12I2(0) = I(cid:48)
I(cid:48)
l2(cid:82)

(cid:34)

I(cid:48)
1(0)∆l

−l2

−

l2(cid:90)

−l2

l2(cid:82)

l2(cid:82)

−l2

−l2

(cid:35)(cid:34) l2(cid:82)

l2(cid:82)

(cid:35)

(A.31)

I0(z(cid:48))G22(z|z(cid:48))I1(z) dz(cid:48) dz

G12(z(cid:48))I1(z(cid:48)) dz(cid:48)

−l2

−l2

I1(z(cid:48))G22(z|z(cid:48))I1(z) dz(cid:48) dz

Now Equations (A.30) and (A.31) can be solved analytically to find the impedance pa-

rameters Z22 and Z12. After further simplification, the impedance parameters are given as

followed in Equation (A.32). The parameters A, B, C, D, E, F, G, andH are given in Equa-

tion (A.33)

(cid:21)
(cid:21)

(A.32)

(cid:20)
(cid:20)∆le−jβr0

Z22 = j30

Z12 = j30

(A + jB) − (C + jD)2
(cid:21)(cid:20)
(E + jF )

G − H

(C + jD)
(E + jF )

r0

157

S(4L) + 4cosL(LcosL − sinL)(cid:2)ln

− C(2L)(cid:3)

4L
α

(cid:21)

αsin2L − 2cos2Lsin2L

− C(4L) − (4Lcos2L − 2sin2L)S(2L) + sin22L

(cid:21)

(cid:20)

(cid:20)
(cid:20)

(cid:20)

1

(1 − cosL)2
− 8
π

(1 − cosL)2

1

1

A =

B =

C =

D =

(sinL)S(4L) + (1 − cosL)C(4L) + 4cosL(1 − cosL − LsinL)C(2L)

(1 − cosL)
+ 4cosL(1 − cosL)C(L) + 4sinL(LcosL − sinL)ln
− 8
π

αsinL(1 − cosL) − (sin2L)2

(cid:21)

4L
α

+ 4(1 − cosL)2ln

2L
α

1

(1 − cosL)S(4L) + (sinL)C(4L) + 4cosL(1 − cosL − LsinL)S(2L)

(1 − cosL)
+ 4cosL(1 − cosL)S(L) + 2sin2Lsin2L

E =2cosL(1 − cosL)S(4L) + 2sinL(1 − cosL)C(4L) + 4(1 − cosL)2S(2L)

− 4L(sin2L)C(2L) + 8sinL(1 − cosL)C(L) +(cid:2)4Lsin2L − 8sinL(1 − cosL)(cid:3)ln

4L
α

− 24
π

α(1 − cosL)2 − 2sin2Lsin2L

F =2sinL(1 − cosL)S(4L) − 2cosL(1 − cosL)C(4L) − 4L(sin2L)S(2L)

− 4(1 − cosL)2C(2L) + 8sinL(1 − cosL)S(L) + 4sin4L

2

(sinL − LcosL)

G =

(1 − cosL)

H =2(cid:2)2(1 − cosL) − LsinL(cid:3)

where

L = βl, α = βa, C(x) =

(A.33)

1 − cosu

u

du, S(x) =

(cid:90) x

sinu

0

u

du

(cid:90) x

0

158

APPENDIX B

Copper Etching Process

Procedure: 1. Substrate Preparation and Cleaning.

(a) Cut the substrate to desired dimensions.

(b) Rinse the substrate in acetone for 30 seconds.

(c) Rinse the substrate in methanol for 30 seconds.

(d) Rinse the substrate in DI water for 30 seconds.

(e) Rinse the substrate in isopropanol alcohol (IPA) for 30 seconds.

(f) Blow dry using Nitrogen.

(g) Bake the substrate at 1050C for 3 minutes.

2. Spin-Coat Photoresist.

(a) Set the rpm and time on Laurell Technologies Spinner (WS400B-6NPP LITE). Typi-

cally, for 17 µm - 30 µm thick Copper laminates, rpm is 3000 for 30 seconds.

(b) Spin the substrate after dropping positive photoresist S1813 over the substrate.

(c) Soft bake the substrate at 950 for 45 seconds.

3. Expose and Develop

159

(a) Calculate the expose time based on the power of the UV lamp in MJB 3 Mask Aligner

and the spin coated thickness of the photoresist.

(b) Typically, expose for 45 seconds (3000 rpm, 30 sec - spin).

(c) Bake the substrate at 1050 for 30 seconds.

(d) Develop using MF-319 for 30 seconds.

(e) Rinse with DI water and blow dry using Nitrogen.

(f) Hard bake at 1150C for 5 minutes.

4. Copper Etching

(a) Prepare the etchant by mixing Sodium per sulphate in DI water with 1:4 ratio (250 g

in 1000 ml).

(b) Heat the etchant at 800C.

(c) Immerse the exposed substrate until the desired copper pattern is etched.

(d) Rinse with acetone, methanol for 30 seconds each.

(e) Rinse with DI water and bake at 1050 for 5 minutes.

160

BIBLIOGRAPHY

161

BIBLIOGRAPHY

[1] Y. Ma, X. Hui, and E. C. Kan, “Harmonic-WISP: A passive broadband harmonic RFID
IEEE, 2016,

platform,” in IEEE MTT-S International Microwave Symposium (IMS).
pp. 1–4.

[2] S. Mondal, S. Karuppuswami, D. Kumar, and P. Chahal, “Scope and application of
IEEE,

harmonic RFID for implanted body area network,” in IEEE RFID Conference.
2020, pp. xx–xx.

[3] S. Mondal, D. Kumar, and P. Chahal, “A wireless passive pH sensor with clutter

rejection scheme,” IEEE Sensors Journal, vol. 19, no. 9, pp. 3399–3407, 2019.

[4] S. Karuppuswami, M. I. M. Ghazali, S. Mondal, and P. Chahal, “Wireless EAS sensor
tags for volatile profiling in food packages,” in 2018 IEEE 68th Electronic Components
and Technology Conference (ECTC).

IEEE, 2018, pp. 2174–2179.

[5] M. Yuan, P. Chahal, E. C. Alocilja, and S. Chakrabartty, “Wireless biosensing using
silver-enhancement based self-assembled antennas in passive radio frequency identifi-
cation (RFID) tags,” IEEE Sensors Journal, vol. 15, no. 8, pp. 4442–4450, 2015.

[6] C. Occhiuzzi, S. Cippitelli, and G. Marrocco, “Modeling, design and experimentation of
wearable RFID sensor tag,” IEEE Transactions on Antennas and Propagation, vol. 58,
no. 8, pp. 2490–2498, 2010.

[7] Y. Ma and E. C. Kan, “Accurate indoor ranging by broadband harmonic generation in
passive nltl backscatter tags,” IEEE transactions on microwave theory and techniques,
vol. 62, no. 5, pp. 1249–1261, 2014.

[8] M. J. W. Rodwell, “Picosecond electrical wavefront generation and picosecond opto-

electronic instrumentation,” Ph.D. dissertation, Stanford University, 1987.

[9] F. Yu, K. G. Lyon, and E. C. Kan, “A novel passive rfid transponder using harmonic
generation of nonlinear transmission lines,” IEEE Transactions on Microwave Theory
and Techniques, vol. 58, no. 12, pp. 4121–4127, 2010.

[10] S. Ahuja and P. Potti, “An introduction to RFID technology,” Communications and

Network, vol. 2, no. 3, pp. 183–186, 2010.

[11] J. Riley, A. Smith, D. Reynolds, A. Edwards, J. Osborne, I. Williams, N. Carreck,
and G. Poppy, “Tracking bees with harmonic radar,” Nature, vol. 379, no. 6560, p. 29,
1996.

162

[12] D. Zhang, L. T. Yang, M. Chen, S. Zhao, M. Guo, and Y. Zhang, “Real-time locating
systems using active RFID for Internet of Things,” IEEE Systems Journal, vol. 10,
no. 3, pp. 1226–1235, 2016.

[13] G. EPCglobal, “Epc radio-frequency identity protocols generation-2 UHF RFID; spec-
ification for RFID air interface protocol for communications at 860 MHz–960 MHz,”
EPCglobal Inc., November, 2013.

[14] J. Landt, “The history of RFID,” IEEE potentials, vol. 24, no. 4, pp. 8–11, 2005.

[15] A. R. Koelle, S. W. Depp, and R. W. Freyman, “Short-range radio-telemetry for
electronic identification, using modulated RF backscatter,” Proceedings of the IEEE,
vol. 63, no. 8, pp. 1260–1261, 1975.

[16] S. Kumar, E. Swanson, and T. Tran, “RFID in the healthcare supply chain: usage and
application,” International Journal of Health Care Quality Assurance, vol. 22, no. 1,
pp. 67–81, 2009.

[17] C. Costa, F. Antonucci, F. Pallottino, J. Aguzzi, D. Sarri´a, and P. Menesatti, “A
review on agri-food supply chain traceability by means of RFID technology,” Food and
bioprocess technology, vol. 6, no. 2, pp. 353–366, 2013.

[18] C. Amador, J.-P. Emond, and M. C. do Nascimento Nunes, “Application of RFID
technologies in the temperature mapping of the pineapple supply chain,” Sensing and
Instrumentation for Food Quality and Safety, vol. 3, no. 1, pp. 26–33, 2009.

[19] S. Piramuthu and W. Zhou, RFID and sensor network automation in the food industry:
Ensuring quality and safety through supply chain visibility. Wiley Online Library, 2016.

[20] D. L. Brock, “The Electronic Product Code (EPC)-A naming scheme for physical

objects,” White paper, 2001.

[21] ——, “Integrating the Electronic Product Code (EPC) and the Global Trade
Item Number (GTIN),” White Paper available at www. autoidcenter. org/pdfs/MIT-
WUTOID-WH-004. pdf, vol. 25, 2001.

[22] A.-I. Center, “860MHz-960MHz Class I radio frequency identification tag radio fre-
quency & logical communication interface specification proposed recommendation Ver-
sion 1.0.0,” AutoID Center, MIT, 2002.

[23] S. Sarma, “Towards the 5 cents tag,” Auto ID White Paper MIT-AUTOID-WH-006,

2001.

[24] F. Bauchot, J.-Y. Clement, G. Marmigere, and P. Secondo, “Method and system for

locating tires using RFID,” Jan. 6 2009, uS Patent 7,474,214.

163

[25] A. Dolgui and J.-M. Proth, “RFID technology in supply chain management: state of
the art and perspectives,” IFAC Proceedings Volumes, vol. 41, no. 2, pp. 4464–4475,
2008.

[26] T. Zefferer, “A survey and analysis of nfc based payment solutions for smartphones,”

in IADIS International Conference e-Society 2013, 2013, pp. 275–282.

[27] V. Coskun, B. Ozdenizci, and K. Ok, “A survey on near field communication (nfc)

technology,” Wireless personal communications, vol. 71, no. 3, pp. 2259–2294, 2013.

[28] E. Haselsteiner and K. Breitfuß, “Security in near field communication (nfc),” in Work-

shop on RFID security.

sn, 2006, pp. 12–14.

[29] S. Roy, V. Jandhyala, J. R. Smith, D. J. Wetherall, B. P. Otis, R. Chakraborty,
M. Buettner, D. J. Yeager, Y.-C. Ko, and A. P. Sample, “Rfid: From supply chains to
sensor nets,” Proceedings of the IEEE, vol. 98, no. 9, pp. 1583–1592, 2010.

[30] A. P. Sample, D. J. Yeager, and J. R. Smith, “A capacitive touch interface for passive
rfid tags,” in 2009 IEEE International Conference on RFID. IEEE, 2009, pp. 103–109.

[31] M. Buettner, R. Prasad, A. Sample, D. Yeager, B. Greenstein, J. R. Smith, and
D. Wetherall, “Rfid sensor networks with the intel wisp,” in Proceedings of the 6th
ACM conference on Embedded network sensor systems. ACM, 2008, pp. 393–394.

[32] “RFID based sensors,” http://https://axzon.com/products//, [Online; accessed 14-

October-2019].

[33] Y.-Y. Hu, “Back-scattering cross section of a center-loaded cylindrical antenna,” IRE

Transactions on Antennas and Propagation, vol. 6, no. 1, pp. 140–148, 1958.

[34] C. Levis and C. Tai, “A method of analyzing coupled antennas of unequal sizes,” IRE

Transactions on Antennas and Propagation, vol. 4, no. 2, pp. 128–132, 1956.

[35] R. F. Harrington, “Small resonant scatterers and their use for field measurements,”
IRE Transactions on Microwave Theory and Techniques, vol. 10, no. 3, pp. 165–174,
1962.

[36] B. S. Ciftler, A. Kadri, and I. G¨uven¸c, “IoT localization for bistatic passive UHF RFID
systems with 3-D radiation pattern,” IEEE Internet of Things Journal, vol. 4, no. 4,
pp. 905–916, 2017.

[37] D.-Y. Kim, H.-G. Yoon, B.-J. Jang, and J.-G. Yook, “Effects of reader-to-reader in-
terference on the UHF RFID interrogation range,” IEEE Transactions on Industrial
Electronics, vol. 56, no. 7, pp. 2337–2346, 2009.

164

[38] X. Hui and E. C. Kan, “Monitoring vital signs over multiplexed radio by near-field

coherent sensing,” Nature Electronics, vol. 1, no. 1, p. 74, 2018.

[39] M. I. M. Ghazali, S. Karuppuswami, and P. Chahal, “Embedded passive RF tags to-
wards intrinsically locatable buried plastic materials,” in IEEE 66th Electronic Com-
ponents and Technology Conference (ECTC).

IEEE, 2016, pp. 2575–2580.

[40] S. Mondal, D. Kumar, M. I. Ghazali, P. Chahal, L. Udpa, and Y. Deng, “Monitoring
and localization of buried plastic natural gas pipes using passive RF tags,” in AIP
Conference Proceedings, vol. 1949, no. 1. AIP Publishing LLC, 2018, p. 020020.

[41] S. Mondal and P. Chahal, “A passive harmonic RFID tag and interrogator develop-
ment,” IEEE Journal of Radio Frequency Identification, vol. 3, no. 2, pp. 98–107,
2019.

[42] J. Breitbarth, “Design and characterization of low phase noise microwave circuits,”

Ph.D. dissertation, University of Colorado at Boulder, 2006.

[43] H. Wu, B. Tao, Z. Gong, Z. Yin, and H. Ding, “A fast uhf rfid localization method
using unwrapped phase-position model,” IEEE Transactions on Automation Science
and Engineering, vol. 16, no. 4, pp. 1698–1707, 2019.

[44] H. Ma and K. Wang, “Fusion of RSS and phase shift using the Kalman filter for RFID

tracking,” IEEE Sensors Journal, vol. 17, no. 11, pp. 3551–3558, 2017.

[45] L. M. Ni, D. Zhang, and M. R. Souryal, “RFID-based localization and tracking tech-

nologies,” IEEE Wireless Communications, vol. 18, no. 2, pp. 45–51, 2011.

[46] D. Kumar, S. Mondal, A. Kaur, Y. Deng, and P. Chahal, “An RF backscatterer system
for real-time multi-sensing applications,” Sensors and Actuators A: Physical, vol. 303,
p. 111685, 2020.

[47] P. Chahal, D. Kumar, S. Mondal, and S. Karuppuswami, “Systems and methods for a

multiband sensing platform,” May 28 2020, uS Patent App. 16/695,758.

[48] S. Mondal, K. Wijewardena, S. Karrapuswami, D. Kumar, P. Chahal, M. Ghannam,
and M. Cuddihy, “A Wireless Battery-less Seat Sensor for Autonomous Vehicles,” in
2020 IEEE 70th Electronic Components and Technology Conference (ECTC).
IEEE,
2020, pp. 2289–2294.

[49] D. Kumar, S. Mondal, Y. Deng, and P. Chahal, “Wireless Battery-Free Harmonic
Communication System for Pressure Sensing,” Micromachines, vol. 11, no. 12, p. 1043,
2020.

165

[50] S. Mondal, D. Kumar, S. Karuppuswami, and P. Chahal, “A harmonic RF phase-shifter
based wireless pH sensor,” in 2018 IEEE 68th Electronic Components and Technology
Conference (ECTC).

IEEE, 2018, pp. 796–801.

[51] S. Karuppuswami, S. Mondal, D. Kumar, A. Kaur, and P. Chahal, “Thin-film based
passive RFID sensor tag for detection of packaged food volatiles,” in International
Symposium on Microelectronics, vol. 2019, no. 1.
International Microelectronics As-
sembly and Packaging Society, 2019, pp. 000 599–000 602.

[52] S. Karuppuswami, S. Mondal, M. I. M. Ghazali, and P. Chahal, “A multiuse fully 3-D
printed cavity sensor for liquid profiling,” IEEE Sensors Letters, vol. 3, no. 2, pp. 1–4,
2018.

[53] ——, “A Reusable 3D Printed Cavity Resonator for Liquid Sample Characterization,”
International Mi-

in International Symposium on Microelectronics, vol. 2018, no. 1.
croelectronics Assembly and Packaging Society, 2018, pp. 000 389–000 392.

[54] S. Karuppuswami, S. Mondal, D. Kumar, and P. Chahal, “A Compact Wireless Passive
Breath Analyzer for Health Monitoring,” in 2020 IEEE 70th Electronic Components
and Technology Conference (ECTC).

IEEE, 2020, pp. 1010–1015.

[55] M. Philipose, J. R. Smith, B. Jiang, A. Mamishev, S. Roy, and K. Sundara-Rajan,
“Battery-free wireless identification and sensing,” IEEE Pervasive computing, vol. 4,
no. 1, pp. 37–45, 2005.

[56] A. B. Miller, M. E. Sears, L. L. Morgan, D. L. Davis, L. Hardell, M. Oremus, and C. L.
Soskolne, “Risks to health and well-being from radio-frequency radiation emitted by
cell phones and other wireless devices,” Frontiers in Public Health, vol. 7, p. 223, 2019.

[57] A. Bortkiewicz, “Health effects of radiofrequency electromagnetic fields (RF EMF),”

Industrial health, vol. 57, no. 4, pp. 403–405, 2019.

[58] R. Singh, R. Nath, A. K. Mathur, and R. S. Sharma, “Effect of radiofrequency radiation
on reproductive health,” The Indian journal of medical research, vol. 148, no. Suppl 1,
p. S92, 2018.

[59] C. Mutti and A. Wittneben, “Robust signal detection in passive RFID systems,” in
Proceedings of the First International EURASIP Workshop on RFID Technology, 2007,
pp. 39–42.

[60] P. H. Cole, D. M. Hall, M. Loukine, and C. D. Werner, “Fundamental constraints
on RF tagging systems,” in Third Annual Wireless Symposium and Exhibition, Santa
Clara, 1995, pp. 294–303.

166

[61] P. V. Nikitin and K. Rao, “Performance limitations of passive UHF RFID systems,”
IEEE, 2006, pp.

in Antennas and Propagation Society International Symposium.
1011–1014.

[62] X. Hui and E. C. Kan, “Radio ranging with ultrahigh resolution using a harmonic
radio-frequency identification system,” Nature Electronics, vol. 2, no. 3, pp. 125–131,
2019.

[63] L. Zhang, D. Karasiewicz, B. Cifctioglu, and H. Wu, “A 1.6-to-3.2/4.8 GHz dual-
modulus injection-locked frequency multiplier in 0.18 µm digital CMOS,” in Radio
Frequency Integrated Circuits Symposium, (RFIC).

IEEE, 2008, pp. 427–430.

[64] F. Benson and J. Last, “Nonlinear-transmission-line harmonic generator,” in Proceed-

ings of the Institution of Electrical Engineers, vol. 112, 4.

IET, 1965, pp. 635–643.

[65] F. Yu, K. G. Lyon, and E. C. Kan, “A novel passive RFID transponder using harmonic
generation of nonlinear transmission lines,” IEEE Transactions on Microwave Theory
and Techniques, vol. 58, no. 12, pp. 4121–4127, 2010.

[66] Y. Zhao, J. R. Smith, and A. Sample, “NFC-WISP: A sensing and computationally en-
hanced near-field RFID platform,” in IEEE International Conference on RFID. IEEE,
2015, pp. 174–181.

[67] S. Bhadra, D. S. Tan, D. J. Thomson, M. S. Freund, and G. E. Bridges, “A wireless
passive sensor for temperature compensated remote pH monitoring,” IEEE Sensors
Journal, vol. 13, no. 6, pp. 2428–2436, 2013.

[68] E. Recommendation, “70-03 relating to the use of short range devices (SRD),” Version

of, vol. 18, 2009.

[69] H. M. Lu, C. Goldsmith, L. Cauller, and J.-B. Lee, “MEMS-based inductively coupled
RFID transponder for implantable wireless sensor applications,” IEEE Transactions
on Magnetics, vol. 43, no. 6, pp. 2412–2414, 2007.

[70] M. Adnan and E. Afshari, “Efficient microwave and millimeter-wave frequency multi-
pliers using nonlinear transmission lines in cmos technology,” IEEE Transactions on
Microwave Theory and Techniques, vol. 63, no. 9, pp. 2889–2896, 2015.

[71] M. J. Rodwell, M. Kamegawa, R. Yu, M. Case, E. Carman, and K. S. Giboney, “Gaas
nonlinear transmission lines for picosecond pulse generation and millimeter-wave sam-
pling,” IEEE Transactions on Microwave Theory and Techniques, vol. 39, no. 7, pp.
1194–1204, 1991.

167

[72] S. Mondal, M. I. M. Ghazali, S. Karuppuswami, A. Kaur, and P. Chahal, “A nonlinear
transmission line based harmonic rf tag,” in 2017 IEEE 67th Electronic Components
and Technology Conference (ECTC).

IEEE, 2017, pp. 2237–2242.

[73] K. S. Champlin and D. Singh, “Small-signal second-harmonic generation by a non-
linear transmission line (short paper),” IEEE transactions on microwave theory and
techniques, vol. 34, no. 3, pp. 351–353, 1986.

[74] F. Callegari, J. C. Boggio, H. Fragnito, and R. Arradi, “Mixing of signals in two-
pump fiber optic parametric amplifiers and wavelength converters,” in Optical Fiber
Communication Conference. Optical Society of America, 2004, p. MF18.

[75] Y. Chen and A. W. Snyder, “Four-photon parametric mixing in optical fibers: effect

of pump depletion,” Optics letters, vol. 14, no. 1, pp. 87–89, 1989.

[76] K. S. Rao, P. V. Nikitin, and S. F. Lam, “Antenna design for UHF RFID tags: A
review and a practical application,” IEEE Transactions on antennas and propagation,
vol. 53, no. 12, pp. 3870–3876, 2005.

[77] M. Sasaki, Q. Chen, and K. Sawaya, “Radiation efficiency of small loop antennas for
IEEE,

pager,” in Antennas and Propagation Society International Symposium, vol. 1.
1999, pp. 10–13.

[78] G. Marrocco, A. Fonte, and F. Bardati, “Evolutionary design of miniaturized meander-
line antennas for RFID applications,” in Antennas and Propagation Society Interna-
tional Symposium, vol. 2.

IEEE, 2002, pp. 362–365.

[79] O. O. Olaode, W. D. Palmer, and W. T. Joines, “Effects of meandering on dipole an-
tenna resonant frequency,” IEEE Antennas and Wireless Propagation Letters, vol. 11,
pp. 122–125, 2012.

[80] G. Marrocco, “Gain-optimized self-resonant meander line antennas for RFID applica-
tions,” IEEE Antennas and Wireless propagation letters, vol. 2, no. 1, pp. 302–305,
2003.

[81] K. W. Lui, O. H. Murphy, and C. Toumazou, “32-µm wirelessly-powered sensor plat-

form with a 2-m range,” IEEE Sensors Journal, vol. 12, no. 6, pp. 1919–1924, 2012.

[82] J. F. Dickson, “On-chip high-voltage generation in MNOS integrated circuits using an
improved voltage multiplier technique,” IEEE Journal of solid-state circuits, vol. 11,
no. 3, pp. 374–378, 1976.

[83] G. G. Galloway, “Addressable underground marker,” May 14 2002, US Patent

6,388,575.

168

[84] S. Hemour, Y. Zhao, C. H. P. Lorenz, D. Houssameddine, Y. Gui, C.-M. Hu, and
K. Wu, “Towards low-power high-efficiency RF and microwave energy harvesting,”
IEEE transactions on microwave theory and techniques, vol. 62, no. 4, pp. 965–976,
2014.

[85] A. P. Sample, D. J. Yeager, P. S. Powledge, and J. R. Smith, “Design of a passively-
powered, programmable sensing platform for UHF RFID systems,” in IEEE Interna-
tional Conference on RFID.

IEEE, 2007, pp. 149–156.

[86] R. Barnett, G. Balachandran, S. Lazar, B. Kramer, G. Konnail, S. Rajasekhar, and
V. Drobny, “A passive UHF RFID transponder for EPC Gen 2 with-14dBm sensitivity
in 0.13 µm CMOS,” in IEEE International Solid-State Circuits Conference, (ISSCC).
IEEE, 2007, pp. 582–623.

[87] V. Palazzi, F. Alimenti, M. Virili, C. Mariotti, G. Orecchini, P. Mezzanotte, and
L. Roselli, “A novel compact harmonic RFID sensor in paper substrate based on a
variable attenuator and nested antennas,” in IEEE MTT-S International Microwave
Symposium (IMS).

IEEE, 2016, pp. 1–4.

[88] J.-W. Lee and B. Lee, “A long-range UHF-band passive RFID tag IC based on high-
Q design approach,” IEEE Transactions on industrial electronics, vol. 56, no. 7, pp.
2308–2316, 2009.

[89] “Monza

4

RFID

IC,”

https://support.impinj.com/hc/en-us/articles/

202756908-Monza-4-RFID-Tag-Chip-Datasheet, [Online; accessed 08-October-2018].

[90] S. Gollakota, V. Talla, B. Kellogg, B. Ransford, S. Naderiparizi, and J. R. Smith,
“Power transmission using wireless communication signals,” Aug. 13 2019, uS Patent
10,383,126.

[91] J. R. Smith, K. P. Fishkin, B. Jiang, A. Mamishev, M. Philipose, A. D. Rea, S. Roy,
and K. Sundara-Rajan, “RFID-based techniques for human-activity detection,” Com-
munications of the ACM, vol. 48, no. 9, pp. 39–44, 2005.

[92] A. P. Sample, D. J. Yeager, P. S. Powledge, and J. R. Smith, “Design of a passively-
powered, programmable sensing platform for UHF RFID systems,” in 2007 IEEE in-
ternational Conference on RFID.

IEEE, 2007, pp. 149–156.

[93] D. J. Yeager, J. Holleman, R. Prasad, J. R. Smith, and B. P. Otis, “Neuralwisp: A
wirelessly powered neural interface with 1-m range,” IEEE Transactions on Biomedical
Circuits and Systems, vol. 3, no. 6, pp. 379–387, 2009.

[94] N. Xiao, W. Yu, and X. Han, “Wearable heart rate monitoring intelligent sports

bracelet based on Internet of things,” Measurement, vol. 164, p. 108102, 2020.

169

[95] M. Chung, G. Fortunato, and N. Radacsi, “Wearable flexible sweat sensors for health-
care monitoring: a review,” Journal of the Royal Society Interface, vol. 16, no. 159, p.
20190217, 2019.

[96] P. A. Hammond, D. Ali, and D. R. Cumming, “A system-on-chip digital pH meter for
use in a wireless diagnostic capsule,” IEEE Transactions on Biomedical Engineering,
vol. 52, no. 4, pp. 687–694, 2005.

[97] J. Kim, S. Imani, W. R. de Araujo, J. Warchall, G. Vald´es-Ram´ırez, T. R. Paix˜ao,
P. P. Mercier, and J. Wang, “Wearable salivary uric acid mouthguard biosensor with
integrated wireless electronics,” Biosensors and Bioelectronics, vol. 74, pp. 1061–1068,
2015.

[98] S. Baliga, S. Muglikar, and R. Kale, “Salivary pH: A diagnostic biomarker,” Journal

of Indian Society of Periodontology, vol. 17, no. 4, p. 461, 2013.

[99] M. Fiyaz, A. Ramesh, K. Ramalingam, B. Thomas, S. Shetty, and P. Prakash, “Asso-
ciation of salivary calcium, phosphate, pH and flow rate on oral health: A study on 90
subjects,” Journal of Indian Society of Periodontology, vol. 17, no. 4, p. 454, 2013.

[100] K. Rajesh, S. H. Zareena, and M. A. Kumar, “Assessment of salivary calcium, phos-
phate, magnesium, pH, and flow rate in healthy subjects, periodontitis, and dental
caries,” Contemporary clinical dentistry, vol. 6, no. 4, p. 461, 2015.

[101] H. Cao, V. Landge, U. Tata, Y.-S. Seo, S. Rao, S.-J. Tang, H. F. Tibbals, S. Spechler,
and J.-C. Chiao, “An implantable, batteryless, and wireless capsule with integrated
impedance and pH sensors for gastroesophageal reflux monitoring,” IEEE Transactions
on Biomedical Engineering, vol. 59, no. 11, pp. 3131–3139, 2012.

[102] S. Bhadra, G. E. Bridges, D. J. Thomson, and M. S. Freund, “Electrode potential-
based coupled coil sensor for remote pH monitoring,” IEEE Sensors Journal, vol. 11,
no. 11, pp. 2813–2819, 2011.

[103] W.-D. Huang, S. Deb, Y.-S. Seo, S. Rao, M. Chiao, and J. Chiao, “A passive radio-
frequency pH-sensing tag for wireless food-quality monitoring,” IEEE Sensors Journal,
vol. 12, no. 3, pp. 487–495, 2012.

[104] B. E. Horton, S. Schweitzer, A. J. DeRouin, and K. G. Ong, “A varactor-based, induc-
tively coupled wireless pH sensor,” IEEE Sensors Journal, vol. 11, no. 4, pp. 1061–1066,
2011.

[105] S. Bhadra, D. J. Thomson, and G. E. Bridges, “Monitoring acidic and basic volatile
concentration using a pH-electrode based wireless passive sensor,” Sensors and Actu-
ators B: Chemical, vol. 209, pp. 803–810, 2015.

170

[106] T. Litovitz, N. Whitaker, L. Clark et al., “Preventing battery ingestions: an analysis

of 8648 cases,” Pediatrics, vol. 125, no. 6, p. 1178, 2010.

[107] P. Gou, N. D. Kraut, I. M. Feigel, H. Bai, G. J. Morgan, Y. Chen, Y. Tang, K. Bocan,
J. Stachel, L. Berger et al., “Carbon nanotube chemiresistor for wireless pH sensing,”
Scientific reports, vol. 4, p. 4468, 2014.

[108] S. Mondal, K. Wijewardena, S. Karuppuswami, N. Kriti, D. Kumar, and P. Chahal,
“Blockchain inspired RFID based information architecture for food supply chain,”
IEEE Internet of Things Journal, vol. 6, no. 3, pp. 5803–5813, 2019.

[109] P. Hashemi, P. L. Walsh, T. S. Guillot, J. Gras-Najjar, P. Takmakov, F. T. Crews, and
R. M. Wightman, “Chronically implanted, nafion-coated Ag/AgCl reference electrodes
for neurochemical applications,” ACS chemical neuroscience, vol. 2, no. 11, pp. 658–
666, 2011.

[110] H. Suzuki, A. Hiratsuka, S. Sasaki, and I. Karube, “Problems associated with the
thin-film Ag/AgCl reference electrode and a novel structure with improved durability,”
Sensors and Actuators B: Chemical, vol. 46, no. 2, pp. 104–113, 1998.

[111] Y. Devrim, S. Erkan, N. Bac, and I. Eroglu, “Improvement of PEMFC performance
with Nafion/inorganic nanocomposite membrane electrode assembly prepared by ul-
trasonic coating technique,” International Journal of Hydrogen Energy, vol. 37, no. 21,
pp. 16 748–16 758, 2012.

[112] T. Instruments, 2015, ADS7040 ultra-low power, ultra-small size, 8-bit, 1-MSPS, SAR

ADC. [Online]. Available: http://www.ti.com/lit/ds/sbas676c/sbas676c.pdf

[113] S. Mondal, S. Karuppuswami, R. Steinhorst, and P. Chahal, “A wearable passive
pH sensor for health monitoring,” in 2019 IEEE 69th Electronic Components and
Technology Conference (ECTC).

IEEE, 2019, pp. 1240–1245.

[114] F. Kong, C. Qi, H. Lee, G. D. Durgin, and M. Ghovanloo, “Antennas for intraoral
Tongue Drive System at 2.4 GHz: design, characterization, and comparison,” IEEE
Transactions on Microwave Theory and Techniques, vol. 66, no. 5, pp. 2546–2555,
2018.

[115] D. Ma, C. Mason, and S. S. Ghoreishizadeh, “A wireless system for continuous in-
mouth pH monitoring,” in 2017 IEEE Biomedical Circuits and Systems Conference
(BioCAS).

IEEE, 2017, pp. 1–4.

[116] F. Xu, G. Yan, K. Zhao, L. Lu, J. Gao, and G. Liu, “A wireless capsule system with
ASIC for monitoring the physiological signals of the human gastrointestinal tract,”
IEEE transactions on biomedical circuits and systems, vol. 8, no. 6, pp. 871–880, 2014.

171

[117] A. Lahtela, M. Hassinen, and V. Jylha, “RFID and NFC in healthcare: Safety of hos-
pitals medication care,” in 2008 Second International Conference on Pervasive Com-
puting Technologies for Healthcare.

IEEE, 2008, pp. 241–244.

[118] S. Mondal, S. Karuppuswami, and P. Chahal, “ID integrated batteryless wireless digital

pH sensor,” IEEE Sensors Journal, vol. 19, no. 24, pp. 12 079–12 086, 2019.

[119] A. P. Sample, D. J. Yeager, P. S. Powledge, A. V. Mamishev, and J. R. Smith, “Design
of an RFID-based battery-free programmable sensing platform,” IEEE transactions on
instrumentation and measurement, vol. 57, no. 11, pp. 2608–2615, 2008.

[120] A. C. Cabezas, R. G. Medina, N. M. Pe˜na, and M. A. Labrador, “Low energy and
low latency in wireless sensor networks,” in 2009 IEEE International Conference on
Communications.

IEEE, 2009, pp. 1–5.

[121] S. Hemour, Y. Zhao, C. H. P. Lorenz, D. Houssameddine, Y. Gui, C.-M. Hu, and
K. Wu, “Towards low-power high-efficiency RF and microwave energy harvesting,”
IEEE Transactions on Microwave Theory and Techniques, vol. 62, no. 4, pp. 965–976,
2014.

[122] S. Mandal and R. Sarpeshkar, “Low-power CMOS rectifier design for RFID applica-
tions,” IEEE Transactions on Circuits and Systems I: Regular Papers, vol. 54, no. 6,
pp. 1177–1188, 2007.

[123] C. Lauterbach, W. Weber, and D. Romer, “Charge sharing concept and new clock-
ing scheme for power efficiency and electromagnetic emission improvement of boosted
charge pumps,” IEEE Journal of solid-state circuits, vol. 35, no. 5, pp. 719–723, 2000.

[124] X. Liu, L. Huang, K. Ravichandran, and E. S´anchez-Sinencio, “A highly efficient
reconfigurable charge pump energy harvester with wide harvesting range and two-
dimensional MPPT for Internet of Things,” IEEE Journal of Solid-State Circuits,
vol. 51, no. 5, pp. 1302–1312, 2016.

[125] G. Papotto, F. Carrara, and G. Palmisano, “A 90-nm CMOS threshold-compensated
RF energy harvester,” IEEE Journal of solid-state circuits, vol. 46, no. 9, pp. 1985–
1997, 2011.

[126] Z. Luo, M.-D. Ker, W.-H. Cheng, and T.-Y. Yen, “Regulated charge pump with new
clocking scheme for smoothing the charging current in low voltage CMOS process,”
IEEE Transactions on Circuits and Systems I: Regular Papers, vol. 64, no. 3, pp.
528–536, 2016.

[127] A. Chasin, V. Volskiy, M. Libois, K. Myny, M. Nag, M. Rockel´e, G. A. Vandenbosch,
J. Genoe, G. Gielen, and P. Heremans, “An integrated a-IGZO UHF energy harvester

172

for passive RFID tags,” IEEE Transactions on Electron Devices, vol. 61, no. 9, pp.
3289–3295, 2014.

[128] T. Tanzawa and T. Tanaka, “A dynamic analysis of the Dickson charge pump circuit,”

IEEE Journal of solid-state circuits, vol. 32, no. 8, pp. 1231–1240, 1997.

[129] A. Ballo, A. D. Grasso, G. Giustolisi, and G. Palumbo, “Optimized charge pump with
clock booster for reduced rise time or silicon area,” IEEE Transactions on Circuits and
Systems II: Express Briefs, vol. 66, no. 12, pp. 1977–1981, 2019.

[130] S. A. Maas, “Microwave mixers,” ah, 1986.

[131] J.-h. Ou, S. Y. Zheng, A. S. Andrenko, Y. Li, and H.-Z. Tan, “Novel time-domain
Schottky diode modeling for microwave rectifier designs,” IEEE Transactions on Cir-
cuits and Systems I: Regular Papers, vol. 65, no. 4, pp. 1234–1244, 2017.

[132] Q. Chen, X. Chen, H. Cai, and F. Chen, “Schottky diode large-signal equivalent-
circuit parameters extraction for high-efficiency microwave rectifying circuit design,”
IEEE Transactions on Circuits and Systems II: Express Briefs, 2020.

[133] C. A. Costa, M. Dheans, P. Baltus, and H. Gao, “Characterization of damper-to-
damper wireless channel in small cars,” in 2018 IEEE MTT-S International Wireless
Symposium (IWS).

IEEE, 2018, pp. 1–3.

[134] T. R. Rao, D. Balachander, P. Sathish, and N. Tiwari, “Intra-vehicular RF propagation
measurements at UHF for Wireless Sensor Networks,” in 2012 International Conference
on Recent Advances in Computing and Software Systems.
IEEE, 2012, pp. 214–218.

[135] T. ElBatt, C. Saraydar, M. Ames, and T. Talty, “Potential for intra-vehicle wireless
IEEE, 2006, pp.

automotive sensor networks,” in 2006 IEEE Sarnoff Symposium.
1–4.

[136] J.-R. Lin, T. Talty, and O. K. Tonguz, “On the potential of bluetooth low energy
technology for vehicular applications,” IEEE Communications Magazine, vol. 53, no. 1,
pp. 267–275, 2015.

[137] S. Karuppuswami, S. Mondal, D. Kumar, and P. Chahal, “RFID Coupled Passive Dig-
ital Ammonia Sensor for Quality Control of Packaged Food,” IEEE Sensors Journal,
vol. 20, no. 9, pp. 4679–4687, 2020.

[138] J. S. McLean, “A re-examination of the fundamental limits on the radiation Q of
electrically small antennas,” IEEE Transactions on antennas and propagation, vol. 44,
no. 5, p. 672, 1996.

173

[139] W. Geyi, “Physical limitations of antenna,” IEEE Transactions on Antennas and Prop-

agation, vol. 51, no. 8, pp. 2116–2123, 2003.

[140] M. Villegas, J.-L. Polleux, X. Begaud, M. Pigeon, C. Delaveaud, L. Rudant, and
K. Belmkaddem, “Miniature directive antennas,” International Journal of Microwave
and Wireless Technologies, vol. 6, no. 1, p. 45, 2014.

[141] R. M. Fano, “Theoretical

limitations on the broadband matching of arbitrary

impedances,” Journal of the Franklin Institute, vol. 249, no. 1, pp. 57–83, 1950.

[142] R. F. Harrington, “Effect of antenna size on gain, bandwidth, and efficiency,” J. Res.

Nat. Bur. Stand, vol. 64, no. 1, pp. 1–12, 1960.

[143] P. J. Chahal and S. Mondal, “Harmonic RFID Tag-Reader System For Long Range

Sensing Identification And Security,” Nov. 28 2019, uS Patent App. 16/420,711.

174