<%BANNER%>

Electronic Pills for Medication Compliance Monitoring

Permanent Link: http://ufdc.ufl.edu/UFE0041596/00001

Material Information

Title: Electronic Pills for Medication Compliance Monitoring
Physical Description: 1 online resource (132 p.)
Language: english
Creator: Yu, Hong
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: antenna, capsule, compliance, electronic, integrated
Electrical and Computer Engineering -- Dissertations, Academic -- UF
Genre: Electrical and Computer Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Medication compliance is the degree to which a medication is taken according to a prescribed treatment and is usually measured in terms of percent of doses taken over a given interval. It is estimated 125,000 people die of treatable ailments because of poor adherence. A tenth of hospital admissions are associated with noncompliance at a healthcare services expense of approximately $15.2 billion annually. Motivated in part by the need for new alternatives for medication compliance, the primary objective of this dissertation is to investigate the feasibility of small electronic transponders as potential means for low cost and reliable detection schemes of orally ingestible electronic pills (e-pills). With continued advances in RF biotelemetry, it is envisioned that an external monitoring point-of-care device or a body-worn electronic sensor can be used to detect the presence of the pill in the stomach or GI-tract after ingestion. The proposed medication compliance device comprises of an electronic microchip and an antenna inlay placed on the surface of a standard 0 or 00 sized capsule. These antennas can be made of conductive bio-compatible coatings by incorporating a metal, which can dissolve, such as silver, under a temporary protective layer such as polyglycolic acid, or by incorporating particles that are non-toxic by virtue of being non-absorbable. Thus, the substrate for the antenna and the electronic device can be directly applied onto the surface of the drug delivery device without affecting the volume reserved for the medication. The electronic pill in this system-on-acapsule can be potentially manufactured in large scales using a thin, mechanically compliant and small antenna pill inlay under a biocompatible protective coating that is excreted via the GI tract. This work begins with an overview of medication compliance and current techniques of measuring medication adherence. A brief review of potential alternatives based on electrical identification technologies is presented. We devote a large portion of this thesis to understanding the transmission characteristics of small electronic radiating elements inside the human body. Initial feasibility studies are carried out using small coil antennas inside human phantom solutions that mimic the electrical properties of the human body. Studies are supported by extensive simulations of radiating elements inside the human body using the finite difference time domain (FDTD) technique. The radiated field intensity over several US Federal Communications Commission (FCC) telemetry bands was characterized to determine optimal UHF transmission frequencies and suitable locations for external reader placement. This dissertation also investigates a variety of capsule antenna designs, direct on-capsule printing methods and measurement infrastructure to experimentally characterize the antenna pills. This work introduces a novel radio frequency identification design, called ?e-Burst?, which enables detection of passive electronic pills inside the human body. The critical insight of the proposed RF tagging architecture is the use of an asymmetric powering and communication scheme to circumvent problems associated with signal attenuation inside the human body and poor radiation efficiency of electrically small antennas. Since the power levels required to activate a tag are orders of magnitude larger than what is detectable externally, the tag employs a galvanic coupling method to energize the microchip at low frequencies where in-body attenuation is lowest, and generates RF bursts at higher frequencies where the efficiency and fractional bandwidth of a capsule-sized antenna is higher. The operation of the tagging system can be treated as a two-step energy conversion process where low frequency energy transferred from the reader device is first converted into DC to energize the tagging device, and the stored DC energy on the tagging device is then converted to the UHF RF bursts to show the presence of the tagging device. Critical design parameters for overall system level analysis incorporating experimentally characterized channels using human cadaver are presented. Based on these studies, a proof-of-concept asymmetric RF tagging device was fabricated using 130nm CMOS technology and validated inside phantom solutions.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Hong Yu.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Bashirullah, Rizwan.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-04-30

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0041596:00001

Permanent Link: http://ufdc.ufl.edu/UFE0041596/00001

Material Information

Title: Electronic Pills for Medication Compliance Monitoring
Physical Description: 1 online resource (132 p.)
Language: english
Creator: Yu, Hong
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: antenna, capsule, compliance, electronic, integrated
Electrical and Computer Engineering -- Dissertations, Academic -- UF
Genre: Electrical and Computer Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Medication compliance is the degree to which a medication is taken according to a prescribed treatment and is usually measured in terms of percent of doses taken over a given interval. It is estimated 125,000 people die of treatable ailments because of poor adherence. A tenth of hospital admissions are associated with noncompliance at a healthcare services expense of approximately $15.2 billion annually. Motivated in part by the need for new alternatives for medication compliance, the primary objective of this dissertation is to investigate the feasibility of small electronic transponders as potential means for low cost and reliable detection schemes of orally ingestible electronic pills (e-pills). With continued advances in RF biotelemetry, it is envisioned that an external monitoring point-of-care device or a body-worn electronic sensor can be used to detect the presence of the pill in the stomach or GI-tract after ingestion. The proposed medication compliance device comprises of an electronic microchip and an antenna inlay placed on the surface of a standard 0 or 00 sized capsule. These antennas can be made of conductive bio-compatible coatings by incorporating a metal, which can dissolve, such as silver, under a temporary protective layer such as polyglycolic acid, or by incorporating particles that are non-toxic by virtue of being non-absorbable. Thus, the substrate for the antenna and the electronic device can be directly applied onto the surface of the drug delivery device without affecting the volume reserved for the medication. The electronic pill in this system-on-acapsule can be potentially manufactured in large scales using a thin, mechanically compliant and small antenna pill inlay under a biocompatible protective coating that is excreted via the GI tract. This work begins with an overview of medication compliance and current techniques of measuring medication adherence. A brief review of potential alternatives based on electrical identification technologies is presented. We devote a large portion of this thesis to understanding the transmission characteristics of small electronic radiating elements inside the human body. Initial feasibility studies are carried out using small coil antennas inside human phantom solutions that mimic the electrical properties of the human body. Studies are supported by extensive simulations of radiating elements inside the human body using the finite difference time domain (FDTD) technique. The radiated field intensity over several US Federal Communications Commission (FCC) telemetry bands was characterized to determine optimal UHF transmission frequencies and suitable locations for external reader placement. This dissertation also investigates a variety of capsule antenna designs, direct on-capsule printing methods and measurement infrastructure to experimentally characterize the antenna pills. This work introduces a novel radio frequency identification design, called ?e-Burst?, which enables detection of passive electronic pills inside the human body. The critical insight of the proposed RF tagging architecture is the use of an asymmetric powering and communication scheme to circumvent problems associated with signal attenuation inside the human body and poor radiation efficiency of electrically small antennas. Since the power levels required to activate a tag are orders of magnitude larger than what is detectable externally, the tag employs a galvanic coupling method to energize the microchip at low frequencies where in-body attenuation is lowest, and generates RF bursts at higher frequencies where the efficiency and fractional bandwidth of a capsule-sized antenna is higher. The operation of the tagging system can be treated as a two-step energy conversion process where low frequency energy transferred from the reader device is first converted into DC to energize the tagging device, and the stored DC energy on the tagging device is then converted to the UHF RF bursts to show the presence of the tagging device. Critical design parameters for overall system level analysis incorporating experimentally characterized channels using human cadaver are presented. Based on these studies, a proof-of-concept asymmetric RF tagging device was fabricated using 130nm CMOS technology and validated inside phantom solutions.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Hong Yu.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Bashirullah, Rizwan.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-04-30

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0041596:00001


This item has the following downloads:


Full Text

PAGE 1

1 ELECTRONIC PILLS FOR MEDICATION COMPLIANC E MONITORING By HONG YU A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Hong Yu

PAGE 3

3 Dedicated to my parents and my wife

PAGE 4

4 ACKNOWLEDGMENTS I would like to first thank my advisor Dr. Rizwan Bashirullah for giving me this splendid opportunity to work towards a Ph.D under his supervision. His constant guidance and encouragement provided me a clear path for my study, and I have truly enjoyed working with him over the years acquiring technical knowledge as well as other soft skills. I would also like to thank Dr. William Eisenstadt, Dr. Jenshan Lin and Dr. Christopher Batich for their valuable time and for being on my Ph.D committee. I feel very fortunate to have worked together with all my colleagues, especially Chun -ming Tang, Chung -ching Peng and Zhiming Xiao in the ICR group, and David M. Peterson of the RF Lab, Brain Institute, UF, whose helpful discussions, suggestions and friendship have greatly improved the quality of my work. Wit hout them, the completion of this project would not have been possible. Finally, I would like to acknowledge the love and continuous encouragement from my parents and my wife, Juan Yu, to whom I dedicate this work.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................................... 4 LIST OF TABLES ................................................................................................................................ 8 LIST OF FIGURES .............................................................................................................................. 9 ABSTRACT ........................................................................................................................................ 12 CHAPTER 1 INTRODUCTION ....................................................................................................................... 15 1.1 Overview of Medication Compliance .................................................................................. 15 1.2 Significance of Medication Compliance Monitoring ......................................................... 15 1.3 Current Techniques in Measuring Compliance ................................................................... 18 1.3.1 Direct Methods ......................................................................................................... 19 1.3.2 In-direct Methods ..................................................................................................... 19 1.4 Revi ew of Electronic Detection Methods ............................................................................ 20 1.4.1 Magnetic Methods ................................................................................................... 21 1.4.2 Low Frequency Resonant Tags ............................................................................... 21 1.4.3 RFID Technologies .................................................................................................. 24 1.4.3.1 Near -field RFID ....................................................................................... 24 1.4.3.2 Ultra high f requency RFID ...................................................................... 25 1.5 Towards an Electronic Pill ................................................................................................... 27 1.6 Dissertation Organization ..................................................................................................... 27 2 FEASIBILITY S TUDY OF E PILL DETECTION ................................................................. 30 2.1 Introduction ........................................................................................................................... 30 2.2 Biocompatible Inks ............................................................................................................... 32 2.2.1 Ink Preparation ......................................................................................................... 32 2.2.2 Biocompatibility and Silver Toxicity ..................................................................... 33 2.3. Modeling of A ntennas inside the Human Body ................................................................. 36 2.3.1 Computational Method ............................................................................................ 36 2.3.2 Simulation Setup ...................................................................................................... 37 2. 3.3 Radiation in FCC Regulated Frequency B ands ..................................................... 38 2.4 Phantoms Experiment ........................................................................................................... 40 2.4.1 Ph antom Preparation ................................................................................................ 40 2.4.2 Insertion Loss Measurements .................................................................................. 41 2.5 Conclusions ........................................................................................................................... 43

PAGE 6

6 3 CAPSULE ANTENNA DESIGN AND MEASUREMENT ................................................... 45 3.1 Introduction ........................................................................................................................... 45 3.2 Capsule Antenna: Materials and Methods ........................................................................... 46 3.2.1 Capsules .................................................................................................................... 47 3.2.2 Printing Methods ...................................................................................................... 48 3.2.2.1 In k jet printing .......................................................................................... 48 3.2.2.2 Screen printing ......................................................................................... 49 3.2.2.3 Pad printing............................................................................................... 50 3.3 Screen P rinted Capsule Dipole A ntenna ............................................................................. 51 3.3.1 Capsule Antenna ...................................................................................................... 51 3.3.2 Antenna Measurement Setup Validation ................................................................ 52 3.3.3 Zig-Z ag Capsule Antenna Measurement ................................................................ 54 3.3.4 Radiation inside the Human Body .......................................................................... 55 3.3.4.1 Model setup .............................................................................................. 56 3.3.4.2 Near -field analysis ................................................................................... 57 3.3. 4.3 Far -field patterns ...................................................................................... 59 3.4 Screen Printed Capsule Antennas for RFID ....................................................................... 60 3.5. Pad -Printed Capsule Antenna ............................................................................................. 65 3.5.1 Asymmetric Antenna Design and Measurements .................................................. 66 3.5.2 Antenna Validation Using Existing RFID Technology ......................................... 71 3.5.3 Pill -to -Pill Communication ..................................................................................... 73 3.6 Conclusions ........................................................................................................................... 75 4 CHIP DESIGN FOR E -PILL SYSTEM .................................................................................... 77 4.1 Introduction ........................................................................................................................... 77 4.2 Power and Communication for Passive In-Body Microsystems Using the Human Body as a Transmission Medium ..................................................................................... 78 4.2.1 An Electronic Medication Compliance Device -Associated Challenges .............. 79 4.2.2 The Human Body as a Powering and Communication Medium .......................... 81 4.2.3 Low Frequency Human B ody Channel Characterization ...................................... 83 4.2.4 Safety and Regulations ............................................................................................ 86 4.3 InB ody Electronic Burst (E Burst) RF Transponder: System Optimization ................... 87 4.3.1 System Modeling ..................................................................................................... 89 4.3.2 Tag Activation: Required Reader Output Voltage ................................................ 94 4.4 Circuit Implementation ....................................................................................................... 100 4.4.1 System Architecture ............................................................................................... 100 4.4.2 RF DC Multiplier .................................................................................................. 101 4.4.3 Power Conditioning ............................................................................................... 102 4.4.4 Modulator ............................................................................................................... 104 4.4.5 Demodulator ........................................................................................................... 108 4.5 Experimental Results .......................................................................................................... 110 4.5.1 Test Bench Results ................................................................................................. 110 4.5.2 Antenna Fabrication and Tag Assembly ............................................................. 114 4.5.3 Characterization of the Assembled Tagging Device ........................................... 115 4.6 Conclusions ......................................................................................................................... 118

PAGE 7

7 5 SUMMARY AND FUTURE WORKS ................................................................................... 119 5.1 Summary of this Dissertation ............................................................................................. 119 5.2 Suggested Future Works ..................................................................................................... 121 APPENDIX: MATLAB CODE FOR MULT IPLIER DESIGN ..................................................... 122 LIST OF REFERENCES ................................................................................................................. 124 BIOGRAPHICAL SKETCH ........................................................................................................... 132

PAGE 8

8 LIST OF TABLES Table page 1 1 List of direct and indirect methods of medication compliance ........................................... 18 3 1 Summary of different types of printed capsule antenna. ..................................................... 76 4 1 RMS induced and contact current limits for continuous sinusoidal waveforms ................ 87 4 2 Summary of the measured prototype e -pill performance .................................................. 117

PAGE 9

9 LIST OF FIGURES Figure page 1 1 The Pulse -Listen methodology of detecting chipless tags. ............................................. 22 1 2 Illustration of UHF backscattering RFID tagging system. .................................................. 26 2 1 Passive miniaturized electroni c chip for medication compliance ....................................... 30 2 2 Simulated electrical small antennas and human body model. ............................................. 38 2 3 Simulated results of electrical field distribution for electrical small antennas. .................. 39 2 4 Measured S11 when narrow band probe is loaded with the phantom solution and human body within the MICS band. ..................................................................................... 41 2 5 Experiment setup of measuring the attenuation caused by the phantom solution at MICS band. ............................................................................................................................. 42 2 6 Measured scattering parameters (S11, S21) using the se tup in Figure 2 5. ....................... 43 3 1 Multi turn coils printed on gelatin capsule using modified inkjet printer. ......................... 49 3 2 Screen printed Zigzag dipole antenna, and its SEM picture. .............................................. 51 3 3 The measurement setup for the electronic pill. .................................................................... 53 3 4 HFSS simulation of effect of metal holding structure to the antenna impedance. ............. 54 3 5 Capsule antenna ...................................................................................................................... 55 3 6 Simulation setup for zig -zag capsule antenna radiating from inside of the human body. ........................................................................................................................................ 57 3 7 Cross -section of human body model torso indicating the normalized radiation field intensity contours for the WMTS 608614MHz, ISM 902928MHz, ISM 2.4 2.45GH z, and ISM 5.7255.825GHz FCC frequency bands ............................................... 58 3 8 Far field radiation patterns for the tuned zigzag pill antenna inside of the human body. ........................................................................................................................................ 60 3 9 Two inductive capsule antennas designed to be compatible with the frontier RFID technology. .............................................................................................................................. 61 3 10 The measured and simulated (XFDTD) input impedance of the inductive capsule antennas. ................................................................................................................................. 63

PAGE 10

10 3 11 The simulated near field electrical field distribution inside and outside of human body. ........................................................................................................................................ 64 3 12 The simulated far field radiation patterns of the two inductive antennas. .......................... 65 3 13 Asymmetric electrical loop antenna designed to maximize the gain of a capsule sized antenna. ................................................................................................................................... 67 3 14 Simulated and measured input impedance of the proposed asymmetric electrical loop antenna. ................................................................................................................................... 68 3 15 Simulated open-circuit RCS of the self resonant antenna for different illuminating angles. ..................................................................................................................................... 69 3 16 Attenuation contours around and inside of human body for the electrical asymmetric loop antenna placed inside of the GI tract ............................................................................ 70 3 17 Validation of the asymmetric loop antenna using existing te chnology from Alien system. .................................................................................................................................... 72 3 18 Measurement setup for pill to -pill communication. ............................................................ 73 3 19 The measured coupling gain of the capsule antenna pair for pill to -pill communication. ...................................................................................................................... 74 4 1 Illustration of transmission through the human body .......................................................... 82 4 2 Measured channel loss from various locations on a cadaver to the inside of stomach and in phantom solution. ........................................................................................................ 84 4 3 Symbolic representation of the tagging device and its two operating phases. ................... 88 4 4 Calculated charging current Ic for different burst rate, assuming Eb/N0=30dB, = 50dB, Nf = 20dB, Vhigh = 1.2V, and Vlow = 0.6V. ................................................................ 92 4 5 Required storage capacitance Cs as a function of D, assuming Eb/N0=30dB, = 50dB, Nf = 20dB, Vhigh = 1.2V, and Vlow = 0.6V. ................................................................ 93 4 6 Required burst power Pburst from the tagging device for various period T, to achieve 30dB Eb/N0, assuming = 50dB, Nf = 20dB, Vhigh = 1.2V, and Vlow = 0.6V ............ 93 4 7 Finding the minimum VTX by incorporating the low frequency channel to the multiplier design. .................................................................................................................... 94 4 8 Design the multiplier using charge equilibrium based on the behavior of a single diode device. ........................................................................................................................... 95

PAGE 11

11 4 9 Calculated multiplier parameters based on charge equilibrium for four input amplitude values ..................................................................................................................... 98 4 10 Calculated requirement for external voltage to activate the ingested tagging device. ....... 99 4 11 The system level block diagram of the asymmetric tagging IC. ....................................... 100 4 12 Schematic of the RF DC multiplier optimized for 13.56MHz. ......................................... 1 02 4 13 The schematic of the power conditioning circuits. ............................................................ 103 4 14 Schematic of the RF modulator composed of the DCO and the class D PA. .................. 105 4 15 Class D PA optimization. .................................................................................................... 107 4 16 The demodulator and the clock and data recovery (CDR) circuits. .................................. 109 4 17 Micrograph of the tagging IC. ............................................................................................. 111 4 18 The measured sensitivity of the fabricated tagging IC. ..................................................... 111 4 19 Measured time -domain response of the demodulator and CDR. ...................................... 112 4 20 Tunability of the DCO. ........................................................................................................ 112 4 21 Measured DCO phase noise at 915.6MHz with 1.2 supply voltage. ................................ 113 4 22 Small capsule antenna designed for maximizing the output power .................................. 114 4 23 Picture of the tagging device assembly. .............................................................................. 115 4 24 Characterization of Pburst, the radiated power from the assembled tag ............................. 116 4 25 Prototype e pill verification using the phantom solution. ................................................. 117

PAGE 12

12 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ELECTRONIC PILLS FOR MEDICATION COMPLIANC E MONITORING By Hong Yu May 2010 Chair: Rizwan Bashirullah Major: Electrical and Computer Engineering Medication compliance is the degree to which a medication is taken according to a prescribed treatment and is usually measured in terms of percent of doses taken over a given interval. It is estimated 125,000 people die of treatable ailments because of poo r adherence. A tenth of hospital admissions are associated with noncompliance at a healthcare services expense of approximately $15.2 billion annually. Motivated in part by the need for new alternatives for measuring medication compliance, the primary obje ctive of this dissertation is to investigate the feasibility of small electronic transponders as potential means for low cost and reliable detection schemes of orally ingestible electronic pills (e pills). With continued advances in RF biotelemetry, it is envisioned that an external monitoring point -of -care device or a body -worn electronic sensor can be used to detect the presence of the pill in the stomach or GI -tract after ingestion. The proposed medication compliance device comprises o f an electronic mic rochip and an antenna inlay placed on the surface of a standard 0 or 00 sized capsule These antennas can be made of conductive bio-compatible coatings by incorporating a metal, which can dissolve, such as silver, under a temporary protective layer such as polyglycolic acid, or by incorporating particles that are non toxic by virtue of being nonabsorbable. Thus, the substrate for the antenna

PAGE 13

13 and the electronic device can be directly applied onto the surface of the drug delivery device without affecting the volume reserved for the medication. The electronic pill in this system ona capsule can be potentially manufactured in large scales using a thin, mechanically compliant and small antenna pill inlay under a biocompatible protective coating that is excreted via the GI tract. This work begins with an overview of medication compliance and current techniques of measuring medication adherence. A brief review of potential alternatives based on electrical identification technologies is pr esented. We devote a large portion of this thesis to understanding the transmission characteristics of small electronic radiating elements inside the human body. Initial feasibility studies are carried out using small coil antennas inside human phantom sol utions that mimic the electrical properties of the human body. Studies are supported by extensive simulations of radiating elements inside the human body using the finite difference time domain (FDTD) technique. The radiated field intensity over several US Federal Communications Commission (FCC) telemetry bands was characterized to determine optimal UHF transmission frequencies and suitable locations for external reader placement. This dissertation also investigate s a variety of capsule antenna designs, dir ect on -capsule printing methods and measurement infrastructure to experimentally characterize the antenna pills. This work introduces a novel radio frequency identification transponder design, called e Burst, which enables detection of passive electronic pills inside the human body. The critical insight of the proposed RF tagging architecture is the use of an effective signal coupling methodology and an asymmetric powering and communication scheme to circumvent problems associated with signal attenuation inside the human body and poor radiation efficiency of electrically small antennas. Since the power levels required to activate a tag are orders of magnitude larger than what is detectable externally, the tag employs a galvanic coupling method

PAGE 14

14 to energize the microchip at low frequencies where in -body attenuation is lowest, and generates RF bursts at higher frequencies where the efficiency and fractional bandwidth of a capsule-sized antenna is higher. The operation of the tagging system can be treated as a two -step energy conversion process where low frequency energy transferred from the reader device is first converted into DC to energize the tagging device, and the stored DC energy on the tagging device is then converted to the UHF RF bursts to show the pr esence of the tagging device. Critical design parameters for overall system level analysis incorporating experimentally characterized channels using human cadaver are presented. Based on the se stud ies a proof -of concept asymmetric RF tagging device was fa bricated using 130nm CMOS technology and validated inside phantom solution s.

PAGE 15

15 CHAPTER 1 INTRODUCTION 1.1 Overview of Medication Compliance Medication compliance can be loosely defined as the degree to which a medication is taken according to a prescribed t reatment, and usually it is measured in terms of percent of dose s taken over a period of time. Medication compliance monitoring is critical in at least three different areas: pharmaceutical clinical trials, geriatrics, and mental health/addiction medicine. In pharmaceutical trials estimates vary, but on average it is likely that 25 50% of the patients are noncompliant to the prescribed regimen [1] [2]. It is estimate d that 125,000 people die because of non-compliance to prescription s every year Almost a tenth of the hospital admissions are associated with noncompliance at a cost of approximately 15.2 billion dollars per year [3]. Psychotics who can be prevented from relapses (by correct medication compliance) can save the healthcare system thousands of dollars per relapse (often enough to pay for the health -care system in preventing only one such relapse). Similarly, the elderly and other high medication groups suffer from poor medication compliance, drug interaction, and improper administration probl ems. In medical development industry, this noncompliance requires large numbers of excess patients to be studied for both efficacy and safety (potentially costing tens to hundred s of millions of extra dollars per trial). Hughes [4] rates the overall cost o f noncompliance at about 100 billion dollar per year, and described it as Americas other drug problem. 1.2 Significance of Medication Compliance Monitoring Over the past three decades the yearly spending on drug development has increased more than 12 t imes in inflation adjusted dollars [5]. A major obstacle t he drug industry faces is in reducing the development times while improving the success rate of a drug trial. Clinical trials consume a major portion of the development time and cost of introducin g a new drug into the

PAGE 16

16 market [ 1], [6 ]. The importance of knowing with certainty the patients compliance to a medication regimen is vital to the outcome of the clinical trial, because without it the resu lt from a trial cannot be accurately interpreted. Despi te the presence of several methods that measure the compliance of patient s to medication, there is no methodology that is considered the gold standard. None of the methods offer s both a qualitative and a quantitative measure of compliance [1][7 ], and thus measuring medication regimen compliance continues to be a major problem. Reports from the Tufts Center for the Study of Drug Development (CSDD) indicate that the average cost to develop a drug is $802 million per drug [5], of which the clinical trial phas e is one of the costliest in drug development. Lowering development costs is not only beneficial for drug companies, but also to health -insurance providers and consumers. The drug industry needs to adapt novel research and development models that stress be tter, faster, and cheaper ways of introducing new drugs to the market. Since the passage of the Prescription Drug User Fee Act in 1992 the average development time for new drugs has dropped from 9.2 years to 6.9 years [5], yet the industry still faces a ma jor obstacle in reducing the development times while improving the success rate of a drug trial. Studies conducted by Tufts CSDD have shown that reducing the development time by half will reduce total costs by 29%. Their analysis also indicates that for a drug developer to reduce development costs by $200 million, the developer must either improve clinical success rate by 30% or cut out -of pocket preclinical costs by 59% Furthermore, the annual cost to the U.S. healthcare system from non-compliance with me dication regimens exceeds $100 billion and pharmaceutical companies lose over $25 billion in revenue from unfilled prescriptions [6][8]. The definition and measurement of adherence varies from study to study, but the quoted definition of compliance is th e extent to which a persons behavior coincides with medical and

PAGE 17

17 health advice [1 ][6] For the purposes of this dissertation the terms adherence and compliance will be used interchangeably, and non-compliant type of behavior will be defined as in [ 1 ], [5 ] [6 ] and [ 9 ]: A delay in beginning and/or the termination of treatment Omissions of one or several doses Errors in the size of the dose taken Inappropriate and irregular timing in administration In a clinical drug trial it is important to kno w the patients compliance to a medication regimen with a high degree of certainty, because without such knowledge the results from a clinical trial cannot be accurately and realistically interpreted. There is no doubt that the outcome of a clinical trial is always correlate d to the medication adherence regimen. For instance, in a beta blocker heart attack trial the death rate was reported at 13.6% in subjects whose compliance was less than 75%, compared to 5.6% in subjects whose compliance was over 75% [10] Furthermore, variations in compliance can alter the number of subjects required to detect a significant difference between treatment and placebo. Knowing the adherence rate is also necessary to examine the dose-response relationship and allow a valid analysis of treatment efficacy. Some trials use a run -in period where prospective subjects are eliminated if they do not meet a pre-determined level of compliance. For example, i n a clinical trial involving low -dosage aspirin it was reported that 11,000 o f 33,000 eligible subjects were exc luded because of low compliance [6], [11] Use of an effective compliance monitor can significantly reduce the cost of drug trials (and thus drug development), increase early detection of non-compliance, possibly increase compliance because subjects know they will be caught, and accurate dismissing of bad data. This will lead to a higher success rate leading to lower costs and less time as described above.

PAGE 18

18 The following points summarize the significance of non-compliance in clinical trials [1 ], [5 ][6 ], [12][13] : $800 million development cost for a new drug. Non -compliance leads to increased drug trial lengths, more subjects, increased development costs, and failure to detect infrequent complications. Poor compli ance records make it difficult to ascertain true outcome of a clinical trial, thus increases the cost of developing a new medication In the broader sense, the following points summarize the significance of non-compliance in general [1 ] [5 ] [6 ], [12][13] : 50% of medication is not taken as prescribed. 10% of hospital admissions are due to patients not taking medication correctly. It is estimated that there are 218,000 deaths annually from not taking medication properly. 100 million Americans are taking medication for chronic conditions, and 50% are not compliant. 1.3 Current Techniques in Measuring Compliance Numerous direct and indirect methods are available for measuring patient adherence to medication regimens, but as documented in a recent JAMA review [2] of medication adherence strategies, none have had a significant effect in determining the validity of compliance [1 ][6 ][14] Direct methods typically provide proof that the patient took the drug, whereas indirect meth ods provide a way to qualitatively measure the adherence. The table below lists both methods. Table 1 1. List of direct and indirect methods of medication compliance Direct Methods Indirect methods Detection of drug or a metabolite in blood or urine Dete ction of a biological marker that is given with the drug Direct observation of patient receiving the medication Self reporting by the patient Pill count Electronic monitoring devices(MEMS cap and metered -dose inhaler) Prescription record review

PAGE 19

19 1.3.1 Direct Methods Direct methods prov ide an answer to the question : H as the patient recently taken a dose of the drug? However, the presence of the drug does not prove compliance, and likewise the absence does not prove non -compliance. Measurements of blood a nd urine drug levels do not depict the complete manner in which a patient has taken the medication, nor do they detect fluctuations in compliance between clinical visits. For instance, if the drug or biological marker has a fairly short half -life examining the drug levels in biological fluids would not detect so called white -coat compliance, in which the patient is non -compliant until shortly before a clinical visit and return s to non-compliant behavior after the clinical appointment. Directly Observed Th erapy (DOT) has been shown to positively impact adherence [6]. Adherence rates of over 80% have been achieved with this novel approach. DOT requires that patients report to a public health facility on a daily basis and that they be observed taking their me dication(s). DOT is usually reserved for diseases that pose a public health risk such as tuberculosis and AIDS. A similar approach has been used for methadone administration. DOT is expensive, and the m ethod is not entirely foolproof; for instance, deliber ate non -compliers can feign swallowing of the medication and then remove it from their mouth when they are not observed. 1.3.2 In -direct Methods Indirect methods provide a more quanti tative measurement of adherence; however patient interviews and questionnaires are generally considered unreliable for accurate assessment of compliance [6]. Gordis et al reported that comparing a urine test to determine whether a dose of penicillin had been taken that morning with patient inte rviews indic ated 69% compliance whereas urine test documented only 33% compliance The accuracy and validity of self -reporting methods is dependent on the interviewers skill and the construction of the questions. The relationship and manner of communicat ion between health care professional s and patient s have

PAGE 20

20 also been found to affect compliance Besides patient interview, the most common method used to determine adherence to medication is counting the number of dosage units that the patient failed to take in between scheduled clinical visits. However, the pill count method does not provide information related to the nature of adherence problem e.g. the pattern of missed doses. A number of electronic devices [6 ][7 ][15] have been created specifically for th e purpose of monitoring medication adherence. These devices record the time and date that a patient obtains a dose of medication by detecting when the pillbox is opened. These devices provide precise data on the manner in which the patient uses the medicat ion, and are far more reliable than self reporting. However, the fundamental drawback with indirect methods is that they cannot provide a qualitative answer to medication compliance, i.e. they cannot corroborate that the patient did indeed took the medicat ion. 1.4 Review of Electronic Detection Methods Biotelemetry is a rapidly growing field tha t may provide the next critical breakthrough in medical monitoring. These devices can be used for tasks such as embedded oxygen monitoring, glucose sensors, fetal mo nitoring, and measurement of hormones such as progesterone to optimize in vitro fertilization. In addition to medical applications, athletes can use biotelemetry to improve training by monitoring the concentration of oxygen or lactic acid in skeletal muscl es. Techniques such as Radio -frequency -Identification ( RFID ) can be adapted to provide biotelemetry by including external sensors into the existing systems. However, RFID was not designed to operate in vivo and the transmission of EM signals from embedded or internal sensors will be hampered by attenuation in human tissue. RF telemetry from within or through the human body needs to be optimized by tailoring the frequency, antenna geometry, sensor design, and power sources. Alternate approaches such as magn etic based sensors are simple to

PAGE 21

21 implement but do not provide a unique signature to distinguish different types of capsules. Below we provide a brief review of different types of electronic detection methods. 1.4.1 Magnetic Methods The use of magnetic pill s in humans is well documented [16] [19] The pills have been mainly used in the research of gastrointestinal motility. Magnetic fields are impervious to materials (apart from iron and a few other materials) it travels through. This would be advantageous i n compliance monitoring, because a magnetic field emanating from inside of a human body will not be distorted due to blood, tissue, and bone all of which distort electric fields. The caveat, however, is the cubic decrease in magnetic field strength with every doubling of distance. Another drawback in magnetic based approaches is that it is difficult to generate a unique signature. Metallic objects such as metal sheet or a cell phone create a response that is simil ar in shape to that of a permanent magnet Their responses, although smaller in strength by roughly an order in magnitude, makes discrimination from the permanent magnet -based tags difficult [20] 1.4.2 Low Frequency R esonant Tags L ow frequency resonant tags are commonly used for electronic article s urveillance (EAS) applications to prevent shoplifting from retail store s or pilferage of books from libraries [21] [23] T hese tags are typically operated within the frequency range from 30 KHz to 300 K Hz or from 3 MHz to 30 MHz and can be implemented using various methods. A magnetic tag is formed by two magnetic strips [ 22][23] : a strip of magnetostrictive ferromagnetic amorphous metal and a strip of a magnetically semi -hard metal which can be used as a biasing magne t to change the signal strength and thus completely deactivate the tag. The amorphous metal is used due to its magneto -elastic coupling ability which means it can convert the coupled magnetic energy into mechanical vibration. The tag is activated by magnetizing the

PAGE 22

22 semi -hard magnet, which helps the amorphous strip generate a much stronger response to the detector; the tag can also easily be deactivated by de -magnetizing the metallic strip, so that the mechanical oscillation from the amorphous metal is too small to be detected by the reader. The magnetic tag operates at the resonant frequency of the amorphous strip, usually in the k Hz range Fig ure 1 1. The Pulse Listen methodology of detecting chip-less tags Low frequency resonant tags can also be implemented using discrete LC components. Most of the commercially available LC tags are made to resonate between 1.75MHz and 9.5MHz, with 8.2MHz being the most popular operating frequency [25]. For some of the tagging systems available in the market, the tag is formed by a discrete capacitor and a planar coil with 5 7 turns These tags can also be de activated by applying a strong electromagnetic f ield at the resonance frequency to induce voltages that exceed the capacitor's breakdown voltage therefore destroying the capacitor and detuning the tag off resonance. Because of its relatively small size and ease of detection at low frequency LC tags ar e more popular than the magnetic tags. EAS Tag 58KHz, 8.2MHz PA Switch Detector Reader bursts Tag response Talk Listen Talk Indicating Tag presence HF RFID Tag

PAGE 23

23 The Talk and -Listen (F igure 1 1) approach is the most popular interrogation technique to detect the low frequency resonant tags. To inquire the presence of the tags, the reader talks first by sending out periodic tonal bursts at th e tag resonant frequency In this phase, oscillating current is induced into the resonant tank, and e lectromagnetic energy is stored within the tag When the reader is quickly shut off and goes into the listening mode, the stored energ y will sustain the tag oscillation for an additional period of time. By detecting this residual oscillation from the tag, the reader can tell that the tag is in its reading range. The Talk and -Listen method is seldom used to detect tags resonated at hig h frequencies since the detection quality of this method depends on the time duration of the tag s residual oscillation, and a tag with longer residual oscillation can be detected more reliably. Because the amplitude of the residual oscillation is damped by a factor eper oscillating cycle [25] in which Q is the quality factor of the resonant tank, long residual oscillation time can only be achieved by forming tags with highQ components. Although c omponents with Q ~ 1000 can be easily implemented at low frequency, it is difficult to obtain such high Q at VHF/UHF bands because of the increased ohmic loss and electromagnetic radiation. Although the low frequency resonant tagging technology has already been widely and reliably used in various applications, there are difficulties to apply it directly for m edication compliance monitoring. For example, the metallic (magnetic) strips used to form the magnetic tags can hardly be made biocompatible, and they are too large to be ingested by the patients. The energy coupling between t he LC resonant tag and the detector relies heavily on the size of the tagging coil, therefore using a coil fit ting the standard capsule, which is used as ingestible vehicle for medi cation, greatly limits the detection range. Furthermore, the low operating

PAGE 24

24 frequency may cause interference problems with some implanted medical device s such as pacemaker s and defibrillator s [24]. 1.4.3 RF ID Technologies Similar to the EAS technology, RFID tags can be applied to or incorporated into a product, animal, or person for the purpose of tracking and identification. In order for the interrogator to distinguish different objects in the detection range at the same time identification codes can b e assigned to different object s Therefore, an integrated circuit (IC) device for storing and processing data (i.e. the identification code) must be included. Based on the operation scheme, RFID tags can be divided into three major categories: the active t ag which is powered by a battery and ca n transmit signals autonomously; the passive tag, which is powered by an external RF source; and the semi active tag, which requires an external source to wake up but is powered by a battery to transmit. For the drug compliance application, we concentrate on the passive tags since any ingested RFID system containing batteries will impose a degree of risk to the patient s safety A p assive RFID tagging device consists of a tagging IC and an antenna. Based on their mechanism of operation e xisting passive RFID systems may further be divided into either near field / in ductive coupling system or far field/ backscattering coupling system 1.4.3.1 Near -field RFID The near -field RFID tag operates at low frequency. Relying on the energy inductively coupled from the reader, the near field RFID is essentially a resonant tank formed by a transponder coil and capacitive tagging IC To detect a tag located in the detection range, the reader transmits a magnetic alternating field at the tag resonant frequency; t he impinging field causes current to circulate in the tag resonant tank and the peak current at the frequency of resonance can in duce a small but detectable change in the reader coil impedance. This change in

PAGE 25

25 impedance is known as reflectance impedance and can be used by tags to transmit information back to the reader. For example, two distinct reflectance impedance values represent ing digital information can be generated by tuning the resonant tank to or away from the coupling frequency. Considering that the l ow frequency RF power can easily transmit through the human body without incurring too much attenuation, low frequency taggi ng system s seem attractive However, t here are drawbacks associated with the near field technique that limit its usefulness in this application. S imilar to the EAS, the major obstacle of using the low frequency resonant RFID technology for medication compliance monitoring is the relative large size of the tags. In order to induce enough current on the resonant tag to sustain the IC operation the tag coil must be either comparable to the operating wavelength, or must consisted of multipl e turns. Unfortunately, the envisioned RF structure around the capsule is to be made using a single layer of biocompatible conductive inks. While this definition simplifies the manufacturing process, it makes the multi turned structure in compatible. The i nductive coupling is very sensitive to the relative position between the reader and the tagging device Once ingested into the human body, the tagging device moves along the GI tract, and possi bly rotates with in the stomach. T herefore a stable inductive co upling for IC activation cannot be maintained. 1 .4.3.2 Ultra high f requency RFID Recently, telemetry system operating at Ultra High Frequency (UHF) has drawn interest among researchers around the world due to its potentially small size and long operating range [26] [39]. Functioning in the far field loosens the requirement for tag reader alignment, and the e fforts to minimiz e tag antennas [40][47] make this approach more suitable for the proposed

PAGE 26

26 Figure 1 2. Illustration of UHF backscattering RFID tagging system application. The basic operation principle of the UHF RFID tagging system can be illustrated as shown in F igure 1 2. The reader transmits high power at UHF frequency to inquire the presence of the tagging device, while t he RFID tagging device, including the an tenna modulator, the oscillator and the digital core with the unique identification code is energized by the power extracted from the UHF carrier. Once activated, the tagging device communicates with the reader by backscattering the electromagnetic power impinging on the tag antenna T he signal backscattered from the tagging device is generally ~60dB smaller (including the two -way path loss between the reader and tagging device, and the scattering loss on the tag antenna) than the UHF carrier at the reader input In order to distinguish the backscattered signal from this large interferer the tag antenna as an EM scatter is periodically modulated by the tagging IC and therefore produces two backscatter ing signatures. In frequency domain, these two signatures are UHF carrier PA Detector T TAG Circulator LNA Mixer BPF Modulator Frequency Domain Carrier Modulation frequency Sidebands to show the presence of the tag

PAGE 27

27 represented as sidebands on both side of the UHF carrier ; t he reader can then detect the tagging device by recognizing these sidebands. The UHF RFID technology seems to be a good candidate for m edical compliance monitoring, as the tag size can be reduced by simply designing the system at higher frequency. However, this technology is not designed to operate in vivo : when transmitted from embedded or internal sensors to a reader located outside of the body, UHF electromagnetic signals are severely attenuated by the human body. 1.5 Towards an Electronic Pill An ideal compliance monitoring method would consist of both direct and indirect methods for monitoring a patients adherence to a regimen [1 ] [6] Our concept entails using a point -of care (POC) handheld monitor that consists of an RF transceiver and antenna that detects the presence of the electronic pill (e -pill) in the GI tract after it has been ingested by the patient. Without interfering wit h the capsule as a transport for medication, t he e -pill, coded with a unique identification tag, is integrated on to the outer surface of a standard -size d capsule. The patient orally ingests the capsule and activates the POC handheld monitor. The monitor is placed at close proximity to the patients abdomen, enabling the detector to scan for the presence of the e pill in the patients GI -tract. Upon detecting the unique signature of the e pill, the detector stores a tim e -stamped reading of a positive complia nce; i f the detector fails to register the correct e -pill it will alert the patient for failed compliance. The entire history of the patient compliance can be stored inside the detector and relayed to the health care provider for study and analysis 1.6 Dissertation Organization The proposed research will focus on the design and implementation of electronic pills: standard sized capsule s with small electronic devices attached to their outer surface, which can only be detected after ingestion. The primary goals are to understand how the human body

PAGE 28

28 affects the radio -frequency wave transmission, and to design the preliminary system with the necessary capsule structures and integrated circ uit chip based on th ose findings. In C hapter 2, we present a feasibility study of small capsule antennas for medication compliance monitoring. The finite difference time domain (FDTD) method and experiments using biological phantoms (i.e. solutions that resemble the electrical properties of the human body) are used to determine the radiation characteristics of simple loop and dipole antenna structures from inside the human torso. In addition, we provide a brief overview of the development and for mulations of custom silver nano particle inks as the capsule antenna material. Chapter 3 shows our effort to design and characterize the capsule antennas. To verify the d esign and fabrication process, the first zig -zag antenna printed directly onto the standard caps ules is reported. The capsule antenna was printed using the screen printing method, and was characterized on a probe station with its metallic cover replaced by a plastic replica to reduce the grounding effect. Following the same procedure, s everal inducti ve capsule antennas compatible with the UHF backscattering technology were designed and verified. Among those, the capsule antenna with the maximum radar cross section (RCS), which is a term used to represent the scattering ability of a conductive structur e, was packaged with a commercially available IC device to form a backscattering tag, and its detection was demonstrated using a commercial reader. In C hapter 4 we propose a new powering scheme to activate the tag ging device ingested into the human bod y. This method utilizes the fact that the human body is moderately conductive, and transfers the required power to energizing the tagging IC through the human body channel at low frequency The ingested tagging device can then utilize two conductive pads directly contacting the surrounding tissue environment to extract the required energy. Once

PAGE 29

29 activated, the tagging device periodically generates short burst s at UHF frequency to show its presence. To facilitate the design of this asymmetric tagging device the low frequency powering channel through the human body was first characterized through cada ver and saline experiments. Then, the detailed IC design methodology incorporating the characterized powering channel is discussed. A proof -of -concept tagging IC was fabrica ted using CMOS 130nm technology. After its functionality was verified on a probe station, the fabricated tagging IC was packaged with a capsule antenna to form a prototype electronic pill device T he packaged device was then dipped into the sal ine solution that was used to mimic the human torso and its detection was demonstrated using an Agilent mixed -signal analyzer. The research effort to develop the electronic pill for medication compliance monitoring is summarized in Chapter 5, as well as suggestions to future work s that can further improve the design methodology

PAGE 30

30 CHAPTER 2 FEASIBILITY STUDY OF E PILL DETECTION 2.1 Introduction Motivated in part by the need for new alternatives for medication compliance, the primary objective of this chapter is to investigate the feasibility of small printed antennas on standard sized capsules as potential radiating elements for low cost and rel iable detection schemes of orally ingestible electroni c pills (e -pills). As shown in F igure 2 1, the outcome of this study may lead to passive or battery -less radio -frequency tags for standard capsules. With continued advances in RF biotelemetry, it is env isioned that an external monitoring point -of -care device or a body-worn electronic sensor can be used to detect the presence of the pill in the stomach or GI tract after ingestion. Figure 2 1. Passive miniaturized electronic chip for medication complian ce. The proposed micro -system will be approximately 1mm by 2mm and less than 0.5mm in thickness. It is attached to a conductive antenna printed or adhered onto the capsule. The electronic tape is enclosed with biocompatible coating. This chapter presen ts design and implementation aspects of small antennas and their detectability inside the body We employ both simulation tools based on the finite difference time domain (FDTD) method and experiments using biological phantoms (i.e. solutions that

PAGE 31

31 model th e electrical properties of the human body) to determine the radiation characteristics of simple loop and dipole antenna structures. In addition, we present a brief summary of custom silver nano particle inks as a potential capsule antenna material. Detailed studies of the silver inks are presented in [48] Unlike previous studies, this work investigates the feasibility of patterning antennas on the outer surface of a capsule using silver conductive inks. The uniqueness of this approach stems from utilizing the capsule surface area to print electromagnetic radiative elements which could be made to dissolve or detach from the capsule. Antenna s can be made of conductive biocompatible co atings by incorporating a metal that can dissolve, such as silver, under a temporary protective layer such as polyglycolic acid, or by incorporating particles that are non toxic by virtue of being nonabsorbable. The medication capsule can therefore house on its outer surface an electronic compliance device. Thus, the substrate for the antenna and the drug delivery device are the same and the volume reserved for the medication remains unchanged. The RF tagging electronic chip can be made thin, mechanically compliant and smaller than 0.5mm2 [49] under a biocompatible protective se alant and excreted via the GI track. Should direct printing onto capsules become feasible, it could lead to low cost and high volume manufacturing of medication capsules using specialized offset -gravure machines commonly used in the pharmaceutical industry Alternatively, a ready -made electronic compliance tape on biocompatible substrate can be transferred onto the capsule using similar printing systems. This passive system -on a -capsule can potentially incorporate sensors in addition to an RF subsystem for wireless telemetry, and can potentially be made very inexpensive in large numbers for highvolume medication compliance applications.

PAGE 32

32 2.2 Biocompatible Inks Silver has a long history as a material of practical use in civilizations. Besides well known uses in coins, photography, silver is widely used in surgical procedure s prescribed medicine and food. For example, silver acetate (AgC2H3O2) was used for over a decade as a smoking cessation ingredient in both Europe and the United States; even today, pure s ilver -made foil is ingested together with food in India and other countries. In this section, we present a brief overview of formulations and experimental c haracterizations of silver nanoparticle (SNP) based inks. These studies were performed in collaborat ion with researchers from the Department of Material s Science at the University of Florida. The preparation method and biocompatibility of the silver ink is then briefly reviewed, and a detailed discussion of methods, formulations and experimental results can be found in [48]. 2.2.1 Ink Preparation Silver was the chosen capsule antenna material because of its biocompatibility and low sintering temperature when used in colloidal form. Colloidal silver has a sintering temperature about 1/10 the normal silver melting temperature (100C). The key component to creat e an ink suitable for printing as presented in [48] is the protection of silver colloids during silver nanoparticle synthesis; this controls SNP size by limiting agglomeration and particle coalescence Much scientific progress has been made when polyvinylpyrrolidone (PVP) is used for protection of silver ion by forming the Ag -PVP com plex that limits silver growth Under such conditions, stable silver colloids can be formed within the narrow size distribution of 2070 nm, which decreases the melting temperature of the silver particles to below 100C. The silver ink methodology was prepared via reduction with formaldehyde using the PVP as a protective agent to limit the growth of silver -nano particl es [48] The conduc t ive inks can be re -dispersed in either water or alcohol solutions and adapted for a wide variety of printing

PAGE 33

33 requirements. Possible printing applications include inkjet printing (with wi de nozzles) using water and die hyleneglycol, scree -terpineol and ethylcellulose, and pad printing using propylene glycol or an organic binder such as ethyl terpineol. The organic binders are used to adjust the ink viscosity for different printing methods. The ink viscosity determines the thickness of the printed trace and the detail of the resolution. For screen printing, diethylene glycol was added and centrifuged into solution for a viscosity of approximately 20kcps. 2.2.2 Biocompatibility a nd Silver Toxicity A full biocompatibility study of various silver compounds and particles is beyond the scope of this study. Since the target of the proposed application is to have a silver antenna printed or attached on a gelatin capsule or a biocompatible substrate, a general review of silver toxicity presented in [48] is repeated herein for completeness Silver toxicity has been reviewed quite thoroughly with the main focus of the research concerned with cosmetic changes associated with colloidal silve r ingestion and its anti -microbial effects. Argyria, or skin discoloration due to colloidal silver ingestion, is a common side effec t with large doses of silver The EPA publishes a Reference Dose (Rfd) of many major chemicals, which estimates the daily ex posure that will likely be of little physiological harm to single individual during their lifetime. For oral silver, the exposure recommendation target is less than 5 In exposure. Drake and Hazelwood [50] gave a thorough literature review of the health effects of silver and silver compounds. The authors indicated that the toxicity of a metal was influenced by several factors, including the solubility of the metal, the ability of the metal to bind to biological

PAGE 34

34 sites, and the degree to which metal complexes are sequestered or metabolized and excreted. The major exposure pathways of silver into the human body include the ingestion of silver compounds and colloidal silver proteins, the inhalation of dust or fumes in occupational settings, skin contact from burn creams and jewelry, as well as many other medical applications. The greate st toxicity differences between the various silver forms are between soluble and insoluble silver, with soluble silver compounds being indicated to be more easily absorbed. Soluble silver forms complexed primarily with proteins and RNA and DNA by binding t o sulfhydryl, amino, carboxyl, phosphate, and imidazole groups On the other hand, metallic silver and insoluble silver compounds were not readily taken up by the body, although Drake and Hazelwood cited a few studies where argyrosis or localized argyria (such as in the lungs) are developed from silver metal. A number of studies cited by Drake and Hazelwood showed that metallic silver was not soluble in aqueous solutions nor were they solubilized by any physiological mechanism. Therefore, they were more l ikely to be excreted from the body. These considerations as to the differences between the effects of the form of absorbed silver will be critical in future biocompatibility of a silver antenna that may or may not be protectively sealed. Malcolm et al. [5 1] indicated that silver acetate (AgC2H3O2) was used for over a decade as a smoking cessation ingredient in Europe. Three controlled, double -blind European studies were cited as effective in reducing the number of cigarettes smoked per day. In Malcolms st udy, smokers were given silver acetate gum that contained ~6 mg of silver acetate, which yielded 3.9 mg of silver in each stick of gum. This is estimated to be the typical amount of silver contained in one capsule antenna, depending on the final design. Fo r three weeks, t he gum was chewed 6 times per day for 30 minutes each time The total dose per day was then about 23.2 mg (actual percent absorption not known) and the total study dose was 487 mg. Argyrism was not found in

PAGE 35

35 any patients, although argyrism w as one side effect in other silver acetate studies, occurring only after 50 times the normal dose was taken and in lozenge form. Silver acetate was found by an FDA Advisory to be safe as a nonprescription treatment for smoking cessation and allowed for OTC use, but later studies proved the gum ineffective for smoking cessation. This study showed a large dose of silver taken by ingestion could be tolerated with side effects linked more closely to the type of silver ingested (silver acetate) rather than to me tallic silver. Das et al. reviewed the use of food grade silver foils. This study gives evidence for the overall safety of ingested silver metal foil that may be analogous to sintered silver antennas in their material properties. Silver has been used for c oloring agents for confectionary and alcoholic beverages. Foil designs have been used to add a shiny glaze to food preparations in South East Asian and Middle East Asian countries. European Union legislation from 1994 allowed sliver powder or tiny sheets t o be used as colorants in specific foods if the silver was more than 99.5% pure. Das et al. noted that the FDA in 2003 granted approval to silver ion technology as contact substances that could be used for antibacterial applications. In the Indian subconti nent, consumption of silver through silver foils averaged 25 50 mg per day. Previous studies indicated that consumption produced moderate analgesic effects. Human volunteers fed 50 mg of silver foil daily for 20 days showed significant hypcholesterolaemic and hypoglycaemic effects. Intake of good-quality silver foils was said to be considered safe by the authors and there was no mention as to the effects of argyria induced in persons taking silver foil this may indicate that metallic silver is no t absorbe d to any great extent While the previous studies do not definitively argue for or against the ultimate biocompatibility of silver, they offer significant evidence for a continued biocompatibility study of silver metal in the digestive system. For the pur poses of this application, argyria due to silver

PAGE 36

36 colloid ingestion could be mitigated by the sintering treatment applied to the silver traces during the processing of silver capsule antennas. Should silver prove to be the most cost and manufacturing friendly of capsule antenna materials, epoxy-coating and other polymer -sealing methods may prove suitable to protect direct biological degradation and absorption of silver. 2.3. Modeling of A ntennas inside the Human Body In the United States, increasing in terest in wireless medical implants and telemetry devices has prompted the FCC to allocate specific frequency bands for use with medical devices. For example, these include the 402 to 405MHz for Medical Implant Communications Service (MICS) band, the 608 t o 614 MHz for Wireless Medical Telemetry Service (WMTS) band, and the 902928MHz, 2.4 2.483GHz and 5.7255.825GHz Industrial Scientific -Medical (ISM) bands. In order to evaluate the field distributions of electrically small antennas inside the human body i n the above FCC regulated frequency bands, we employed the finite difference time domain (FDTD) method. For this analysis, the capsule antenna s were tuned to the desired frequency bands and the radiation behavior was characterized inside and outside of the human body. 2.3.1 Computational Method Finite difference time domain (FDTD) [ 52] is a popular method to solve problems involving the interaction between electromagnetic waves and complex structures. This method is a grid -based time -domain numerical method : the simulated structure is segmented into boxshaped cells which are small compared with the wavelength. On each of those cells, time dependent Maxwell equations in partial differential forms are approximated into corresponding discrete forms, and the re sulting difference equations are solved to obtain the spatial electrical and magnetic vectors at each time step, which represents the time needed for electromagnetic

PAGE 37

37 waves to travel from one cell to the next. The calculation continues until a state of conv ergence is achieved. The major advantage of using the FDTD method is its excellent scaling performance as the problem to be solved becomes very large. As the number of unknowns increase s due to the increase of the structure size, this method is more efficient and outperforms most other simulation methods. Furthermore, due to its time domain nature, in contrast to most frequency domain simulation methods, the FDTD method can be used to obt ain a broad band solution from a single board -band excitation such as a short pulse, which gives quick and accurate results over a very large frequency band. Liao absorbing boundary conditions [ 53] is used in the simulation instead of perfect matched layer (PML) boundary conditions. B oth bounda ry conditions only allow electromagnetic wave s to travel outward of the simulation space B ut unlike Liao, PML increases the simulation space by defining additional padding layers composed of artificial absorbing mat erials beyond the initial simulation boundary, at the cost for longer simulation time Therefore, simulation using L iao boundary condition s needs a smaller simulation space, which leads to faster convergence. 2.3.2 Simulation Setup The human torso model with a vertically oriented coil antenna in the transverse or XY plane of the torso is shown in F ig ure 2 2 To verify the simulation setup, e xtensive preliminary simulations were performed to calibrate the results against theoretical b enchmarks of well characterized loop and dipole antennas. The process of calibration included determining the appropriate mesh size, cell size, time steps, radiation boundaries, padding cells and source type to yield convergence within a reasonable time. O ur setup consisted of a grid size of 650 by 425

PAGE 38

38 Figure 2 2. Simulated electrical small antennas and human body model. by 140 cells of the torso area with a resolution or mesh size of 1.5mm and an absorbing boundary (Liao absorbing boundary conditions LIAO) with 40 cells of padding. Simulations were carried out for a range of FCC regulated frequency bands spanning 400MHz 1200MHz. In each case, the coil or dipole antenna was first designed to resonate at the aforementioned frequencies by forcing a singl e cell -sized gap at the antenna feed and inserting a complex source impedance of the appropriate value, which was found via Gaussian source excitation simulations. 2.3.3 Radiation in FCC Regulated Frequency B ands A single turn ed loop antenna with diameter 8mm and a dipole antenna with length 2cm were simulated inside the human body model for an American adult male, which is composed of 23 different types of body tissues with different electrical properties. Figure 23 shows the Pill Location X Y X Y Z Single turned loop With diameter 8mm Dipole with each leg length 1cm

PAGE 39

39 Figure 2 3. Simulated resu lts of electrical field distribution for electrical small antennas: (a). vertically oriented single turned loop with diameter 8mm; (b). vertically oriented dipole antenna with total length 2cm. simulated E -field distributions of those antennas outside of t he human torso at the MICS, WMTS, and ISM 915MHz bands. In those simulations, both the loop and the dipole antennas were aligned in the vertical direction, and placed slightly offset from the body axis. A comparison of the antennas operated in the MICS band shows that the single turned loop outperforms the 2 -cm dipole antenna by about 10 15dB. The E -field distributions for frequencies ranging from 400MHz to 1200MHz at ~200MHz steps for horizontally positioned antennas were also carried out. As in previously published studies [54][55], the simulation results indicate that the signal attenuation varies depending on the location of the pill as well as that of the receiving antenna. For both loop and dipole antennas, the maximum E -field intensity occurs within t he range of 600MHz 900MHz. Based on our initial studies, the single -turned loop antenna exhibits good isotropic characteristics inside the human body, and the simulated attenuation caused by the human body was about 40dB at 915MHz ISM band. (a) (b)

PAGE 40

40 2.4 Phantoms E xperiment To validate our theoretical analysis, we performed experiments consisting of measuring the insertion loss of an electrically small loop antenna in electrically equivalent human phantoms. The single turned loop was fabricated by wrapping a piece of copper tape ( with 3 mm wid th ) around a 00-sized capsule [56]. The loop antenna was tuned to 402 MHz MICS band by soldering a n additional chip capacitor across the two ends of the loop antenna. 2.4.1 Phantom P reparation Phantoms are solutions that exhibit electrical properties that are approximately equivalent to biological tissue and are used herein to characterize signal attenuation of electrically small loops in the human torso. While phantom preparations vary in complexity and accuracy, the simplest fo rmulation, which has been used in our experiments, is based on sodium chloride (NaCl) and distilled water; however, improved gel based phantoms consisting of polysaccharide gel, NaCl, alum inum powder and s ucrose can also be used. The phantom solution was f illed in a Rubbermaid container with a slot on top to allow the insertion of the resonated loop a nd the broadband feeding probe which was used for energizing the capsule antenna by inductive coupling. H ow close the phantom solution can model the human to rso should be verified, and this was achieved by measuring the loaded return loss (S11) of a broadband RF probe using a vector signal analyzer (VNA). To take the loading effect into consideration, t he broadband probe was pressed firmly against the back of a human torso and the sidewall of the phantom box, and the measured S11 in those two cases were compared T he relative concentration of NaCl in roughly 4 liters of distilled water was then varied until the loaded S11 on the sidewall of the phantom box matched the S11 when loaded with the human back. The loaded S11 for the broadband probed

PAGE 41

41 Figure 2 4. Measured S11 when narrow band probe is loaded with the phantom solution and human body within the MICS band. a gainst the human back and the finalized phantom solution was in good agreement, as shown in F igure 2 4. 2.4.2 Insertion Loss Measurements To experimentally characterize the radiation property of a small capsule antenna inside the phantom solution, we measured the transmission loss between the resonated capsule antenna inside the phantom box and a n outside broadband receiving coil antenna The experim ent setup is shown in F igure 2 5. To prevent the 50 from adversely affecting the quality factor of the ca psule antenna the 8mm-loop antenna was powered through inductive coupling from a broadband probe in close proximity (~2cm). The receiving coil antenna was located outside of the phantom box, and was matched to 50 antenn a and the receiving coil antenna through the phantom solution to mimic the longest expected distance when the e pill is ingested into the GI tract. Ferrites were used in all cables 3.8 3.9 4 4.1 4.2 4.3 x 108 -100 -50 0 50 100 150 390 400 410 420 430 -100 -50 0 50 100 150 380 Freq(MHz)Probe input impedance ( )loaded with human back loaded with phantom box Real Imaginary Frequency (MHz)

PAGE 42

42 Figure 2 5. Experiment setup of measuring the attenuation caused by the phantom solution at MICS band. connecting to the antennas in order to suppress radiative currents induced along the cable ground shield s The broadband probe was implemented using a multi turn coil (~1.5cm diameter) with self resonant frequency much hig her than the interested frequency range, and therefore was completely unmatched to the VNA input impedance. When the resonated capsule loop antenna was placed in close proximity, because more power was coupled to the capsule antenna, the broadband probe became more matched, and the measured S 11 was improved to ~ 10dB (F igure 2 6). Correspondingly, because the capsule antenna stored and re radiated the coupled energy, more power can be received by the receiving coil outside of the phantom box Neglecting the coupling loss from the broadband probe to the capsule antenna, which is reasonable as their sizes are comparable and they are placed in very close proximity, the path loss for a small capsule loop antenna radiating through the saline solution can be e stimated based on the measured insertion loss (S21) between the powering probe and t he receiving antenna. Shown in F igure 2 6, this path

PAGE 43

43 Figure 2 6. Measured scattering parameters (S11, S21) using the setup in Figure 25. loss was approximately 30dB at the resonant frequency of the capsule antenna which was in good agreement with the FDTD simulation s 2.5 Conclusions This chapter presents the feasibility study of using electronic pills for medication compliance monitoring. The design of such a devic e poses many challenges, such as antenna printing, biocompatible ink development, and the RF detection scheme. Our investigations suggest that silver colloidal ink s are a promising candidate balancing the antenna performance and printability. The silver methodology of bin ary MW PVP protected silver nan o particles yields acceptable electrical resistivities for capsule antennas, and can potentially lead to direct application to gelatin or hypermellose capsules using localized sintering. While previou s studies on the toxicity of silver are not definitive, they offer significant evidence for a continued biocompatibility study of silver metal in the digestive system. In particular, silver in metallic form is generally insoluble in aqueous solutions and m ore likely to be excreted from the body. Nonetheless, concerns relating to silver toxicity can be entirely dB0 -10 -20 -30 -40 -50 60 300 350 400 450 500 Frequency (MHz) S11 of the feeding probe S21 Capsule resonance

PAGE 44

44 eliminated by using a biocompatible sealant. From an overall manufacturing perspective, transferring an assembled medication compliance device onto t he surface of a capsule may prove to be the most viable approach. This would allow printing, sintering and packaging of the antenna and RF tagging microchip in a separate transferable biocompatible matrix under protective coating. In order to determine the radiation characteristics of small capsules antennas inside the human body, a finite difference time domain (FDTD) mesh model of an average American male with 23 different tissue types was used. Both a single turn 8mm loop and a 2cm dipole antenna were si mulated inside the human body model at various FCC approved telemetry bands ranging from 400MHz to 1GHz. As expected, the field intensity was found to vary with frequency and the maximum radiation appeared to be between 600MHz and 900MHz. To experimentally verify the possibility and to prove the correctness of the simulation, we measured the insertion loss between a resonated 8mm -coil antenna around the 00 -sized capsule placed inside of the phantom solution and a n outside receiving coil The results showed that the attenuation for the capsule loop radiating through the phantom solution was approximately 30dB at its resonant frequency, which roughly matched the FDTD simulation results.

PAGE 45

45 CHAPTER 3 CAPSULE ANTENNA DESI GN AND MEASUREMENT 3.1 Introduction This chapter presents design and implementation aspects of small printed capsule antennas. Designing miniature antennas that fit within the limited available surface area of standard -sized capsule s is one of the major challenges towards the implementation of the electronic pill. Unlike previous studies, this work investigates the feasibility of applying antennas on the outer surface of a capsule using conductive inks either by directly patterning the capsule or transferring a flexible antenna inlay onto it The uniqueness of this approach stems from utilizing the capsule surface area to print electromagnetic radiative elements which could be made to dissolve or detach from the capsule. Antennas can be made of conductive bio -compatible coatings by incorporating a metal that can dissolve, such as silver under a temporary protective layer such as polyglycolic acid, or by incorporating particles that are non -toxic by virtue of being nonabsorbable. Thu s, the substrate for the antenna and the drug delivery device are essentially the same and the volume reserved for the medication remains unchanged. This chapter begins with a brief introduction of the capsule and antenna materials. This is followed by a brief discussion of printing techniques used to transfer patterns onto non -planar surfaces. Since antenna designs may require different printing area and axial coverage around the capsule three commonly used printing method s are briefly discussed in sect ion 3.2. I n section 3.3, we report the first antenna printed directly onto the outer surface of a 00-sized capsule using the screen printing method. The antenna was characterized on a custom probe station, and its radiat ion proper ties from inside the human body w ere evaluated through various FDTD simulations The simulation results suggested that the maximum radiation occurs between 600MHz 900MHz. Based on this study the design for two screen printed capsule antennas with

PAGE 46

46 more complex structures are d iscus sed in section 3.4 The antennas were designed to be inductive, and therefore are compatible with UHF backscattering RFID technology. FDTD simulations were carried out to study the 915MHz near -field and far -field radiation generated by the two antennas located inside the body; the return loss of the two antennas was also experimentally verified using the custom probe station. A nother 915MHz capsule antenna aiming to fully utilize the capsule surface is reported in section 3.5 The antenna co vered ~300o axially around the capsule, and was first pr inted onto a flexible substrate which could be wrapped around the capsule. The antenna was printed using the silver conductive ink s, which can be made bio -compatible using methods described in [48]. In order to further validate its performance the antenna was packaged with a commercial available IC to form the prototype backscattering tagging device, and the detection of the assembled tag in air was demonstrated using a commercial RFID reader. The pos sibility of realizing pill -to -pill communication between different capsules is also demonstrated in this chapter 3.2 Capsule Antenna: Materials and Methods The capsule antenna is an electromagnetic radiating element placed directly onto the surface of a hard -shell capsule. Unlike previous studies of wireless transmission using antennas inside ingestible capsules [ 57], this work investigates the radiation properties of antennas placed on the surface of the capsule. Thus the capsule itself, one of the oldest forms of oral dosage, can potentially serve as the printing substrate in addition to a drug delivery device. In the pharmaceutical industry, capsules are routinely patterned and colored in volume by specialized offset -gravu re machines. The coloring inks formulated for specific printing processes are only used for esthetic and brand name labeling purposes. For the capsule antenna in this application, conductive inks could be employed to achieve electrical conduction and electromagnetic

PAGE 47

47 radiation. In t his section, we present a brief discussion of the materials and methods used for the design and fabrication of the capsule antennas 3.2.1 Capsules A hard shell gelatin capsule (Capsugel Division of Pfizer Inc) was used in this study for initial prototypin g of the capsule antenna. Gelatin based capsules are essentially made of a mixture of gelatin, water and colorants that gels into a hard polymer by a simple change in temperature. Capsule formation during manufacturing takes place as stainless steel pins a re dipped into the gelatin solution in a controlled temperature and moisture environment, forming a homogenous film around the mould pins. Another predominant form of the hard shell capsule is the hypromollose capsule, also known as HPMC, which is manufact ured in a similar fashion but use s small quantities of carraageenan and potassium chloride in its solution matrix to facilitate the gelation process during manufacturing [ 58]. Extensive studies have been reported on both types of capsules [ 59] [60]. Although the capsules differ somewhat in their moisture content and appearance, and to some extent in their mechanical properties, their basic properties are similar; they are sufficiently inert and stable, mechanical ly strong for handling during fil l in g and printing, and exhibit desired release and solubility properties in a biological system. The physical specifications for the hard polymer capsules are generally defined in terms of the outer capsule diameter, height or locked length (i.e. cap and body), and the actual volume. For most human oral dosage applications, standard capsule size ranges from 11.1mm in height or joined length for Size 5 capsules to 26.14mm for Size 000 capsules. Generally, the capsule body is larger than the cap, and provides a larger surface area for printing an antenna. For instance, the body of a size 00 capsule has an outer diameter of 8.53mm and axial length of approximately 17mm (this includes only the uniform printing surface, not the actual height of the body). The total surface area for printing an antenna, assuming a full 360 degree printing coverage, is

PAGE 48

48 approximately 17mm by 30mm, or ~510mm2. In practice, high resolution printing is difficult over the entire capsule surface area and the actual area available for an ant enna is much smaller. 3.2.2 Printing Methods To obtain the desired radiation properties, the antenna patterns have to be accurately transferred either directly onto the outer capsule surface, or on to a thin flexible substrate, which can be wrapped aroun d the capsule. As a low cost alternative to the offset-gravure printing machines used in the pharmaceutical industry, we have explored several compatible small scale patterning approaches for antenna printing. 3.2.2.1 Ink jet p rinting Inkjet printing method reproduces an image by propelling variably -sized droplets of liquid ink onto a substrate. Recent advances in inkjet print ing technology for printing nano -scaled inks composed of gold or silver particles has shown good uniformity and conductivity. Current technology uses piezo print heads to deliver pico liter sized droplets of nanoparticles onto various substrates. Those dro plets are then dried to form a thin film, which could show excellent conductivity. Considering the similarity of the printing procedure, an Epson R220 inkjet printer was modified mechanically to allow printing on a circular substrate. This printer has piez oelectric heads with 12.4 micrometer openings and a droplet volume of 3pL, fo r a reported resolution of 5760 x 1400 dot s -per inch (dpi). Preliminary printing on gelatin based substrate indicated line resolution s of about 400 m. Multi turn coils antennas printed on 00 sized capsules us ing inkjet printers are shown in F ig ure 3 1. The initial experiments of printing thick traces around the capsules was rationalized due to the following reasons: t hick traces are among the si mplest geometry to be printed; b y printing simple traces, the procedure that allows directly printing onto the circular surface of standard sized capsules can be established; t he existing printing technology can be directly applied in the proposed applications. Although t he initial

PAGE 49

49 Figure 3 1. Multi turn coils printed on gelatin capsule using modified inkjet printer. results were promising, this printing method is mere suitable for small scale prototyping instead of large scale manufacture. 3.2.2.2 Screen p rinting The screen printing method involves a finely woven metallic mesh stretched by a solid mesh frame. The mesh contains the negative of the pattern that is to be printed. The size of each woven cell of the area containing no antenna pattern is much smaller compared with the particle size of the ink, while the area with the negative antenna pattern contains cells with much larger openings, such that the ink can only pass through these areas. A squeegee made by soft plastics pushes the ink through the area with large r openings, as it is pressed both downward and across the mesh as the same time. This printing procedure leaves the positive image of the antenna pattern on the substrate.

PAGE 50

50 The screen printing method has limited axial coverage for circular surfaces. For example, when printing directly onto a gelatin capsule, the capsule has to be rolled across the metallic screen. Without a stable rotation axis, the accuracy of the pattern cannot be preserved. Through numerous trials of printing in the lab environment, we f ound out that only about 1/4 of the curvature surface area around an 000 sized capsule can be used to correctly transfer the designed pattern, if printed by hand in lab environment This indicates that the screen printing method is best suited when printin g a sleeve that consists of a heat treatable polymer substrate which could be wrapped around the gelatin capsule. Besides the limitation of printing on curved surfaces screen printing also requires an additional cleaning step after every few printings to prevent the residual ink particles from ag glomerating and blocking further ink trespassing. 3.2.2.3 Pad p rinting The pad printing method has been used to print on irregula r objects for over 15 years. This method implements an automated process which allows for rapid prototyping, low ink consumption, repeatability, and is suitable for large scale production. The process consists of a silicon pad of various hardness and shape, which is used to transfer the pattern from the printing plate (clich) to the substrate. The clich is created by etching grooves that are effectively the negative image of the pattern. Ink is applied and kept in the grooves, and is then picked up by the silicon pad. Then the pad is pressed ov er the sub strate to transfer the pattern and the positive image of the antenna will be produced on the substrate. Pads with different softness are chosen based on the different types of the substrate: compressible objects such as the gelatin capsule need a softer pad. A small horizontal shaped pad is ideal for printing radial patterns on a cylinder surface. Unlike inks compatible with the screen printing method, which requires slow drying time to avoid clogging the mesh, inks suitable for pad printing must dry fast to ensure transferring the artwork from the clich to the pad, and from the pad to the substrate. The

PAGE 51

51 Figure 3 2. Screen -printed Zigzag dipole antenna, and its SEM picture. mi nimum resolution of the method is determined by the ink viscosity and study has shown that a line resolution of 25 m can be achieved using the pad printing method. 3.3 Screen -P rinted Capsule Dipole A ntenna 3.3.1 Capsule Antenna The first prototype capsule antenna was fabricated by patterning a simple zig -zag dipole antenna directly onto the surface of a two piece gelatin based capsule using the screen printing method. A dipole antenna was chosen for its ease of printing and well known radiating properties. A size 00 -capsule with outer diameter of approximately 8.53mm and height or locked length (with the cap) of 23.3mm was chosen for this initial prototyping. As shown in F ig ure 3 2 the antenna was printed longitudinally to maximize the surface area used al ong the axial direction and increase the electrical length of the dipole. To facilitate ink transfer using the screen transfer method, printing was limited to within 40 degree s of radial coverage which resulted in an available printed area of approximately 17mm by 3mm. Thus, the dimensions of the antenna were limited to 14.5mm in length and 2.9mm in height. Meandering the antenna in zig -zag fashion resulted in a dipole length increase from 14.5mm to 27mm. To facilitate

PAGE 52

52 measurements and allow the antenna to be contacted directly using RF probes, the separation of the two antenna terminals were kept less than 250 m The silver conductive inks were formulated with a viscosity of approximately ~20k cps and a typical electrical resistance of /sql, which is approximately 30 times hi gher than the bulk resistivity of copper or silver. A scanning electron microscopy (SEM) image shows that the ink lay-down thickness after sintering is approximately 25m. 3.3.2 Antenna Measurement Setup Validation The prototype ca psule antenna w as measured directly on a manual RF p robe sta tion, as shown in F igure 3 3 The return loss and input impedance of the prototype capsule antenna s were measured using an experimental setup composed of an HP8510 network analyzer, a balun, highfrequency RF probes and a manual RF probe station from JMicroTechnologies (model JR 2745). In order to minimize ground plane reflections from underlying metallic structures, the RF probe station was modified using a custom built acrylic structure with r ~ 3.5, thereby increasing the separation between the DUT antenna and the underlying metallic support from a few millimeters to approximately 10cm additional acrylic plastic extension plates allows larger separations of slightly over 15cm. The differentia l input feed of the capsule antenna was provided using a balun to convert the single ended signal from the analyzer to balanced signals for low loss interference. The capsule antenna was contacted using the high-frequency SGS differential probes from GGB i ndustries with a signal to -signal pitch of 300m. The measurement setup was calibrated to the probe tips using short, open and load on a calibration substrate. In order to determine the effect of a metallic ground plane on the accuracy of the measurements the antenna along with a circular ground plane were simulated using an

PAGE 53

53 Figure 3 3. The measurement setup for the electronic pill electromagnetic full -wave structure simulator (HFSSTM). The antenna and the circular metal chuck were placed inside a cyl indrical air box representing the radiation boundary. As a reference, the input impedance of the zig -zag dipole antenna as fabricated in F igure 3 2 was first determined without the effect of a ground plane. As shown in F igure 3.4(a ), the antenna resonates at approximately 6.1GHz. In F igure 3 4 (b ), the simulated real and imaginary parts of the input impedance are plotted as t he separation between the underlying metallic plate and the antenna was varied at 5mm increments. For separatio ns larger than 10cm, the real part of the input impedance is within 5% of the expected value (shown in dashed lines), with an acceptable change in the imaginary component. The effect of the custom probe station for lower UHF frequencies, such as the 915M Hz ISM band, was also characterized. Although the metal antenna separation in this case seems very short the diameter of the underneath metal plate is small compared with the operating

PAGE 54

54 (a) (b) Figure 3 4 HFSS simulation of effect of metal holding structure to the antenna impedance. wavelength ( ~20% of the wavelength ) and therefore should not introduce large measurement errors. To verify whether the measurement setup is also suitable for capsule antenna s designed at 915MHz, we simulated the input res istance and reactance of the zigzag antenna at 915MHz and varied the separation between the under lying metal plate and the antenna The simulation result s show less than ~10% error at 915MHz when the DUT antenna is separated from the underlying metal plate by more than 10cm. 3.3.3 Zig -Zag Capsule Antenna M easurement Single port S -parameter measurements of the zig zag dipole prototype antenna were carried out using the aforementioned setup. As shown in F igure 3 3, the contacts between the antenna terminals and the RF source were established using high frequency RF probes. The 1 2 3 4 5 6 7 8 9 0 200 400 1 2 3 4 5 6 7 8 9 -1500 -1000 -500 0 Im Validation frequency Re 400 200 06 7 8 9 Frequency (GHz) 2 3 4 5 1 0-500 -1000 -1500 300 100Rin ( )Xin ( ) Metal plate antenna 0 20 40 60 80 100 120 -10 0 10 20 30 40 50 60 70 80 90 Distance(mm)Ohms Resistance Reactance 90 70 50 30 10 20 40 60 80 100 120 Simulated impedance without Metal plate at resonant frequency10Zin ( )Separation (mm) Xin@5.8GHz Rin@5.8GHz 0 50 100 150 200 250 2 3 4 5 6 7 8 9 0 50 100 150 200 250 -1500 -1400 -1300 -1200 -1100 -1000 -900 -800 50 100 150 200 Separation (mm) Simulated impedance without Metal plate Xin@0.915GHz Rin@0.915GHz 250 800 900 1000 1100 1200 1300 1400Xin ( )9 8 6 4 2Rin ( )7 5 3 0

PAGE 55

55 (a) (b) Figure3 5 Capsule antenna (a) zig -zag antenna and close up of probe tips on the antenna structure and (b) measured input impedance of the zigzag antenna using the modified probe station. attainable line separation between the antenna input feed pads was approximately 250m, which was sufficient for landing the RF probes with 300m signal to -signal pitch. The measured return loss and input impedance are shown in F igure 3 5. The resulting return loss from measurements is 14.4dB at 5.78GHz, which f alls within the 5.8GHz ISM band, and the 3dB bandwidth is 795MHz or 13.75%. Overall the single port measurements are in good agreement with simulated results. For practical applications, the antenna operating band would have to be tuned to a lower frequency using an electronic microchip placed across the input terminals of the antenna. This can be easily accomplished if the terminating impedance is inductive, which shifts the resonance to lower frequencies. 3.3.4 Radiation inside the H uman Body In order to evaluate the radiation properties of the zig -zag capsule antenna from inside the human body in various FCC regulated frequency bands, we employed the finite difference time

PAGE 56

56 domain (FDTD) method. T he capsule antenna was tuned to the desired frequency bands by assigning a passive port composed of an inductive component across the antenna terminals, and the near -field electrical field distribution s and far -field radiation patterns at the simulated bands were characterized. 3.3.4.1 Model s etup A representative cross -section of the human torso model and the location and orientat ion of the an tenna is shown in F ig ure 3 6 The antenna was located in the stomach area, deep -set in the torso slightly offset from the body axis. We considered both near -field and far -field intensities of a longitudinally (from left to right) oriented zig -zag antenna. A more in -depth study of the effects of antenna orientations for an electrically small antenna is provided in [ 6 1 ]. In our study, the objective was to evaluat e the radiation properties of the zig zag capsule antenna to determine the feasibility of using a surface printed antenna to implement the electronic medication compliance device. The antenna and overall FDTD model setup was calibrated extensively for accuracy and rapid convergence. The calibration process comprised of determining the appropriate mesh size, time steps, radiation boundaries, padding cells and source type to yield convergence within a reasonable time. The simulation setup for the zigzag capsule antenna consisted of a grid size of 650 by 425 by 140 cells of the torso area with a resolution or mesh size of 1.5 x 1.5 x 1.5 mm3 and an absorbing boundary (Liao absorbing boundary conditions LIAO) with 40 cells of padding. An adaptive or localized sub -gridding approach with minimum mesh size of 0.15 x 0.15 x 0.15 m3 was used to better describe the physical and electrical nature of the radiating antenna for improved resolution. For each of the simulated antenna orientations, the input impedance was first determined over a broad frequency range using a narrow Gaussian-shaped pulse. Simulations were then carried out for a range of FCC regulated frequency bands spanning 400MHz 5800MHz, and in each case the antenna was mad e to resonate at the frequency of

PAGE 57

57 Figure 3 6 Simulation setup for zig-zag capsule antenna radiatin g from inside of the human body interest by forcing a single cell sized gap and inserting a complex source impedance of the appropriate value. 3.3.4.2 Near -field analysis Near -field electric field radiation contours of the zig -zag capsule antenna were eva luated in the FCC approved 402405MHz MICS band, the 608614 MHz WMTS band, and the 902 928MHz, 2.4 2.483GHz and 5.7255.825GHz ISM bands. The resulting near -field intensities shown in F igure 3 7 are all normalized to a 0dBm source and the minimum field contour plotted is 75dB. Loading effects of the surrounding tissue on impedance and radiation efficiency were accounted by resonating the antenna (inside the body) at the frequencies of interest using complex source impedance. Our simulation results show t hat the radiated field intensity of the capsule antenna is strongest in the anterior location slightly to the left of the stomach, which would be the optimal location of an external receiving antenna. The radiated intensity first increases and then decreas es with increasing frequency. The low field intensity at lower frequency is because of the low radiation efficiency of the electrical small antennas. As the posterior anterior longitudinaltransverse transverseverticalcut plane capsule antenna

PAGE 58

58 Figure 3-7. Cross-section of hu man body model torso indicating th e normalized radiation field intensity contours for the WMTS 608-614MHz, ISM 902-928MHz, ISM 2.42.45GHz, and ISM 5.725-5.825GHz FCC freque ncy bands. Radiation is most efficient in the 902-928MHz band. The sma ller circles above a nd below the torso represent the arms of the subject. operating frequency increases, radia tion efficiency increases and more power can be radiated through the body. This fact accounts for the initial in crease of electrical inte nsity with frequency. When operating above about 500MHz, high attenua tion is expected due to the increased absorption of electromagnetic power by the body, which causes the increased attenuation as the frequency increases. The radiated fields for the simulated zig-zag antenna are most efficient in the 915MHz ISM band. This agrees well with previous studies fo r electrically small antennas radiating from inside the body [61]. The field in tensity outside the body is approximately -25dB 3 83 83 63 43 23 02 82 62 42 21 4-203 84 25 05 56 06 57 07 5363 84 04 44 85 56 57 5 7 56 04 42 61 42 23 65 03 43 43 8-144 8 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 05 15253545 10 20 3040 cm ISM 2400-2483MHz 7 55 03 52 55 0-60 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 05 152535 45 10 20 30 40 cm ISM 5725-5875MHz 0 10 20 30 40 50 60 70cm 3 23 43 63 84 04 55 04 55 05 56 06 57 07 56 06 57 57 0656 05 55 04 53 03 2-22-24243 8-454 54 03 04 05 55 0-40 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 05 152535 4510 20 30 40 0 10 20 30 40 50 60 70 cmcm WMTS 608-614MHz 4 04 55 05 56 06 56 06 57 07 55 55 04 56 06 57 06 06 57 07 53 8-363 83 42 83 83 03 4-32 -40 -503 64 54 55 56 56 56 5 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 05 15 253545 1020 3040 0 10 20 30 40 50 60 70 cmcm MICS 402-405MHz 6 05 55 04 54 03 83 6343 23 02 82 62 8303 23 42 22 42 62 83 03 23 43 63 84 04 5 5 05 04 55 55 56 06 57 07 57 57 06 56 05 55 04 51 42 01 43 04 0-405 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 05 1525 3545 1020 3040 0 10 20 30 40 50 60 70 cmcm ISM 902-928MHz

PAGE 59

59 at 915MHz and 10dB lower at 402MHz. It is interesting to note that despite the increasing absorption of electro magnetic energy at high frequency bands due to high water content tissues, the radiation field intensity in the front of the torso at the 2.4GHz ISM band is comparable to that of the 915MHz ISM band. This result is largely due to improved radiation efficiency of the zigzag dipole antenna at higher frequencies. Further increase in the operating frequency, as depicted by the radiated f ield intensity at 5.8GHz, results in significant loss in the region directly surrounding the antenna and is likely not suitable for electronic capsules. 3.3.4.3 Far -field patterns The far field patterns of the zig zag capsule antenna radiating from inside the human body were also characterized through simulations. Because the body tissue generates large attenuation within the ISM 5.8GHz band, we only simulated the far -field patterns at frequencies ranging from 4002 400MHz. Figure 3 8 shows the azimuthal rad iation patterns for a longitudinally oriented zig -zag dipole located in the stomach slightly offset from the body axis, with the body model standing upright. For each frequency, both horizontal ly and vertical ly polarized patterns were simulated. Unlike the symmetrical donut -shaped azimuthal pattern for the dipole antenna radiating in free space, radiation from inside of the body is weaker in the back side, because of the high density of muscle and bones which make up the back side of the torso. Al though 1020dB lower compared with the horizontally polarized radi ation, the patterns in F igure 3 -8 show some perpendicular polarized radiation likely caused by re -radiation of the dominant horizontal field components at the interfaces between different bo dy tissues. The overall anterior radiation is better than the posterior far fields which can be attributed to attenuation caused by the thick dorsal muscle mass that runs down the back of the torso. Similar to the near field radiation intensity, the far fi eld radiation is the strongest at 915MHz for the simulated frequency bands;

PAGE 60

60 (a) (b) Figure 3 8 Far field radiation patterns for the tuned zigzag pill antenna inside of the human body: (a). Antenna placed parallel to the abdomen; (b) Antenna placed perpendicular to the abdomen. further increase in the operating frequency results in severe attenuation due to the increased absorption of EM energy. C ompared with 2.45GHz, the antenna radiation is more i sotropic at frequencies below 1GHz. 3.4 Screen -P rinted Capsule Antennas for RFID Backscattering at 915MHz is the most popular technique to implement UHF RFID system s, and t he reading range of such a system, which is defined as the maximum distance for th e tagging device to be detected by the reader, is a critical design parameter The reading range, Rr, of an RFID system can be derived from the well known Friis equation and expressed as in (3 1): -25 dBi -38.75 -52.5 -66.25 -80 0 30 60 90 120 150 180 210 240 270 300 330 402MHz MICS -25 dBi -38.75 -52.5 -66.25 -80 0 30 60 90 120 150 180 210 240 270 300 330 610MHz WMTS -25 dBi -38.75 -52.5 -66.25 -80 0 30 60 90 120 150 180 210 240 270 300 330 915MHz ISM -25 dBi -38.75 -52.5 -66.25 -80 0 30 60 90 120 150 180 210 240 270 300 330 2400MHz ISM -25 dBi -38.75 -52.5 -66.25 -80 0 30 60 90 120 150 180 210 240 270 300 330 402MHz MICS -25 dBi -38.75 -52.5 -66.25 -80 0 30 60 90 120 150 180 210 240 270 300 330 610MHz WMTS -25 dBi -38.75 -52.5 -66.25 -80 0 30 60 90 120 150 180 210 240 270 300 330 915MHz ISM -25 dBi -38.75 -52.5 -66.25 -80 0 30 60 90 120 150 180 210 240 270 300 330 2400MHz ISM Horizontal Polarization Perpendicular Polarization Horizontal Polarization Perpendicular Polarization

PAGE 61

61 thP 2 | | 1tag G Tx G t P 1 = r R 2 4 (3 1) Figure 3 9. Two inductive capsule antennas designed to be compatible with the frontier RFID technology. where Pt is the power transmitted from the reader device, Pth is the m inimum power required for Tx and Gtag are the gain of the reader and tag antennas, and is the reflection coefficient at the tag antenna IC interface. Since t he RFID reading range is maximized when the input impedance of the tagging IC is matched to the tag antenna matching considerations are extremely important However, external matching networks are not feasible in our proposed application and therefore matching must be accomplished through proper antenna design. Since monolithic inductor s at UHF are large and lossy in silicon, and the tagging IC presents a capacitive impedance, it is desirable to design inductive antennas matched to the tag IC impedance at the operating frequency. 8.3mm 15.3mm 15.3mm 5.5mm 105o 70oAntenna 1 Antenna 2

PAGE 62

62 Figure 3 9 shows two inductive capsule antennas The antennas were printed directly onto the capsules using the screen printing method, achieving 105o and 70o of axial coverage around the 000-sized capsule s ( with l ocked height 26.14mm, and diameter 9 .52 mm ) respecti vely Considering the limited capsule surface area available for patterning, line meandering technique were used to increase the antenna s electrical length. By folding or meandering the antenna traces distributed induc tive and capacitive reactance are pr oduced. These distributed reactance impact the antenna input impedance, significantly lowering the resonate frequency at the expense of lower bandwidth and smaller antenna gain [42]. Single port input impedance characterization was carried out for b oth inductive capsule antennas o n the custom probe station T he measure ment result shown in F ig ure 3 10 indicates that the two capsule antennas resonate at 1.6GHz and 1.95GHz, respectively. For the desired 915MHz ISM operating frequency band, t he input impedance of Antenna 1 was measured as 46.7+j196.5 ( and the input impedance of Antenna 2 was 42+j167.1( T h ese inductance values can be used to c ancel approximately 1pF input capacitance presented by the tagging IC The radiation pr operties of the t wo capsule antennas were simulated inside the human body model using the FDTD method Since previous published result s in [ 6 1 ] and the study of the zigzag dipole antenna ( section 3.3) all suggest that electrical small antenna s radiate most efficiently in t he frequency range of 600MHz 1200MHz from inside the human body we only studied the two capsule antennas within the 915MHz ISM band. T o resonate the antennas to the desired frequency, capacitive loads with proper values were assigned across the antenna input terminals. The antennas were placed longitudinally in the stomach within the transverse plane, and their near and far -field radiation patterns with two perpendicular orientations were simulated

PAGE 63

63 (a) (b) F igure 3 10. The measured and simulated (XFDTD) input impedance of the inductive capsule antennas. (a) antenna 1; (b) antenna 2. Similar to the study of the zigzag dipole antenna inside the body, the maximum electrical field intensity (Figure 3 11) of both capsule antennas is observed in the ant erior location of the human body and slightly to the left, which could be the best location for the wireless detection device. T he radiated E -field distribution varies somewhat with capsule antenna orientation, but 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 x 109 -1500 -1000 -500 0 500 1000 1500 2000 800 1200 1600 2000 2400 2800 2000 1000 0 -1000 -1500 500 1500 -500Frequency (MHz)Zin ( ) Measured Simulated 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 x 109 -2000 -1500 -1000 -500 0 500 1000 1500 2000 2500 800 1200 1600 2000 2400 2800 2500 500 1500 2000 500 1500Frequency (MHz)Zin ( ) 1000 0 1000 2000Measured Simulated

PAGE 64

64 (a) (b) Fig ure 3 11. The simulated near field electrical field distribution inside and outside of human body. (a) Antenna 1placed parallel and vertically to the abdomen; (b) Antenna 2 placed parallel and vertically to the abdomen. the optimal reader location does not change significantly. Compared with Antenna 2, Antenna 1 covers a larger surface area on the capsule and as a result it is a more efficient radiator T he simulation results show that the radiated E -field intensity of Antenna 1 in the anterior loc ation of the human body slightly to the left is ~ 20dB, which is about 6dB higher than Antenna 2 at the same location Fig ure 3 12 shows the simulated far -field radiation patterns of the two inductive antennas oriented vertical and parallel to the stomach in the azimuthal plane. Similar to the near field results, the strongest radiation happens in the front side of human body and slightly to the left. In all the simulated patterns, t he back side radiation is less compared with the front side radiati on because of the dense r muscles and bones running down the spine Because the antenna is placed in the transverse plane, most radiated power is horizontal ly polarized ; but there also exists some perpendicular ly polarized radiation (~ 10dB less than the hor izontally polarized power ) The perpendicular radiation is primarily generated by the refraction and scattering of EM wave at interfaces between various types of tissue with different electrical properties. -20-22-26-30-30-34-30-32-34-36-40-40-50-40-45-50-45-45-60 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0 5 15 25 35 45 10 20 30 40 0 10 20 30 40 50 60 70 cmcm -20-22-26-30-32-34-36-40-40-40-45-50-36-40-55-40-40-45-45-50-30-32-36 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0 5 15 25 35 45 10 20 30 40 0 10 20 30 40 50 60 70 cmcm -26-30-32-34-36-40-45-50-55-60-50-55-50-50-50-45-50-60 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0 5 15 25 35 45 10 20 30 40 0 10 20 30 40 50 60 70 cmcm -26-30-32-34-36-40-45-45-45-40-34-40-50-50 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0 5 15 25 35 45 10 20 30 40 0 10 20 30 40 50 60 70 cmcm

PAGE 65

65 (a) (b) Fig ure 3 12. The simulated far field radiation patterns of the two inductive antennas (a) Antenna 1 placed parallel and vertically to the abdomen; (b) Antenna 2 placed parallel and vertically to the abdomen. 3.5. Pad -P rinted Capsule Antenna The radar cross section (RCS) is a quantitative way of measuring the ability of a conductive structure as a n electromagnetic scatter er and it can be defined a s the area intercepting that amount of power which, when scattered isotropically produces at the rec eiver a density which is equal to that scattered by the actual target [62]. In mathematic form, RCS can be expressed as: i s RW W R lim24 (3 2) -5dBi -30 -55 -80 -105 0 30 60 90 120 150 180 210 240 270 300 330 -5dBi -30 -55 -80 -105 0 30 60 90 120 150 180 210 240 270 300 330 Horizontal polarization Perpendicular polarization -5dBi -30 -55 -80 -105 0 30 60 90 120 150 180 210 240 270 300 330 -5dBi -30 -55 -80 -105 0 30 60 90 120 150 180 210 240 270 300 330 Horizontal polarization Perpendicular polarization -5dBi -30 -55 -80 -105 0 30 60 90 120 150 180 210 240 270 300 330 Horizontal polarization Perpendicular polarization -5dBi -30 -55 -80 -105 0 30 60 90 120 150 180 210 240 270 300 330 Horizontal polarization Perpendicular polarization -5dBi -30 -55 -80 -105 0 30 60 90 120 150 180 210 240 270 300 330 -5dBi -30 -55 -80 -105 0 30 60 90 120 150 180 210 240 270 300 330

PAGE 66

66 wh ere Wi is the incident power density, Ws is the scattered power density, and R is the observation distance from the target. For a backscattering RFID system, t he tag sends data by switching its input impedance between two states, effectively changing its RCS and thus modulating the backscattered field. The quality of this communication link is primarily determined by the difference of RCS of the tag antenna between the se two loading states. The RCS of an antenna with specified load impedance can be calcul ated as [63]: 2 22 2 l a az z R G (3 3 ) where G is the tag antenna gain, Za is the antenna complex input impedance (Ra is the input resistance), and Zl is the complex load impedance. T he maxim um RCS difference can be achieved if the load impedance of the antenna is switched between zero and infinity i n which case the quality of the tagto reader link is determined by th e antenna gain G In practice, the gain of an electrically small antenna is proportional to a2, where a is the maximum dimension of the antenna [64]. Therefore, i n order to maximize the antenna gain, and thus improve the RCS, the capsule antenna should be designed to cover as much the capsule surface as possible 3.5.1 Asymmetric Antenna Design and Measurements In this section, we present an inductive capsule antenna that fully utilizes the capsule surface. Due to the inherent limitations of patterning directly onto circular objects, this antenna was first pad -printed onto a flexible PDMS substrate, and then wrapped around the gelatin capsule. The inductiv e loop antenna was designed to fit over a 000-siz ed capsule as shown in F igure 3 13. The antenna m easures 25 x 15.2 mm2, and when wrapped around, it achieves ~300o

PAGE 67

67 Figure 3 13. Asymmetric electrical loop antenna designed to maximize the gain of a capsule sized antenna. of radi al coverage Since the longest dimension of the antenna printing substrate (000-sized capsule, with diameter ~9mm and locked length ~26mm) is only ~10% of the operating wavelength, the antenna trace s are arranged in a meandering fashion in order to increase the electrical length An antenna with regular meandered lines does not exhibit the optimum gain especially when the conductor losses can not be neglected [6 5 ]. Up to the first resonant frequency, the current distributed on the adjacent parallel segments of a regularly meandered antenna have opposite phase. The individual segments can be modeled as current elements that do not contribute significantly to the radiated power, and produce additional losses that degrade the overall antenna gain [6 4 ][6 5 ]. Thus, the antenna design pro cedure consisted of deliberately mismatching the meandered segments in order to minimize the cancelling effects of the distributed current. In the absence of Genetic A lgorithms (GA), simulations for a number of

PAGE 68

68 Figure 3 14. Simulated and measured input impedance of the proposed asymmetric electrical loop antenna. antenna configurations were carried out using the method of moment (provided by ADSTM) The final design shown in Figure 3 13 comprises of mismatched meandered lines that are asymmetric about th e center feeding point. T he printed antenna was characterized using the custom probe station, and the measured input impedance is plotted together with the simulation result s in F igure 3 14. Both the measure d and the simulated results show ed that the c apsule antenna resonates at around 950MHz, and presents inductive impedance within the desired 915MHz ISM band. However, compared with the simulation result s t he bandwidth of the printed antenna appears to be larger, which shows the printed antenna experi ences more ohmic loss This can be attributed to the nonuniform lay down of the conductive ink and the antenna patterning distortion during printing 700 800 900 1000 1100 1200 1300 1400 1500 -3000 -2000 -1000 0 1000 2000 3000 Freq(MHz)Ohm Measured Simulated 700 1100 1200 1300 Frequency (MHz)800 900 1000 1400 1500 -3000 -2000 -1000 0 1000 2000 3000 Zin ( )

PAGE 69

69 (a) (b) Figure 3 15. Simulated open-circuit RCS of the self resonant antenna for different illuminating o; (b). o. As an indication of how well a capsule -sized antenna can be used as electromagnetic scatterer the open-circuit RCS of the capsule antenna to three different solid angles in space was simulated usi ng the FDTD simulator for two orthogonal illuminating angles. The simulation result s (F igure 3 15) clearly demonstrate that the antenna RCS is largest at the antenna resonant 500 750 1000 1250 1500 -120 -110 -100 -90 -80 -70 -60 -50 -40 MHzRCS (dB) 1. 2. 3. X Y Z Incident direction Polarization -50 -40 -60 -70 -80 -90 -100 -110 120RCS( dBsm ) 500 750 1000 1250 1500 Frequency(MHz) 1 2 3 1. =0, =0; 2. =0, = /4; 3. =0, = /2. 500 750 1000 1250 1500 -120 -110 -100 -90 -80 -70 -60 -50 MHzRCS (dB) 1. 2. 3. -60 -50 -70 -80 -90 -100 -110 -120RCS( dBsm )500 750 1000 1250 1500 Frequency(MHz) 2 3 1. =0, =0; 2. =0, = /4; 3. =0, = /2. X Y Z Incident direction 1 Polarization

PAGE 70

70 Figure 3 16. Attenuation contours around and inside of human body for the elec trical asymmetric loop antenna placed inside of the GI tract: (a). Antenna along Y direction; (b). Antenna along Z direction. frequency The RCS of th is pad -printed capsule antenna reaches the maximum value ~ 45dB sm when the longitudinal side of the capsule is perpen dicular to the incoming EM wave. The simulation results also show that most of the backscattered electromagnetic energy maintains the original polarization, which is ~30dB higher than the cross -polarized product. In order to characterize t he antenna as a radiator from inside the human body, the padprinted antenna was also modeled inside the human stomach and simulated using the FDTD method. To help the simulator converge faster, and reduce the simulation time, only a n 8 cm 0 5 15 25 35 45 10 20 30 40 0 10 20 30 40 50 60 70cmcm Vertical oriented 0 5 15 25 35 45 10 20 30 40 0 10 20 30 40 50 60 70cmcm Parallel oriented

PAGE 71

71 thick section of the human body was included in the simulation. While most cells in the simulation space were set to 5 x 5 x 5mm3, the pad printed antenna was meshed to 0.15 x 0.15 x 0.15mm3 using an adaptive grid in order to accurately model the antenna. A s mall capacit ance (~1pF) representing the input reactance of the tagging IC was shunted across the antenna terminals to tune the antenn a down to 915MHz. Absorbing boundary condition ( Liao absorbing boundary conditions ) with 40 padding cells were again utilized in the s imulation to help achieve faster convergence. The near -field electric field intensity of the resona nt antenna in and around the human torso at 915MHz is shown in F igure 3 16. The capsule antenna was oriented in two perpendicular directions, but kept in the same spot inside the stomach. The simulation results show that although both orientations result in similar maximum radiation at the anterior human body slightly to the left ( 14dB for both cases), the pattern of the perpendicular oriented antenna appe ared to be more isotropic. 3.5.2 Antenna V alidation U sing Existing RFID T echnology The pad -printed capsule antenna as a UHF backscattering RFID antenna was further evaluated by packaging the antenna to a commercial RFID chip from Alien technology. The Alien RFID system utilizes the EPC 2 protocol, and operates within the 915MHz I SM band. In this system, e ach tag is assigned a unique 192-bit ta g identification code, which can be backscattered to the reader once the tag is energized. The input impedan ce of the alien chip at 915MHz is 1500 j145 (1.2pF) [66], which can roughly cancel the input inductance of the pad printed capsule antenna. Once the electrical connection s with the tagging IC were established, the antenna was wrapped around a 000-sized c apsule. The experime ntal setup is shown in F igure 3 17. A microstrip antenna contained inside of the white box at the bottom of the table was the reader antenna. A laptop computer in the far end

PAGE 72

72 Figure 3 17. Validation of the asymmetric loop antenna using existing technology from Alien system. was used to analyze the received data, and the green bar shown on the laptop monitor indicated the detection of the RFID tag, with the received 128-bit identification code in its hexadecimal format shown at the bottom. The packaged tag ging device was held ~1.5m from the reader antenna, which was the maximum range that the tag can be detectable. As a comparison, we also experimentally characterized the detection range of Alien tag consisting of the same tagging I C and an Alie n antenna with similar size as the pad -printed UF Antenna Alien AntennaWrapped UF Antenna With Alien chip Receiving Antenna Alien Interface: (Green Bar shows the detection)

PAGE 73

73 capsule ant enna (F igure 3 17). The tag formed by the pad-printed antenna achieved approximately the same detection range as the Alien tag when both antenna s were laid out flat; b ut when wrapped a round the gelatin capsules it appeared that the tag with the pad -printed antenna could achieve ~ 2 times of detection range Figure 3 18. Measurement setup for pill to -pill communication. 3.5.3 Pill-to -P ill Communication The capabilities of pill to -pill communication can prove to be vital for a variety of scenarios to en large system functionality. For instance, the e -pill functionality can be expanded to include various miniature body sensors to enable the measurement of a variety of medical u seful vari ables from inside of the GI tract with pill to pill communication protocols (sensed data collected by any pill could be distributed, shared, uplinked to a centralized external reader). To explore such possibility, we measured the transmission loss between two capsule antennas on a custom probe station, as shown in F igure 3 18. The two pills used in the measurement were made by wrapping the pad printed antennas around 000-sized capsules; e lectrical connections with the pills were es tablished by two SGS RF pr obes with 300 m signal -to -signal pitch. On the Pill_1 Pill_2 45o 15cm SGS Probe SGS Probe

PAGE 74

74 Figure 3 19. The measured coupling gain of the capsule antenna pair for pill to -pill communication custom probe station, the two pills were separated by ~ 15cm from each other (center to -center), and they were tilted 45o apart. RF baluns were used to convert the single -ended signal generated from the VNA into differential form, and the measurement setup was calibrated beforehand to the probe tips using short open, load and through standards on a calibration substrate. Because the capsule antenna is not designed for 50 the antenna -cable interface s and therefore t he characterized S21 along can not correctly predict the transm ission loss between the two capsule antennas in matched condition s To account for these losses, the transmission gain between the antenna pair is characterized use the concept of coupling gain Gc, which can be defined as (3 4 ). ) S )( S ( S = G21 c 2 22 2 11 21 1 (3 4 ) 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 x 109 -90 -80 -70 -60 -50 -40 -30 800 1200 Frequency (MHz)900 1000 1100 1300 1400 1500 1600Coupling Gain (dB) 30 40 50 60 70 80 90 Pill resonant frequency

PAGE 75

75 Using the calibrated setup as in Figure 3 18, scattering parameters including S11, S22, and S21 were captured simultaneously using the VNA, and Gc of the capsule antenna pair was calculated with in the frequency range of 800MHz 1600MHz, as plotted in F igure 3 19. It appears that the maximum coupling gain in this case is approximately 40dB, which is achieved at the resonant frequency of capsule antenna s (~ 950MH z). 3.6 Conclusions Unlike previous studies of wireless transmission using antennas inside ingestible capsules the focus of this study was to design an antenna that can be attached to the outer surface of standard capsules. T he gelatin based capsules are sufficiently inert and stable to serve as the substrate for the antenna and mechanical ly strong for handling during filing and printing S everal small scale patterning approaches to print antenna s directly onto the surface of readily available hard shell capsules have been reviewed. To demonstrate feasibility, a zig zag dipole antenna was first printed onto the 00 -sized capsule using screen printing methods. The printed antenna was experimentally verified on a manual probe station. To reduce reflections du e to effect s caused by the metal plate underneath the DUT antenna, the probe station was modified by removing the top metal cover and replacing it with a plastic duplicate. The near and far -field radiation characteristic of the screen -printed zigzag anten na radiating from inside the human body was simulated using the FDTD method. The analysis for the zigzag antenna shows the 915MHz ISM band as an optimal frequency band to trad e off signal attenuation inside the human body and antenna radiation efficiency A series of more complex antenna designs were patterned using screen -printing method. Compatible with the backscattering RFID technology, the antennas were designed to present an inductive impedance within the 915MHz ISM band and made to cover 70o -100o o f the axial capsule surface All antennas were measured on a probe station to determine the input impedance

PAGE 76

76 and return loss. In addition, numerical simulations using FDTD method were carried out to determine the normalized E -field contours and radiation pa tterns inside the body Table 3 1. Summary of different types of printed capsule antenna. Size(mm 2 ) Axially Coverage Self resonant frequency Max. field intensity outside the body Printing method Zigzag dipole 14.5 x 2.9 40 o 5800MHz 25dB Screen Printing Inductive Antenna 1 15.3 x 8.3 105 o 1500MHz 20dB Screen Printing Inductive Antenna 2 15.3 x 5.5 70 o 1950MHz 25dB Screen Printing Asymmetrical antenna 25x15.2 300 o 950MHz 14dB Pad Printing Due to the stringent space limitations, most of the antennas exhibit a resonant frequency above the desired 915MHz ISM band. However, by fully utilizing the capsule outer surface with 300o radical coverage, a 950MHz antenna was designed. The antenna was pad printed onto a flat PET substr ate, then wrappe d around the 000 sized capsule. To further evaluate it as a backscattering RFID antenna, the pad printed antenna was packaged with a commercial ly available IC to form a 915MHz electronic pill device. Compared with the tagging device formed by the same IC a nd the commercial RFID antenna roughly the same size as the designed antenna, the tag with the pad -printed antenna showed 2 times longer detection range when wrapped around the capsule. The concept of pill -to -pill communication was also demonstrated. The a ntennas discussed in this chapter are summarized in T able 3 1.

PAGE 77

77 CHAPTER 4 CHIP DESIGN FOR E -PILL SYSTEM 4.1 Introduction Chip technology is widely used in a variety of bio -medical related applications, such as a nimal identification systems des igned for animal tracking [68], and safety -sponge systems with embedded transponders to prevent inadvertently leaving sponge inside of the human body after surgical procedures [6 9 ]. Although none of these technologies operates at UHF frequency or requires the signal to transmit through the entire biological object, a similar concept can be applied to the proposed electronic pills. Including a miniature IC chip within the envisioned electronic pills enables increased intelligence as an event marker technology. For example, identification codes can be assigned to each e-pill to store useful information such as the type of medication, or date and place of manufacture. Other features such as anti -collision protocols can enable multiple e -pills to be detected simultaneous ly in the human GI tract In addition, temperature and sensors can easily be integrated to relay this information once the tag enters the digestive tract. The tagging IC in this system ona -capsule can be made thin, mechanically compliant and small under a biocompatible protective sealant and safely excreted via the GI track. This chapter presents the design of an asymmetric tagging IC for the proposed electronic pill device. Once ingested, this device can be activated by power transmitted through the huma n body at low frequency, and generates periodic bursts at UHF frequency to show its presence inside the GI tract. The unique challenges in this particular application are discussed first in section 4.2 Then, the low frequency powering channel is character ized. In section 4.3 a simplified systematic model for the device is proposed, and critical design targets are therefore determined based on the theoretical study. The detailed circuit implementation of the tagging

PAGE 78

78 device is discussed in section 4.4 A pr oof -of -concept tagging IC was fabricat ed using 130nm CMOS technology, and its characterization is shown in s ection 4.5 The IC was packaged with a custom designed capsule antenna to form the prototype electronic pill, which was experimentally validated in phantom solution. Concluding remarks are presented in s ection 4.6 4 .2 Power and Communication for Passive In -Body Microsystems Using the Human Body as a Transmission Medium Passive in -body microsystems refer to small, intelligent biomedical devices situated in deep -set locations within the body and powered from sources external to it. These passive devices, unlike battery operated systems, can be orders of magnitude smaller than the volume occupied by the battery itself and therefore are desirable fo r many applications where size constraints are severe. In the context of small passive or battery less devices, the range of operation is typically limited to a few centimeters due to weak coupling between the external source and the device. This is compou nded by signal attenuation inside the body owing to factors such as frequency, tissue type, distance, antenna size, mismatch losses, etc. The vast majority of passive biomedical devices employ near -field inductive links as the primary method of powering, a nd load impedance modulation for communication [ 70 71]. Many far -field devices, such as passive UHF radio frequency identification (RFID) transponders achieve large distances in free -space, but the system constraints dictated by our envisioned medication c ompliance device precludes the use of traditional backscaterring RFID technology inside the body due to extremely low antenna efficiencies, severe signal attenuation of electromagnetic signals and safety requirements limiting the field exposure to humans. The following sections describe the basic components of the proposed medication compliance device and the associated challenges for communication and powering. Based on experiments and simulations of the human body as a transmission channel, we present our signaling methodologies.

PAGE 79

79 4 .2.1 An Electronic Medication Compliance D evice -Associated Challenges The proposed medication compliance tag comprises of an electronic microchip and a biocompatible antenna inlay placed on the surface of a standard sized capsule It is envisioned that this electronic pill (e -pill) will be monitored using an external point -of -care device such as a body-worn reader to detect the presence of the pill in the stomach or GI tract after ingestion. Since the device is entirely passive, i t must be powered by the external reader (herein we do not consider electrochemical energy derived from stomach acids [ 72]). Moreover, since the electrical pathway from the reader to tag is through the human body and not over the surface of the body (i.e. skin), as in many of the previously reported applications [ 73 75], it is critical to establish the powering and communication methodologies between the reader and the tag, and vice versa. Before describing the methodologies, we present the unique set of ch allenges associated with our medication compliance device. First, the tag construct is envisioned as having two major components: the antenna inlay and the microchip. The inlay is to be made using a biocompatible substrate that is made to dissolve inside the body with a targeted dissolution rate. The antenna pattern, which is printed on the inlay, uses biocompatible silver conductive inks and also made to dissolve, passing through the digestive tract without harmful effects. Details of materials and biocomp atibility studies are beyond the scope of this work, but the interested reader can refer to the following publication [48] for additional information. The second component, the microchip, will be placed under a protective biocompatible coating and excreted via the GI tract. This system definition allows us to encapsulate only a small portion of the inlay (i.e. the microchip) while keeping the much larger antenna and substrate exposed to internal digestive fluids. Since the inlay is in essence wrapped around the pill, the available surface area is related to the size of the pill, which for a typical 00 -sized capsule with an outer diameter of 8.53mm and axial length of 17mm is

PAGE 80

80 approximately 145 mm2. This is large compared to the microchip area of ~1mm2 but sti ll electrically small in relation to low frequency signals. The second challenge stems from the fact that our antenna is to be made using a single layer of biocompatible conductive inks. While this definition simplifies the inlay manufacturing process, it severely limits the types and shapes of antennas available for powering and communication. For instance, coils other than single turn loops are not possible, and therefore signal coupling via inductive methods would be extremely inefficient. Moreover, due to the limited conductivity of biocompatible inks compared to that of bulk material [ 48], the antenna quality factors are low. Space constraints and increased signal attenuation with frequency within the body result in electrically small antennas which ar e inefficient radiators. Another critical challenge relates to safety limits of RF exposure and tag activation energy. Recommendations of safety levels with respect to Human Exposure to Radio Frequency Electromagnetic Fields are described in the IEEE stand ard C95.12005 [ 7 6 ]. These recommendations are expressed in terms of maximum permission exposures (MPE) and specific absorption ratio (SAR) values. For instance, the SAR localized MPE is 2 W/kg averaged over 6 minutes. The SAR can be expressed in terms of 2 3) and E is the rms electric field strength in tissue (V/m). Assuming a nominal tissue density of 1000 kg/m3 and representative tissue conductivity for muscle and skin at 900MHz of ~1S/m [ 7 7 ] for an ISM band UHF reader, the maximum permissible rms electric field is ~45V/m. If the transmitting antenna is placed at 1 cm distance from the body, the corresponding maximum permissible effective isotropic radiated power (EIRP) by the antenna is ~28dBm. State -of -the art far -field passive UHF RFID transponders require anywhere from 10dBm to 15dBm of input power to energize the device. Considering

PAGE 81

81 signal attenuation of ~30 40dB inside the body [ 61] and antenna loss of ~20dB, the amo unt of power that must be deposited into the body would far exceed the maximum field exposure limits. Moreover, practical issues such as signal reflections within the body, antenna directionality and mismatch losses are likely to exacerbate the problem. Wh ile these back of -the -envelope calculations provide only a crude estimate for a link margin analysis, it is clear that alternate signaling methods are required to power the proposed passive microsystem. 4 .2.2 The H uman Body as a P owering and C ommunication M edium Alternatives to near -field magnetic coupling or far -field RF powering as a means to energizing a passive device inside the body can be traced to studies of bioelectricity (Galvani, Volta, Faraday, etc). It is well know that the human body supports current flow via the intra cellular and cellular membranes, and the resulting impedance between two points in the body is a complex function of frequency, tissue type, organs, geometry and distance. A significant amount of literature has been devoted to the study and assessment of bioelectricity and equivalent tissue models [7880]. A very thorough review of permittivity and conductivity of tissues was published by Gabriel [ 7 7 ]. While these models and studies provide excellent cues to understanding the path ways of electrical current flow through the human body, the most relevant information necessary for establishing the human body as a direct transmission channel between two or multiple devices within or on the body is the frequency dependent signal attenua tion characteristics between these nodes. This, in turn, depends both on the method of coupling the signals into the body and the human body channel characteristics. Two basic approaches have been developed to couple signals into the body for signal transm ission, namely capacitive and galvanic coupling, both of which have been predominantly used for data transmission between external body-worn sensors or personal area networks.

PAGE 82

82 (a ) (b) Fi gure 4 1. Illustration of transmission through the human body: (a) capacitive coupling method, (b) galvanic coupling. Figure 4 1 shows a simplified schematic of the capacitive and galvanic body transmissions methods. Capacitive coupled body transmission systems (Figure 4 1 (a )) exploit the capacitive coupling of the body to its environment to establish signal pathways between a source and a detector in close proximity with the body [ 8 1 7 5 8 2 ]. The transmitter and receiver, each with two electrodes, are coupled via electric field lines generated by the transmitter and terminating at the receiver using predominantly the surface of the body as the propagation medium. Return pathways are established by the second electrode of each unit and the surrounding env ironment (or ground). Since the field lines generated by the transmitter electrode escapes through the body to ground and are partially cancelled by the localized return pathways and stray fields, only a small portion of the fields are coupled to the recei ver. In addition, since the return pathways via ground play an integral role in the overall transmission system, the VT x IpriHuman body Isec Detector Transmitter Receiving electrode Ground E1 E2E3 E4 E4E5 E6Capacitive coupling Galvanic coupling

PAGE 83

83 signals detected at the receiver depend not only on the human body but also on its surrounding environment, which is subject to varying conditions. An alternative to capacitive coupling is to directly couple signals into the body using a galvanic method, as shown in Figure 4 1 (b ) Similar to the capacitive approach, the galvanic transmission system relies on a transmitter and receiver, each with two electrodes, coupled instead via transmission of alternating currents through the body and/or over the body surface using electrodes in direct contact with the skin [ 8 3 8 4 ]. Current flow, generated by a differential alternating voltage across the t ransmitter electrodes, is primarily between these electrodes. A portion of the current, however, reaches the receiver to create a detectable differential voltage across the receiver electrodes. This method of transmission depends primarily on the human bod y channel instead of the surrounding environment (i.e. ground) or air to couple the transmitter and receiver via electrical conduction supported by ionic fluids within the body. 4.2.3 Low Frequency Human Body Channel C haracterization Based on the two sig naling approaches presented in s ection 4.2.2 the galvanic method of coupling signals into the body is most suitable for the proposed passive medication compliance device for the following reasons. First, since the proposed tag is physically very small, coupling signals into the tag to extract a differential voltage from the secondary alternating currents induced by the external transmitting electrodes would require only a pair of small electrodes patterned using a single layer of biocompatible conductive i nks. Second, since the body as a whole is a conductive medium, the constraints on the physical transmitting electrode placement on the body can be relaxed, eliminating the need for direct line of site with the ingested tag for increased user comfort. For i nstance, unlike inductively or electromagnetically coupled systems, one can envision a wrist watch or an armband with transmitting electrodes to power a tag inside the stomach without physical alignment constraints. Finally, galvanic coupling minimizes the

PAGE 84

84 dependence on the surrounding environment as the signal transmission is primarily constraint to within the body. To the best of our knowledge, this work presents the first implementation of an in -body electronic tagging device powered entirely from galvan ic electrodes using the human body as a transmission medium. Previously published works have been focused on communication between devices on the body and not inside of it [ 8 1 8 4 ]. Figure 4 2. Measured channel loss from various locations on a cadaver to the inside of stomach and in phantom solution. In order to evaluate the signal attenuation of a galvanic transmission system, a set of initial experiments were carried out on an adult human cadaver. The experiment consisted of placing two small electrode s inside the stomach of the subject to record the differential induced 45 40 35 30 25 20 15 10 5 0 0 5 10 15 20Frequency ( MHz )Signal Attenuation ( dB ) 215 265 350 543 792 1150 1750 Phantom Elbow Hand Chest Arm/Shoulder WristCadaver VT xZs1RIC Zs1 Zs4 Zs2Zs2Human body Vind+ Zs3

PAGE 85

85 voltage generated from two external coupling electrodes. The two coupling electrodes, each measuring 10x3.5 mm2, were used to induce time -varying currents by applying a 10Vp p signal di rectly onto the skin in various body locations, including the hands, elbows, arms and chest; the receiving electrodes in the subject stomach were separated by ~4mm. The induced voltage Vind between the two receiving electrodes was recorded using an oscilloscope with 1M impedance. The resulting channel attenuation in the human body was measured over a frequency range as shown in Figure 4 2. Among the various coupling electrode locations, the channel loss from the right elbow to the stomach appears to be ~10dB lower compared to other locations, and therefore could be the best location for an armband or wristband type wearable transmitter Since access to human cadavers for routine experimentation is limited, we have also created human phantom models. P hantoms are solutions that exhibit electrical properties that are approximately equivalent to biological tissue and are used herein in place of the cadaver to obtain re producible measurement results [8 5 8 6 ]. While phantom preparations vary in complexity and accuracy, the simplest formulation, which has been used in this work, is based on sodium chloride (NaCl) and distilled water, however, improved gel based phantoms consisting of polysaccharide gel, NaCl, aluminum powder and Sucrose can also be used. Our phantom solution consists of 0.45% NaCl and 2 gallons of water filled in a plastic container measuring 40 x 40 x 30cm3, and its conductivity is characterized to be ~0.52S/m at 13MHz and ~1S/m at approximately 900MHz. In our measurements, the four electrod es were immersed inside the phantom box. The two receiving electrodes were located in the center of the phantom box, at a distance of ~15cm from the coupling pads, which were pressed against the box wall. A simple representative circuit model of the electr ical pathway between the coupling electrodes and receiving electrodes is

PAGE 86

86 shown in Figure 4 2 [8 7 89]. This simplified circuit is used to model the channel attenuation through the human body for power transfer to the tagging device. The resistance RIC repre sents the input resistance of the e burst IC which in turn depends on the RF -DC convertor design and the chip power dissipation. To properly capture the interplay between IC parameters (i.e. input impedance) and the channel characteristics, measurements were carried out over a range of RIC values: t he channel attenuation in the phantom was measured for several representative RIC ranging from 97 As shown in Figure 4 2, the phantom experiments are in good agreement with the cadaver experiment. Note that t he phantom experiments tend to overestimate the losses, especially at lower frequencies, providing added margin for error. As expected, increasing RIC decreases the attenuation but only to the extent that is not l imited by the inter electrode re sistance Zs3 of the receiving pads. For instance, at 13.56MHz doubling RIC from 97 to 215 60% increase in Vind; whereas increasing RIC beyond 500 yields only a 10% improvement in received amplitude. 4.2.4 Safety and R egulations Safety restrictions [7 6 ][89] are established to limit the exposure of human body to electromagnetic energy, and therefore provide protections against known adverse health effect, which appears to be painful electro-stimulation in the frequency range of 3kHz 5MHz, and adverse heating in the frequency range of 100kHz 300GHz. Among various regulations for maximum permission exposures (MPE), we have particular interests in the limitation of contact or induced current and SAR, since the former is related to the tag act ivation and the latter determines the maximum allowed radiation from the capsule antenna. Table 4 1 summarizes the rms induced or contact current limits for continuous sinusoidal waveform specified in [ 7 6 ], and the frequency f in the table is in the unit of kHz. It shows that the contact current limitation, which generates perception and pain, is proportional to the

PAGE 87

87 operating frequency in the range of 3kHz 100kHz, and remains as a constant above 100kHz. Derived from the table, the maximum allowed current induced by touching is 16.7mA at above 100kHz for general public. Table 4 1. RMS induced and contact current limits for continuous sinusoidal waveforms Condition General public (mA) In controlled environments (mA) 3kHz 100kHz 100kHz 110MHz 3kHz 100kHz 100kHz 110MHz Both feet 0.90 f 90 2.00 f 200 Each foot 0.45 f 45 1.00 f 100 Contact, grasp ----1.00 f 100 Contact, touch 0.167 f 16.7 0.50 f 50 Exposure to low -frequency electric and magnetic fields normally results in negligible absorption. However, exposure to high -frequency electromagnetic energy can lead to significant absorption therefore increasing tissue temperature via heat dissipation. Ge nerally, temperature rise of less than 1 degree will not result in adverse heating effects, and therefore are considered to be safe. In the guidelines specified in [ 7 6 ] and [ 89], the absorption of electromagnetic energy is expressed in the term of SAR, and the localized exposure limit averaged over 6 minute ( which corresponds to the capsule antenna radiating from inside of the body) in the frequency range of 100kHz 3GHz is 2W/kg for general public and 10W/kg for persons in controlled environment. 4.3 In -Body Electronic Burst (E -Burst) RF Transponder: System Optimization The critical insight of the proposed RF tagging architecture is the use of an asymmetric powering and communication scheme to circumvent problems associated with signal attenuation inside the human body and poor radiation efficiency of electrically small antennas. Since the power levels required to activate a tag are orders of magnitude larger than what is detectable externally, the tag is energized at low frequencies where in body attenuat ion is low, and RF bursts are transmitted at higher frequencies where the efficiency of a capsule -sized antenna is higher.

PAGE 88

88 Figure 4 3. Symbolic representation of the tagging device and its two operating phases. The operation of the tagging system can be treated as a two-step energy conversion process. The low frequency energy transferred from the reader device is first converted into DC using an RF DC convertor to energize the tagging device, and the stored DC en ergy on the tagging device is then converted to UHF bursts to show the presence of the tagging device. Therefore, the tagging device can be symbolically represented using the simple circuit model as shown in Figure 4 3 with its operation divided into a ch arging phase and a burst phase. In the charging phase, the device is energized by the power transferred through the low frequency human body channel. The resulting output current, Io, supports the nominal current consumption of the tagging device, represe nted by Is, and Ic, the current supplied to the storage capacitor Cs. Once the voltage established on Cs reaches the pre -determined value Vhigh, the tagging IC enters the burst phase. In this phase, the stored energy is used to sustain a short UHF burst re presented by load current IL, causing the voltage across Cs to discharge When the voltage Low Frequency Powering Channel Is Cs IC UHF Human body channel RX Supply voltage Vhigh VlowCharging Phase(LF DC) Burst Phase (DC UHF) UHF high Power Burst DT (1 D)T Ic IL Io

PAGE 89

89 of Cs is reduced to Vlow, the tagging device again enters the charging phase and begins re charging Cs. This cycle repeats every period T and the system enters the burst phase for DT, where D is the duty cycle. Our proposed design procedure for the tagging device begins by considering the relationship between required receiver specifications and the radiated power during the burst phase; the tag must momentarily deli ver enough power so that communication through the human body channel results in an acceptable signal level at the receiver input. The system level design is optimize d by creating a system that meets the required specifications. Circuit level parameters ar e then selected to ensure that the system level goals are met. 4.3.1 System M odeling In order to successfully detect the tagging device inside the body the periodic bursts generated within the body must be reliably detected by the external receiver. This minimum detect able received power can be uniquely defined as a function of the signal energy -to -noise ratio Eb/N0, assuming the receiver incorporates a matched filter [9 0 ]. Therefore, the required average radiated power PT_ rad from the tagging device that achieves a specified Eb/N0 is first derived through basic link budget analysis. Estimated c ircuit level parameters (storage capacitor size, average charging current, etc.) are then calculated using this minimum necessary PT_ rad at the tag. The required mi nimum input power, o r the sensitivity of a RF receiver can be represented as BW R N E BW N kT SNR BW N kT Pb f f rx 0 0 0 (4 1) where kT0 is the thermal noise factor in units of watts/Hz, Nf is the noise figure of the receiver, BW is t he bandwidth of the receiver frontend, and R is the communication dat a rate. In its

PAGE 90

90 simplest form, presence of each UHF burst can be represented as single bit of information, and thus the resulting dat a rate R in ( 4 1) is essentially 1/T where T is the transmission period If the power attenuation of the UHF signal traveling out through the human body is represented as t he required Prad from the tagging device for a specified Eb/N0 is then T N E N kT P Pb f rx rad T10 0 (4 2) Since this system level expression for PT_ rad has been defined, circuit level parameters can now be selected by modeling the load placed on the tag as a constant current source, IL. Assuming Cs is linearly discharged from Vhigh to Vlow by IL, which represents the average current consumption of the tagging device during the burst phase, the total energy dissipation ET_ dis can be calculated by integrating the instantaneous power dissipation from (1 D)T to T, as shown in equation (4 3). DT I V V dt ) t ( V I EL low high T T ) D ( L dis T 21 (4 3) T he radiated energy is then the product of ET_ dis and RF ( where RF is the overall DC RF energy conversion efficiency), and PT_ rad can be written as D I V V E T PL low high RF dis RF rad T 2 (4 4) The available energy to drive PT_ rad is also related to the average charging current Ic through charge conservation on Cs. The charge lost from Cs during the burst phase (due to IL) must be equal to the charge placed on Cs throughout T (even during the burst phase, the comparably small IC continues to charge the capacitor), as shown in ( 4 5). DT I T I dt ) t ( I QT L c c 0 (4 5)

PAGE 91

91 From ( 4 5), IL in ( 4 4) can be replaced with Ic/D By equating ( 4 2 ) and (4 4 ), the required average charging current for the specified Eb/N0 is T V V NE kT N Ilow highRF b f c 0 02 (4 6) Comb in ing (6) with the charging equation in the charging phase Ic(1 -D)T = Cs(Vhigh-Vlow), the required storage capacitor Cs can then be evaluated as ( 4 7). 2 2 0 01 2low high RF b f sV V D N E kT N C (4 7) Numerical values for IC and Cs can then be estimated by considering practical concerns associated with the variables of the theoretical equations (4 6) and ( 4 7). The trip voltages Vhigh and Vlow, are limited by the IC fabrication process Vhigh is chosen as the nominal device voltage, in this case 1.2V for the UMC 130nm CMOS process Vlow is set to 0.6V to ensure adequate operation o f support ing circuits during standby Based on the previously discussed FDTD study of small antenna s radiating inside the GI tract in Chapter 3 the attenuation factor in this calculation is set to 50dB ; this is the worst case value taken from the front side of the human torso. As a reasonable estimate of practical receiver performance, w e specify the received signal energy to noise ratio Eb/N0 to be 30dB, and the receiver noise figure Nf to be 20dB. Assuming T is the range of ~1mS, these values amount to an estimat ed receiver sensitivity of ~ 90dBm, using (4 1). As a design aid, the required Ic for three different values of T is calculated using ( 4 6), and plotted in F igure 4 4 versus the total DC RF. As expected, in order to store the sam e amount of energy on Cs within a constant period T, less charging current is required with higher values of RF Although Ic cannot be exactly estimated without knowledge

PAGE 92

92 Figure 4 4 Calculated charging current Ic for different burst rate, assuming Eb/N0=30dB, = 50dB, Nf = 20dB, Vhigh = 1.2V, and Vlow = 0.6V. RF, which can only be determined after the modulator circuit is fully designed, F igure 4 4 shows that Ic is in the micro amp range with an RF larger than 2.5%. Given the duty cycle D anRF, the required size of the storage capacitance Cs can be evaluated by ( 4 7). F ig ure 4 5 shows that Cs is roughly independent of the duty cycle D if it is smaller than 10% However, a more accurate estimation of Cs depends on RF, and will therefore be determined after the modulator is designed. To derive ( 4 6) and ( 4 7), system level considerations for PT_rad were used. At this point in the design procedure, a numeric value for the required PT_rad is necessary for circuit level modulator design. Since the burst modulator is designed independently, we define Pburst as the required radiated power exclusively in the burst phase. D / P Prad T burst (4 8) Pbur st, calculated using ( 4 8) is plotted in Figure 4 6 versus D for various data rates (1/T values) For the same radiated energy, a longer burst requires smaller Pburst, resulting i n a 0 2 4 6 8 10 12 14 0.00% 2.50% 5.00% 7.50% 10.00%Average charging current Ic ( A) RF 500Bursts/Sec 1000Bursts/Sec 2000Bursts/Sec

PAGE 93

93 Figure 4 5 Required storage capacitance Cs as a function of D, assuming Eb/N0=30dB, = 50dB, Nf = 20dB, Vhigh = 1.2V, and Vlow = 0.6V. Figure 4 6 Required burst power Pburst from the tagging device for various period T, to achieve 30dB Eb/N0, assuming = 50dB, Nf = 20dB, Vhigh = 1.2V, and Vlow = 0.6V. monoto nically decreas ing function of D. A realistic D can be estimated by comparing the average charging current Ic and the average discharging current IL. The discharging current IL includes the current consumption of the UHF oscillator and the power amplifier that drive s the capsule antenna. Because the voltage on Cs, which is used as the supply voltage for the tagging IC, declines rapidly while in the burst phase, the required UHF oscillation is generated by a LC 0 1 2 3 4 5 6 0.0001 0.001 0.01 0.1 1Required Cs(nF)Duty Cycle D RF = 0.02 RF=0.05 RF=0.08 -45 -40 -35 -30 -25 -20 -15 -10 -5 0 5 0.0001 0.001 0.01 0.1 1Required Pburst(dBm) Duty Cycle D 500Bursts/Sec 1000Bursts/Sec 2000Bursts/Sec

PAGE 94

94 oscillator due to its insensitivity to supply voltage variations. Considering the current consumption of an LC-oscillator fabricated us ing a similar CMOS pro cess is on the order of milliamp [91][92], and the estimated Ic from Figure 4-4 is on the order of microamp, it is reasonable to approximate D as ~0.001. Therefore, the targeted Pburst is set to -10dBm for ~1000 bursts/sec tag-to-reader link. Figure 4-7. Finding the minimum VTX by incorporating the low frequency channel to the multiplier design. 4.3.2 Tag Activation: Required Reader Output Voltage In the charging phase, the low frequency energy transmitted through the human body is converted into DC energy to char ge the large storage capacitor Cs at the tagging device, slowly increasing the voltage. In this design, the taggin g device is defined as active (enters the burst phase) when the DC voltage on Cs reaches Vhigh. An RF-DC multiplier, implemented as diode-multip lier, is utilized to convert the extracted AC signal at its input into a DC volta ge. To minimize the external voltage (VTX) required to activate the tag, the previously characterized powering channel must be taken into account during the design procedure because the induced input voltage is a function of the effective input impedance of the tag, RIC. Vin VTXLow Frequency Powering Channel RIC Cse Cse RF-DC Multiplier Voltage doubler (positive half cycle) Is+IcCsVout Cse Cse Voltage doubler (negative half cycle)

PAGE 95

95 Figure 4-8. Design the multiplier using charge equilib rium based on the behavior of a single diode device. The symbolic representation of a RF-DC mu ltiplier driven through the low frequency powering channel is shown in Figure 4-7, in which Vout is the targeted output voltage and Io is the average current consumption at the output. Th e basic cell of a multiplier is voltage doubler, operating in either the positive half cycle or the negative half cycle of the sinusoidal input. If the voltage doubler is composed of ideal diodes and operating with infinite load, the generated DC voltage of a basic cell will eventually reach 2Vin. However, due to the finite diode turn-on voltage and reverse charge leakage of practical diodes, the doubler output voltage is smaller than the ideal case. The presence of static current c onsumption at the output counteracts the charging mechanism, leading to a further voltage decrease. As the input voltage level is comparable to the diode turn-on vo ltage, a multiplier with several stages composed from the basic doubler cells may have to be used to generate a large DC output voltage. Including a ll non-ideal effects such as the diod e finite turn-on voltage and reverse leakage, such a RF-DC multiplier is generally designed based on charge conservation [93]-[96]. This method treats the RF-DC conversion as a charge transferring proc ess, and the charge Vd-DC t Vdiode Vin t Idiode Reverse leakage Vin Ic+IsVd_DC Idiode Cse Tp RrecVse+ + -

PAGE 96

96 distributed to the output of the multiplier must be equal to the difference between the total electrical charge transferred from the input and the charge loss caused by reverse -leakage. Recognizing that the electrical charge transferred to the output of the multiplier propagate s through each diode, the charge transfer basic cell of the multiplier can be modeled u sing the simplified cir cuit in Figure 4 8 [95] In steady state, when the input sinusoidal voltage exceeds Vd DC, plus the forward diode voltage the diode is forward -biased and charge is delivered to its cathode. As the input decreases below this level, charge is transferred backward s to the diode anode via reverse leakage current To reach equilibrium, the overall charge delivered to the cathode during each period must be able to sustain the output current Io, or Ic+Is. Therefore, ( 4 9) must be satisfied over the input period Tp. p o DC d p in T d T dT I dt V T t sin V I dt V Ip p 20 0 (4 9) Note equation (4 9) can be solved by normalizing Tp to unity and key design parameters of the multiplier, such as the number of stages N and the required device size to support the total lo ad current Io can be deduced from (4 9). Specifically, if a diode is implemented with a diode connected MOS topology the W/L ratio can be selected to satisfy the required I -V characteristics. Thus, if the design targets of Vout and Io are specified, the multiplier can be uniquely determined for a certain Vin by solving the integral equation. The matlab code used to solve the integration equation is included in the Appendix. The RF DC power conversion efficiency ld is a figure of merit of multiplier desi gn, and it is defined as the ratio of the power sent to the multiplier output to the total input power Ptotal, which includes both the power consumed by the load and the power loss on the M diodes of an multi -stage RF multiplier Then ld is

PAGE 97

97 diode o high o high diode load load total load ldMP I V I V MP P P P P (4 10) where M is the total number of diodes included in the multiplier design, and Pdiode is the average power consumption of a single diode, which can be found by (4 11). dt V T t sin V I T dt V I T PDC d p in T d p d T d p diodep p 2 1 10 0 (4 11) In this application, a larger ld does not necessarily mean the overall system is more efficient since the multiplier may require a larger voltage across its input, which leads to a larger VTX at the read er Since the ultimate goal is to achieve reliable detection with minimum required input voltage, we optimize the system with minimum VTX as our target. Therefore, the low frequency channel model should be included in this design. From the multiplier design, the overall input resistance RIC of the tagging IC can be determined, which in turn is related to the input amplitude and total power consumption as Vin 2/2Ptotal. Once RIC is known, the required external voltage VTX can be derived from Figure 4 2. As a design example, we calculate the required diode size for a RF -DC multiplier composed of N differential stages. A single differential stage is composed of two basic voltage doubler cells o n e for the positive half cycles and one for the negative half c ycle s of the sinusoidal input. Therefore, assuming the diodes are identical, Vd_DC can be approximated as Vout/4N [ 9 3 ]. T he tag can only be activated if the DC voltage on Cs is higher than Vhigh, thus the targeted Vout is set to 1.2V. Considering th at the static current consumption (Is) of a typical low -power RFID tag is limited to [9 7 ][98], the total load current can be estimated by including the numerical calculation of Ic in the previous section. S ince the system only requires microamps to

PAGE 98

98 Figure 4 9. Calculated multiplier parameters based on charge equilibrium for four input amplitude values: (a) The I -V characteristic of the unit diode device, (b) required diode size normalized to 10x0.12um2, (c) conversion efficiency charge Cs for a data rate of approximately 1000bursts/sec, we specify the worst -case total load current Io H erein the diodes in the multiplier cells are implemented using diode -connected low -Vth PMOS in separate N -wells. To simplify the design procedure, all diodes are normalized to the unit -sized device measuring 10 x 0.12 m2, and a diode measuring M x 10 x 0.12 m2 is assumed to generate M times the current when same bias voltage is applied. The corresponding I -V curve of a unit sized diode -connected PMOS is shown in Figure 4 9( a ). Based on the diode I -V profile the integral equation (4 9) is solved for four different input voltage amplitudes ranging from 185mVp to 245mVp, and the resulting siz e of the diode connected PMOS is plotted versus the number of stage s N, as shown in Figure 4 9(b). The plot shows that if the input amplitude is larger, for a determined N, smaller diodes are required to 0.2 0.3 0.8 1.3 1.8 2.3 2.8 1 0.5 0 0.5 1 I diode ( mA ) V diode (V) 0 10 20 30 40 50 4 5 6 7 8 9 10 Diode size(xUnit) No. of Stages 185mVp 205mVp 225mVp 245mVp 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0.22 0.24 4 5 6 7 8 9 10 Conversion efficiency LD No. of Stages 185mVp 205mVp 225mVp 245mVp A unit sized Low VthDiode connected PMOS: Measuring 10umx0.12um

PAGE 99

99 Figure 4 10. Calculated requirement for external voltage to activate the ingested tagging device: (a) Input impedance of the tagging IC, (b) required external voltage from the transmitter. sustain the same load current as expected. Similarly, decreasing the number of multiplier stages requires l arger diodes to sustain the output current. The corresponding ld for each studied case are calculated using (4 10) and plotted in Figure 4 9(c). For a specified Vin, there always exists a pair of N and device size that maximize ld by trading off the power consumption on a single diode -connected device and the total number of diodes. It also shows that the maximum achievable efficiency increases as the input amplitude becomes larger, because of the smaller device size and smaller number of diod es required to generate the same Vhigh and to support Io. The input resistance of the multiplier is calculated for all the studied cases, using the power -voltage relationship Vin 2/2Ptotal, and plotted in Figure 4 10(a). In addition, the required VTX for ea ch case is derived from Fig.3 at 13.56MHz, and plotted in Figure 410(b). The analysis indicates that the minimum VTX is approximately 21Vpp f or the targeted 1.2V Vhigh and ~ 10 A load current. The chosen RF DC multiplier configuration is comprised of seve n 100 150 200 250 300 350 400 450 500 4 5 6 7 8 9 10 R IC( ) No. of Stage 185mVp 205mVp 225mVp 245mVp 20 22 24 26 28 30 32 34 36 4 5 6 7 8 9 10 V TX at 13.56MHz( Vp p ) No. o f Stage 185mVp 205mVp 225mVp 245mVp Design region

PAGE 100

100 stages, with each diode-connected PMOS measuring 7 x 10 x 0.12 m2. The simulated LF-DC power conversion efficiency ld from the multiplier input to the output in this case is ~ 18%. 4.4 Circuit Implementation Figure 4-11. The system-level block di agram of the asymmetric tagging IC. 4.4.1 System Architecture Figure 4-11 shows a block diagram of the propos ed RF tagging IC, or Electronic Burst (Eburst) device. The tag is powered by applying a 13.56MHz signal to a pair of pads in direct contact with the skin. The signa l travels through the communicati on medium, in this case the human body, which functions as a passive electr ical pathway. Low frequency bio-compatible pads on the transponder capsule are used to capture the transmitted power, energizing the microchip. This human-body channel also serves as a downlink, providing command functions (such as setting the burst frequency) to the ta gging IC. As previously described, an antenna on the surface of the pill is designed for UHF fre quency operation and uplinks to an external detector. The captured low-frequency energy drives an RF multiplier, composed of multiple CsPower detector RF DC Multiplier Demod. Register + LogicData Clock V+ VAnt+ Ant Power_ENData_EN V+ VV+ V-

PAGE 101

101 differential RF DC converter stages, which powers the chip and supplies current to a storage capacitor Cs until the supply voltage exceeds a programmed value set by the power level detector. The power detector is used to monitor the storage capacitor voltage and signals the device when ready for transmission. An output transmitter generates short bursts at a pproximately 915MHz when both the power level and correct downlink data ha ve been received. The frequency of the RF burst is controlled by a 6 -bit digital word, externally programmed by the user via the downlink data stream. A 3 -bit cyclic redundancy -check (CRC) code is used for data integrity and the TX burst is generated only after the CRC has been verified. 4.4.2 RF -DC M ultiplier Unlike multiplier s designed at UHF frequency [93 9 6 ], the voltage drop on the series capacitor Cse must be considered in this application due to the low operating frequency operation. In this prototype, we use 13.56MHz as the powering frequency primarily as a tradeoff between channel attenuation and the area constraint of the multiplier design Following the study in section III B, the RF DC multiplier is comprised of seven differential stages, and the required diodes are implemented using diode -connected low -Vth PMOS devices each measuring 7 x 10 x 0.12 m2, as shown in Figure 4 12. The required value of Cse for a targeted voltage drop can be derived by considering a simple voltage division circuit consisting of Cse itself and a resistance Rrec representing the power consumption of all the circuit blocks (i.e. Prec) connecting to Cse. Utilizing the circuit mo del in Fig ure 4 12, the value of Rrec can be estimated by calculating Prec. o DC d diode se in rec se in recI V P V V P V V R 2 22 2 (4 12)

PAGE 102

102 Figure 4 12. Schematic of the RF DC multiplier optimized for 13.56MHz. If we limit the voltage drop across the coupling capacitor to 10%, the series capacitance Cse can be calculated using (4 12) as ~18pF. In order t o prevent the parasitic capacitance s to ground from degrading the multiplier performance, all the coupling capacitors are implemented using metal to -metal capacit ors The metal to -metal capacitance density in the CMOS process that is 2 [99], which corresponds to Cse of ~8 000 2. Given that 7 stages are used, and each differential stage has 4 coupling capacitors, the total are occupation is ~ 224000 2. 4.4.3 Power Conditioning The DC voltage established on Cs is monitored by the power level detector circuit. Once the voltage on Cs exceeds Vhigh (1.2V), the power detector allows the tagging device to enter the burst phase, and a short 915MHz burst of radiation is generated. In this phase, electrical charge is continuously drawn from Cs, therefore decreasing the stored voltage. The power detector shuts off the burst when the voltage falls belo w Vlow (0 .6V). Subsequently the tagging device re -enters the charging phase for the next burst. RFIN+ 18pF Stage 1 Stage 7 18pF 18pF 18pF 18pF 18pF 18pF RFIN18pF 18pF 10 A 2.4nF 1.2V

PAGE 103

103 Figure 4 13. The schematic of the power conditioning circuits. The schematic of the power detector is shown in Fig ure 4 13. It consists of a voltage sensing leg to monitor the voltage on Cs (also used as the DC supply voltage for the integrated circuit blocks), a bias generator to generate a supplyindependent reference Vref, and a two stage differential compa ra tor with hysteresis The monitoring leg is simply a voltage divider composed of two diode -connected PMOS devices (P1 and P2) and a 2.5M At low supply voltages the two stacked PMOS operate in the sub-threshold region with high impedance; the drain voltage of P2 (Vd2) is therefore pu lled to V As the supply voltage gradually increases, both PMOS devices start to enter the active region. In this case, the 2.5M dominant, which brings Vd2 close r to V+. The two -stage differential comparator with hysteresis is then used to compare Vd2 with Vref, and the required operating voltage range (0.6 to 1.2) is generated from the positive feedback introduced by the cross coupled pair M2 and M3 [ 100 ]. Since the steering current through P3 is limited, a large voltage range can only be achieved by keeping a large ratio (1:32) between M2, M3 and M1, M4. Except P1 and P2, all transistors of Bias_P32:1 1:32 1 : 2 Power Enable P1 M2 M1 M4 M3 P2 R Vref 420K Ohm 1.5MOhm IbM6 M5 Bias generator Power detector P4 P5 V+ Bias_N V+ V IsteerP3

PAGE 104

104 the power detector are biased by current mirrored from the refer ence generator circuit to limit the total current consumption to ~400nA. The required voltage reference Vref and bias voltage Vbiasp for the power detector is generated by the bias generator circuit. Considering that the supply voltage var ies throughout t he tag operation, the basic bootstrapped topology [10 1 ] is chosen because of its relative supplyindependence and simplicity. The design is further simplified by the fact that the tag operates inside of a temperature stable environment, the human body. Ib, the current flowing through M6, is utilized as a unit current and copied to other circuit blocks as a current bias reference. The supply independent Vref is then produced by passing 2x Ib through a 1.5 M 4.4. 4 Modulator The schematic of the bur st modulator composed of a digital controlled oscillator (DCO) core and a class D power amplifier (PA ), or driver, is shown in Figure 4 14. Two conditions have to be satisfied simultaneously to trigger the 915MHz burst: The voltage across Cs must reach th e pre -determined value Vhigh, and the frequency tuning word from the recovered data packet must be received and verified. In addition to the requirements listed above, t he burst must be constrained within the regulated frequency band. Because the modula tor is operating under a varying supply voltage (from Vhigh to Vlow in the burst phase), a negative -gm LC oscillator is used in this design to generate the required 915MHz oscillation. This implementation i s relative ly insensitive to changes in the DC supply voltage when compared to ring oscillators T he chosen oscillator topology utilizes only a cross -coupled NMOS pair used to produce the required negative resistance, with no need for an additional PMOS pair that would have severely limited supply ope rating range. This increased supply voltage variation and range comes at the expense of

PAGE 105

105 Figure 4 14. Schematic of the RF modulator composed of the DCO and the class D PA. l arger phase noise. However, since t owards the end of each burst the oscillator m ust function properly with a supply voltage of ~0.6V (Vlow) it is a reasonable tradeoff for this application To a void a complex and power hungry frequency control scheme s (such as a phase locked loop), digital bits are used to tune the oscillation freque ncy through a set of capacitor arrays placed symmetrically on both side s of the LC oscillator. In total, 6 bits are used to control the oscillating frequency, which is tunable within a ~200MHz frequency range the 915MHz ISM band. In applications requiring extremely low power consumption, a switching-mode power amplifier (PA) is the natural choice because of its high power conversion efficiency [102 ]. In this design, a class D PA is chosen primarily because of its simplicity and robustness of operation. The class D driver consists of two MOS devices connected in a fashion similar to a static inverter. Driven by the DCO, the two MOS devices are alternately switched on and off, ClassD Power Amplifier Dummy buffer Antenna V+ V 3.5pF 10nH 10nH Power_EN Data_EN V Bias 3.5pF CTUNE <5:0> 1x 2x 32x 1xC=128fF B0 B1 B5 6 bits control V+ > 1.2V Power_EN Packet received Data_EN 75/0.2 168/0.2 L Rant

PAGE 106

106 providing a current flowing either from the power supply to the capsule antenna, or from the capsule antenna to the IC ground. Energy is therefore radiated from the capsule antenna at the same frequency as the driving signal in the form of a square wave A 2ndorder band pass filter formed by an integrated capacitor and the inductive ca psule antenna help to remove the harmonics and constrain the UHF radiation within the specified frequency band. T he power conversion efficiency of the class D driver is key to the design. T he conversion efficiency PA, of the PA, can be defined as the ra tio of the power transmitted to Rant and the total power drawn from the power supply, and is mainly limited by the parasitic s of the switching MOS devices, including the junction capacitance and the non-zero resistance Ron. Because the dynamic charging los s of the parasitic capacitance is proportional to the square of the supply voltage and the conduction loss of Ron is inversely proportional to the supply voltage, there exists an optimal driver size that minimizes the total PA power loss and achieves the m PA [102 ]. Considering that the parasitic losses of MOS switches varies with the supply voltage, the class PA during the entire burst phase. Since the supply voltage is allowed to vary from Vhigh to Vlow, i n order for the modulator to generate the targeted 10dBm output power even with the smallest supply voltage, we first optimize the class PA at 0.6V. The optimized PA design is then verified over the other supply voltage s, by determining whether the PA output power meets the design target. The optimization procedure of the clas s -D PA is demonstrated in Figure 4 15. According to [ 102 ], for a targeted output power Pburst, the antenna inp ut resistance Rant can be uniquely

PAGE 107

107 determined by the supply voltage Vdd and the parasitic resistance of the switching device Ron, as in ( 4 13). on burst burst on burst antR P V V P V V R P V V R 22 2 2 2 2 2 (4 13) Fig ure 4 15. Class D PA optimization: (a) required loading resistance Rant to generate 13.5dBm DH with 0.6V supply, (c) the output power with various supply voltage when Rant=130 DH with various supply voltage when Ra nt=130 Based on ( 4 13), the required Rant to generate a 10dBm Pburst is calculated and plotted v ersus the driver size in Figure 4 15( a )PA clearly demonstrates the trade -off (a) (b) 14 12 10 8 6 4 2 0 20 40 60 80 100 P rad ( dBm ) with R ant = 130 Width of the driver NMOS( m) 0.6V 0.8V 1.0V 1.2V 0.3 0.35 0.4 0.45 0.5 0.55 0.6 20 30 40 50 60 70 80 90 100 PA efficiency ) with R ant = 130 Width of the driver NMOS( m) 0.6V 0.8V 1.0V 1.2V (c) (d) 80 90 100 110 120 130 140 150 160 170 180 60 70 80 90 100 Required R ant ( ) with V dd =0.6V Width of the driver NMOS( m) 0.415 0.42 0.425 0.43 0.435 0.44 0.445 0.45 0.455 60 70 80 90 100 PA efficiency with V dd =0.6V Width of the driver NMOS( m) Maximizing PA efficiency P cap dominant P res dominant

PAGE 108

108 between the conduction loss and the charging loss of the cl assD driver. T he calculated efficiency reaches a maximum of 2 and with a resistance Rant of ~130 The optimized Rant is then simulated with other supply voltage values (0.8V, 1.0V and 1.2V) in order to verify compliance with the PA output power specification PA and Pburst are plotted in Figure 4 15( c ) and ( d ). The figure shows that PA is maximized with relatively smaller driver sizes and high supply voltages T h is trend can be explained by considering th at dynamic charging loss es decrease with smaller switch sizes (and thus smaller parasitic capacitance) but eventually higher switch resistances dominate. Although the class -D PA sometimes operates below the optimized effi ciency (because of the varying supply voltage), simulation results show that the optimized driver can generate the required -10dBm power during the entire burst phase. Figure 4 15( d ) shows that higher voltages yield greater output power even though the PA is not optimized at these voltages. RF can be estimated with consideration of the completed modulator design. By combining the simulated power consumption of both the DCO and the classRF is found to be ~5%. 4.4. 5 Demodulator Besides providing the required activation power, a data packet including the frequency tuning word is sent to the tagging device through the low frequency human body channel. For constant and reliable power to the tagging device, a p ulse -position-modulation (PPM) scheme is used to encode data with narrow gaps with each lasting approximately 3.25s. The modulated data packet is composed of 23 bits and is 1.136ms in total duration. Since the pulses occur at approximately 30% or 70% of t he bit time duration, the full rate non return -to -zero (NRZ) clock

PAGE 109

109 edge information is readily available, and automatically synchronized with the transmitted data. Example waveforms are shown in Fig ure 4 16. The schematic of the integrated receiver includ ing the demodulator and clock data recovery (CDR) module is also shown in Figure 4 16. The demodulator extracts the envelope of the power carrier, and converts the detected notches into rail to rail pulses for further processing. The extracted envelope is copied into two different low -pass -filtered signal paths, one having a ~10x lower cut off frequency Utilizing the output of the signal path with lower cutoff frequency as the voltage reference, the full -swing PPM pulse train Vdemod can be recovered by comparing the output of the other signal path with the reference. The required clock is t hen generated by simply passing the full -swing Vdemod through a rising -edge activated toggle flip -flop (T FF) T he transmitted information e mbedded in the time domain is converted into a voltage by charging capacitor Cp with constant current Ip. To achieve fast switching and minimize charge integration Fig ure 4 16. T he demodulator and the clock and data recovery (CDR) circuits. VdemodClock VPNRZ Data30% 70% Carrier Bit 1 Bit 0 30% 70% Pad1 Pad2VENV IpCPVTHVP Sample SW1 T FF ClockSampleVdemod FF NRZ dataDemodulator Clock and data recovery Charge pumpIpM7 M8 P5 P6 P7 P8 B B 1x 10x V+Clock V V V+ V -Clock Comp.

PAGE 110

110 error s such as charge injection and clock feed -through, the charge pump proposed in [ 103 ] is utilized. The circuit is compose d of three pairs of MOS devices : NMOS pair M7 and M8 act a s the current switching pair, steer ing current through either pair P7 and P8 or directly through pair P5 and P6. When the clock signal is high, the bias current Ip is mirrored to P6 which charge s Cp. After the discharging signal is received, Ip is steered to P8 pull ing up the gate of P6 gradually to the power supply, thus stopping the charging process. At the same time, SW1 is turned on and quickly discharg es the capacitor. A sampling comparator sa mples the voltage established on Cp at the falling edge of the clock and compares it with a pre -determined voltage reference to recover the NRZ data. A small sampling window is generated by adding a delay in between the charging and discharging p hases, ens uring a constant signal level and leaving time for an accurate comparison The recovered bit is then stored in a latch composed of two crossed coupled NAND gates and synchronized to the clock rising edge at the beginning of the next bit period. 4. 5 Experi mental Results The proof -of -concept IC was fabricated using a CMOS 130nm process. Figure 4 17 shows the micrograph of the fabricated IC with a total chip area of 0.8 x 1.5mm2. The low frequency RF DC multiplier and demodulator occupies 40% of the chip spac e, with most of the area utilized by the twenty -eight series capacitors (18pF each) included in the multiplier. Based on the estimated DC -RF (~5%), the storage capacitor Cs is chosen to be 2.4 nF and is implemented using PMOS gate capacitance because of its high capacitance density. 4.5.1 Test Bench R esults T he basic functionality of the fabricated IC device was first verified via probe station measurement The measured sensitivity of the tagging IC is shown in Figure 418. The s ensitivity is defined as the minimum input power that results in ~1.2V DC voltage on Cs. The

PAGE 111

111 Figure 4 17. Micrograph of the tagging IC. Fig ure 4 18. The measured sensitivity of the fabricated tagging IC. tagging IC require s higher input power for activat ion because of increased voltage dro o p on the series capacitance Cse, resulting in lower RF DC efficiency at long switching periods. The 1500um800um DCO + Class D PARF DC Multiplier Demodulator CDR+ FIFO MOSCAP (Storage caps) 14 12 10 8 6 4 2 0 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 Sensitivity (dBm) Frequency (MHz) 8.7dBm @ 13.56MHz

PAGE 112

112 Figure 4 19. Measured time -domain response of the demodulator and CDR Figure 4 20. Tunability of the DCO: (a) Measured tuning range with 6 bit control word, (6) tuning of DCO around 915MHz. sensitivity reaches a minimum value of 12dBm when operated at approximately 50MHz Above this frequency, the effective parasitic capacitance at the input of the mult iplier becomes the dominant loss mechanism degrading the sensitivity. At the specified operating frequency of 13.56MHz, the sensitivity measured 8.7dBm. The input resistance of the tagging IC with the 8.7dBm input was characterized as 328 amoun ts to a 15.6% deviat ion from the theoretical study presented in Figure 4 11 and discussed in section 4.3 Supply voltage Recovered data Burst enable 600mV 1.2V 1.136mS 23bit Packet header 915MHz 910MHz 905MHz 920MHz 925MHz Frequency 915.6MHz ~3MHz ~3MHz 905 910Frequency (MHz)915 920 ~3MHz ~3MHz 915.6MHz(a) (b) 800 820 840 860 880 900 920 940 960 980 1000 0 8 16 24 32 40 48 56 64 Frequency (MHz) Tuning Word 925

PAGE 113

113 Figure 4 21. Measured DCO phase noise at 915.6MHz with 1.2 supply voltage. The basic functionality of the demodulator and CDR circuit was tested by feeding the 23 bit data packet required for bursting directly into the tagging IC in a periodic manner The data packet was sent at a data rate of ~20Kbit/s, and the RF burst was ther efore enabled every 1.136mS. Four complete oper ating cycles are shown in Figure 4 19. As shown in the figure, Burst enable triggers the 915MHz burst, signaling that the supply voltage has reache d 1.2V and at the same time, the 23-bit data packet has be en verified by the CRC check. Figure 4 20 shows the characterized tunability of the DCO. All 64 possible combinations of the 6 -bit frequency tuning word were covered in this measurement. The measurement result shows that the DCO can be tuned from 805MHz t o 985MHz, and the frequency tuning resolution within the 915MHz ISM band is ~3MHz. As an additional test, a stable DC supply of 1 .2 V was used directly, temporarily replacing the RF DC block. The phase noise of the DCO when tuned to 915.6MHz was characte rized by 120 110 100 90 80 70 60 50 40 30 20 15 150 Phase noise (dBc/Hz) Frequency (KHz) 1000 -108dBc/Hz

PAGE 114

114 feeding the output to a spectrum analyzer with 50 input impedance and is plotted in Figure 4 21. At a 1MHz deviation from the center frequency, the measured phase noise is 108dBc/Hz. 4.5.2 Antenna Fabrication and Tag Assembly A small meanderin g antenna was designed to fit the surface of a 00-sized capsule while maximizing the PA efficiency. The antenna together with the two low frequency pads measure s 28 x 15.7mm2, and is screen printed onto a thin biocompatible polydimethylsiloxane (PDMS) substrate The printing substrate is flexible allow ing the printed antenna to be wrapped around standard capsules. The conductive ink used to create the printed antenna has a measured sheet resistance ~40 milliohms/sq/.5mil [ 104 ]. The antenna was characterized experimentally using a custom designed plastic probe station with impedance of 125 + j280 ( ) at 915MHz. The layout and measured input impedance are shown in Figure 4 22. (a ) ( b ) Figure 4 22. Small capsule antenna designed for maximizing the output power: (a) antenna geometry, (b) measured antenna impedance. In order to experimentally verify the E burst taggi ng device, a prototype electronic pill was assembled. Gold stud bumps were first placed on the pad openings on the fabricated IC. The bonded IC device was flipped over, and dipped into a very thin layer of conductive epoxy. 400 200 0 200 400 600 800 1,000 750 850 950 1050 1150 1250 1350 1450 Z in ( ) Frequency (MHz) Rant Xant 28mm15.7mm UHF antenna with Low frequency pads 125+j280( ) @915MHz

PAGE 115

115 Because the gold stud bumps are extruded above the IC surface, the IC surface will not be contaminated by the epoxy. The IC device is then flipped, accurately aligned and bonded to the printed pattern using a flip chip bonder. A second layer of PDMS was added to isolate the 915MHz antenn a from the surrounding environment (either the human body or phantom solution) while the low -frequency pads are exposed. The assembled tagging device is shown in Figure 4 23. Figure 4 23. Picture of the tagging device assembly. 4.5.3 Characterization of the Assembled Tagging Device The radiated power from the assembled tagging device was characterized using the test setup shown in Figure 424(a). The tagging device was activated by directly feeding the 13.56MHz power carrier and the 23 bit data packet in to the low frequency pads, and the UHF bursts were therefore periodically generated. A folded dipole antenna placed ~2cm away from the tagging device was used as the receiving antenna, and the captured power spectrum was displayed on a spectrum analyzer. B y approximating each burst as a square -shaped pulse with

PAGE 116

116 constant amplitude, according to the peak locations in the power spectrum, the time duration of the burst can be deduced as ~350nS. Considering that burst period is 1.136mS, the peak received power i s ~ 10dBm. (a) (b) Figure 4 24. Characterization of Pburst, the radiated power from the assembled tag: (a). Measurement setup; (b). de -embedding the channel loss and receiving antenna mismatch. To estimate the radiated power Pburst from the tagging device, the pathloss of the uplink channel and the mismatch loss of the receiving antenna must be de -embeded from the measurement setup. For this purpose, a second measurement was carried out, as shown in Figure 4 24(b). This time, the assembled tagging de vice was replaced by the capsule antenna, and the power loss from the signal generator to the spectrum analyzer was characterized to be ~18dB. By Spectrum Analyzer Assembled tagging device 860 870 880 890 900 910 920 930 940 950 960Frequency (MHz)100 505dB/div ~5MHz 76dBm Burst Spectru mPowering measuring setup Measured burst spectrum Spectrum Analyzer Signal Generator Balun Uplink channel

PAGE 117

117 Figure 4 25. Prototype e -pill verification using the phantom solution. adding back the mismatch loss of the capsule antenna, the communication passloss and the receiver mismatch loss together can be calculated as ~10.5dB. Therefore, Pburst is estimated to be approximately 0.5dBm. Table 4 2. Summary of the measured prototype e -pill performance Technology UMC 130nm CMOS Operating range ~0.6V -~1.2V Down/ Up link frequency 13.56MHz/915MHz Data rate ~20 Kb/s (Downlink) Burst Duration ~35 0ns Burst power 0.5dBm VCO tuning range 805MHz 985 MHz S ensitivity for activation 8.7dBm @ 13.56MHz Chip area 1500m x 800m The packaged prototype e -pill was experimental ly verified as shown in Figure 4 25. Phantom solution contained in the same Rubbermaid container was utilized to mimic the human torso. An amplitude modulated RF signal generator was used to transmit the required 13.56MHz 90dBV 30dBV10dB/div Tag IC Phantom Solution 13.56MHz Power link 915MHz link 50 s /div 1V/div 50 0ns /div In+ In (In+) (In ) (In+) (In ) In+ In 1 0 1 1 0 1 Input data sequence 2V/div 50 s /div 1V/div 50 0ns /div In+ In (In+) (In ) (In+) (In ) In+ In 1 0 1 1 0 1 Input data sequence 50 s /div 1V/div 50 0ns /div In+ In (In+) (In ) (In+) (In ) In+ In 1 0 1 1 0 1 Input data sequence 2V/div 50 s /div 1V/div 50 0ns /div In+ In (In+) (In ) (In+) (In ) In+ In 1 0 1 1 0 1 Input data sequence RF source 13.65MHz Oscilloscope MXA signal analyzer 0mS 10mS 1.136mA 62dBV

PAGE 118

118 power and data package, and the two transmitting pads placed against the phantom box were directly contacting the solution from the inside. The modulated power carrier pi cked up by an additional pair of pads immersed on the opposite side of the saline box was first displayed on the oscilloscope and verified. While the voltages generated in each probe, labeled In+ and In -, were nearly in phase, a sufficiently large differen tial voltage (In+ In ) was induced across the probes to power the tag. The prototype e -pill was then immersed in the center of the phantom box (~20cm from the sidewalls, ~15cm from the top), and the generated 915MHz short bursts were captured by a receiv ing folded -dipole antenna placed outside of the phantom container. A vector signal analyzer with a built in demodulation function (Agilent N9020A) was used as the RF receiver. With approximately 2 8 Vp -p 13.56MHz carrier transmitted from the RF transmitter, a periodic downlink data packet activated the transponder at 1.136ms intervals, generatin g RF bursts of approximately 350 ns in duration and a measured signal level of 62dBV (at the receiving antenna). The performance of the prototype e pill is summarized in Table 4 2. 4.6 Conclusions An E -burst RF capsule tagging system is proposed for medication capsule monitoring. The RF tag utilizes an asymmetric powering and communication scheme to circumvent problems associated with signal attenuation inside the huma n body and poor radiation efficiency of electrically small antennas. Since the power levels required to activate a tag are orders of magnitude larger than what is detectable externally, we employ an asymmetric RF link to energize the transponder at low fre quency and transmit RF bursts at higher frequencies. Based on the study of the human body channel, the design of the tagging device operating in 13.56MHz/915MHz ISM bands is optimized. The proof -onconcept device was fabricated in 130nm CMOS, and experimentally validated in phantom solution.

PAGE 119

119 CHAPTER 5 SUMMARY AND FUTURE WORKS 5.1 Summary of th is D issertation Medication compliance is very important to pharmaceutical clinical trials and it is desirable to develop new methods besides the direct observation method that can conveniently and reliably monitor the medication compliance. Small electronic devices attached to the outer surface of standard capsule can potentially serve as a cost -effective method of validating medication compliance via electronic detection of an ingested pill inside the digestive tract This work intends to prove the possibility and report our effort to design suc h an electronic device The possibility of de tecting the e -pill was demonstrated through FDTD simulations of small dipole and coil antennas placed inside of the human body. The small antennas were resonated to various FCC regulated bands by shunting an additional reactance across the antenna terminal s T he simulation results suggest that the maximum radiation appear s to be between 600MHz and 900MHz, and the resulting one -way path loss is about 30dB. To verify the simulation results, we measured the transmission loss between a 8mm-coil antenna and a re ceiving antenna through the phantom solution, which is used to mimic the electrical properties of human torso. The coil antenna was wrapped aroun d the 00-sized capsule and was resonated to 402MHz by soldering a discrete chip capacitor across the terminals T he insertion loss which was characterized using a vector network analyzer was approximately 30dB at the resonant frequency of the capsule antenna The measurement result matches the FDTD simulation results therefore further proves the possibility of using capsule -sized antenna s for medication compliance monitoring. In order to evaluate the performance of the capsule -sized antenna s multiple structures fit ting the outer surface of the sta ndard capsule were designed and printed using either screen or

PAGE 120

120 pad printing method. In C hapter 3, we reported the first capsule antenna that was directly printed onto a capsule s outer surface. In order t o experimentally characterize the capsule antenna, a probe station was modified, and its top metallic cover was re placed by a plastic replica to minimize the ground reflection. The zigzag antenna was resonated to various FCC regulated frequency bands, and its radiation property from inside of the human body was studied using the FDTD method Based on the study, multiple inductive capsule antennas compatible with the popular backscattering technology at 915MHz were designed and printed Among those, t he antenna that covers ~300o axially around a 000 -sized capsule was printed using the biocompatible SNP ink and was pack aged with a commercial available IC to form a backscattering tag; the tag detection was demonstrated by using the corresponding reader device. The concept of pill to pill c ommunication was also discussed in C hapter 3. Including an IC chip brings more flexibility to the e pill design and therefore is key in this study. In Chapter 4, we discussed the associated challenges of reliably and constantly powering such a device, and proposed a new tag activation scheme : ingested electronic tagging device can b e powered entirely from galvanic electrodes using the human body as a transmission medium. Using this method, the ingested tagging device can just utilize two low -frequency pads in direct contact with the surrounding environment to extract the required act ivation energy Based on the study of capsule -sized antennas radiation from inside of the body, the tag -to reader link was designed at 915MHz ISM band to balance the radiation efficiency of ca psule -sized antennas and the transmission attenuation through th e human body. Based on the design methodology, a proof -of -concept asymmetric tagging device operating at 13.56/915MHz was fabricated using CMOS 130nm technology. T he functionality of the tagging IC was tested on probe station To form the prototype electro nic pill, the tagging IC was packaged with a custom designed antenna

PAGE 121

121 to maximize the output power from the tagging device The prototype e -pill was experimentally verified in the phantom solution. 5.2 Suggested Future Works The possibility of using a n in g ested small electronic device fi t t ing the outer surface of the standard capsule for medication compliance monitoring has been demonstrated in this work. The key blocks of the e pill such as the capsule antennas and tagging IC have been designed and proved functional. Suggested future works are described below In this work, the functionality of the e -pill was verified inside the phantom solution. Although the phantom solution was carefully designed to match the electrical properties of the human body, it has its limitations. Un like the phantom solution which is composed of a single material, t he real human body consists of various types of tissues, and the interface between different tissue materials, such as the muscles and bones, might bring additional concerns due to RF reflection and refraction. Also, the conductivity and dielectric constant of the human body not onl y change s with physical conditions such as temperature and humidity, they also vary from person to person Therefore, the RF channel and especially the powering channel through the human body should be studied more carefully and extensively. Once a stati stical channel model is established, the tagging IC can be redesigned by simply following the provided methodology discussed in Chapter 4 The functionality of the e pill can be further expanded by including an anti -collision mechanism. Unique IDs can be assigned to different pills to represent information such as the type of medication, or count of pills. Therefore when multiple pills are ingested, they can be distinguish ed from each others from inside of the human body.

PAGE 122

122 APPENDIX MATLAB CO DE FOR MULTIPLIER DESIGN The matlab code to calculate the corresponding diode size (normalized to a unit device measuring 10 x 0.12 m2) for the 130nm CMOS technology is provided in this section. The calculation flow is controlled by the main function, and the integration equation (4 9 ) is solved by the function opti. FUNCTION output = main % Defines the parameters needed for optimization Vout = 1.3; %V in = 0.26; % amplitude of the input sinusoidal signaling Iload = 4e 6; frequency = 915e6; Cpa_unit = 0.95e 15; % calculated parasitic capacitance to the ground for a unit diode. for ii = 1:20 Vinin(ii)= 0.13+(ii 1)*0.01 Vin = Vinin(ii); % input amplitude of the multiplier for ll = 1:8 N = ll + 6; % number of stages stages = N; Eload = Vout*Iload; %Energy delivered to the load per cycle result = opti (stages,Vout,Vin,Iload,frequency); size(ii,ll) = result(1); Ein(i i,ll) = 4*stages*result(2); % Energy input to the multiplier per cycle Pin(ii,ll) = Ein(ii,ll)+Eload Pindbm(ii,ll) = 10*log10((Ein(ii,ll)+Eload)*1000); effi(ii,ll) = Eload/Pin(ii,ll); % efficiency of the RF multiplier, given Vin, Vout, N. Rin_p(ii,ll) = Vinin(ii)^2/(2*Pin(ii,ll)); % Calculated input resistance of the IC device. end e nd FUNCTIO N output = opti (stages,Vout,Vin,Iload,frequency) % T his function calculate the required size of diode for a fixed number of stages N, fixed Vout, and fixed Vin. N = stages; % Number of stages ; Vo = Vout/(2*N); % considering differential RF -DC converter is used ;

PAGE 123

123 I = Iload; % This is the average load current for other blocks of the RFID chip; freq = frequency; Va = Vin ; % amplitude of the input RF signaling ; load lvtP0.txt; It = lvtP0(:,2)/28; instant current flowing through a single diode(the measured result includes 28 unit -sized diodes); Vt = lvtP0(:,1); % instant voltage appeared in the measured data ; Ql = I*1; % This is the charge flowing through the chip load; Normalized to frequencies. delta = 100; error = 2e 9; %This is the limit of integration error; did = 1000; %Divide the period into 'did' pieces, and using rectangular integration method. M_min = 0.01; M_max = 50000; while (delta >= error) deltaT = 1/did; sum = 0.0; %sum2 = 0.0; intev = (M_max + M_min)/2; for kk = 1:did Vins = Va*sin (6.28*(kk1)*deltaT) Vo/2; % interpolated voltage Vdiode; sum = sum + intev*interp1(Vt,It,Vins)*deltaT; vtest(kk) = Vins; itest(kk) = intev *interp1(Vt,It,Vins); end if (sum Ql > 0) M_max = intev; end if (sum Ql < 0) M_min = intev; end delta = abs(sum Ql); sum ; end %%Calculating the power consumption of the diode device with the calculated size E_cycle = 0.0; for ll = 1:did Vinsll = Va*sin(6.28*freq*(ll 1)*deltaT) Vo/2 E_cycle = E_cycle + Vinsll*intev*interp1(Vt,It,Vinsll,'linear')*deltaT; end output = [M_min,E_cycle*2]; % return calculation results to the main function.

PAGE 124

124 LIST OF REFERENCES [1 ] K.C. Farmer, Methods for measuring and monitoring medication regimen adherence in clinical trials and clinical practice, Clin. Ther., vol. 21, no. 6, pp. 1074 1090, 1999. [2 ] H.P. McDonald, A.X. Garg, R.B. Haynes, Interventions to enhance patient adherence to medication prescriptions: scientific review, J AMA vol. 288, no.22, pp. 2868 2879, 2002. [3 ] C. Nugent, D. Finaly R. Davies C. Paggetti E. Tamburini and N. Black, Can technology improve compliance to medication, From Smart homes to Smart care IOS press, 2005. [4 ] C.M. Hughes, Medication non adherence in the elderly, Drugs Aging. Vol.21, pp. 793811, 2004. [5 ] Tufts CSDD Outlook Report [Internet]. Tufts Center for th e Study of Drug Development (US) ; [updated 2010 January; cited 2007 April]. Available : http://csdd.tufts.edu/reports/description/tufts_outlook [6 ] W.S. Burke, and I.S. Ockene, Compliance in Healthcare and Research, Futura Publishing Company, Inc., Armonk, NY, 2001. [7 ] T he ultimate patient device [Internet] InforM edix (US); [updated 2009 March; cited 2007 April] Available : http://www.informedix.com/description.html [8 ] W.S. Bond, and D.A. Hussar, Detection methods and strategies for improving medication compliance, Am J Hosp Pharm., vol.48, Issue 9, pp. 1978 1988, 1991. [9 ] P. Nony, M. Cucherat, and J.P. Boissel, Revisiting the effect compartment through timing errors in Drug Administration, Tips, vol.19, pp.49 54, 1998. [10] E.J. Gallagher, C.M. Viscoli, and R.I. Horwitz, The relationship of treatment adherence to the risk of death after myocardial infarction in women, J AMA, vol.270, Issue 6, pp. 742 744, 1993. [11] M. Gibaldi, Failure to comply: a therapeutic dilemma and the bane of clinical trials, J. of Clin. Pharmacol vol.36, Issue 8, pp. 674 682, 1996. [12] W. Senterfitt. The Cha llenge of Medication Adherence. [ Intern et], 1998 June. [cited 2007 April]. Available : http://www.aegis.com/pubs/bala/1998/BA980601.html [13] Patient Compliance Monitoring [Internet]. AL E RT Technology Solution (US); [updated 2009 September, cited 2007 April ]. Available : http://www.alert 1.com/default.aspx

PAGE 125

125 [14] Promoting Adherenc e to HIV Antiretroviral Therapy [Internet]. Institute of Healthcare Improvement (US); [updated 2010 March, cited 2007 April ] Available: h ttp://www.ihi.org/IHI/Topics/HIVAIDS/H IVDiseaseGeneral/Tools/ [15] Medical Facts. [Internet]. The Pill Timer (US); [updated 2002, cited 2007 April] Available : http://www.thepilltimer.com/medicalfacts.html [16] N.M. Prakash, M.C. Brown, F.A. Spelman, J.A. Nelson, P. Read, M.M. Heitkemper, R.W. Tobin, C.E. Pope., Magnetic field goniometry: a new method to measure the frequency of stomach contractions, Dig Dis. Sci. vol.44, Issue 9, pp.1735 1740, Sept.,1999. [17] E. Stathopoulos, V. Schlag eter, B. Meyrat Y. Ribaupierre, P. Kucera, Magnetic pill tracking: a novel non -invasive tool for investigation of human digestive motility, Neurogastroenterol Motil, vol. 17, pp.148 154, Feb.,2005. [18] W. Weitschies, J. Wedemeyer, R. Stehr, L. Trahms, M agnetic markers as a noninvasive tool to monitor gastrointestinal transit, IEEE Trans. on Biomed Eng. Vol.41, pp.192 195, Feb., 1994. [19] J. Wenger J.G. Allee, M.S. Landy, Magnet -tipped tubes for studies of the stomach and duodenum, Am J Dig. Dis., vol. 1 5, Issue 4, pp.383 393, April, 1970. [20] Internal document Electronic Pills for Medication Compliance. [21] G. Herzer, Magnetic materials for electronic article surveillance, Journal of Magnetism and Magnetic Materials vol. 254 255, pp. 598602, 2003. [22] Electronic article surveillance [ Internet ]. Wikipedia (US); [updated 2010 March, cited 2007 April]. Available : http://en.wikipedia.org/wiki/Electronic_article_surveillance [23] How Anti -shoplifting Devices Work [Internet ]. Wikipedia (US); [updated 2010 March, cited 2007 April]. Available : http://electronics.howstuffworks.com/gadgets/high -tech gadgets/anti -shoplifting device.htm [24] J. Rod Gimbel, MD, and James W. Cox, Jr, MD, Elect ronic Article Surveillance Systems and Interactions with Implantable Cardiac Devices: Risk of Adverse Interactions in Public and Commercial Spaces ", Mayo Clinic Proceedings March 2007. [25] K.Finkenzeller, RFID handbook, Fundamental and applications in Contac tless Smart Cards and Identification, 2nd ed. West Sussex, U.K, 2003. [26] Nakamoto, H. Yamazaki, D. Yamamoto, T. Kurata, H. Yamada, S. Mukaida, K. Ninomiya, T. Ohkawa, T. Masui, S. Gotoh, K. Yokohama, A Passive UHF RF Identification CMOS Tag IC Using Ferroelectric RAM in 0.35 IEEE Journal of Solid -State Circuits, vol.42, pp. 101 110, Jan.2007.

PAGE 126

126 [27] Nakamoto, H. Yamazaki, D. Yamamoto, T. Kurata, H. Yamada, S. Muka ida, K. Ninomiya, T. Ohkawa, T. Masui, S. Gotoh, K. Kawasak, A Passive UHF RFID Tag LSI with 36.6% Efficiency CMOS Only Rectifier and Current -Mode Demodulator in 0.35/spl mu/m FeRAM Technology ISSCC, 2006, pp. 1201 1210. [28] V. Pillai, H. Heinrich, D. Dieska, P.V. Nikitin, R. Martinez, K.V.S. Rao, An Ultra Low -Power Long Range Battery/Passive RFID Tag for UHF and Microwave Bands With a Current Consumption of 700 nA at 1.5 V IEEE Journal of Circuits and Systems I, vol.54, pp.1500 1512, July,2007. [29] G uo, L.H. Popov, A.P. Li, H.Y. Wang, Y.H. Bliznetsov, V. Lo, G.Q. Balasubramanian, N. Kwong, D. L., A small OCA on a 10.5-mm2 2.45GHz RFID Tag design and integration based on a CMOS -compatible manufacturing technology IEEE Electron Device Letters, vol.27, pp. 96 98, Feb. 2006. [30] Ray Barnett, Ganesh Balachandran, Steve Lazar, Brad Kramer, George Konnail, Suribhotla Rajasekhar, Vladimir Drobny, A Passive UHF RFID Transponder for EPC Gen 2 with 14dBm Sensitivity in 0.13m CMOS, ISSCC,2007, pp.582 584. [31] Vijay Pillai, Harley Heinrich, David Dieska, Pavel V. Nikitin Rene Martinez, and K. V. Seshagiri Rao An Ultra -Low -Power Long Range Battery/Passive RFID Tag for UHF and Microwave Bands With a Current Consumption of 700 nA at 1.5V IEEE Trans. On Circuits and Systems I vol. 54, Issue 7, pp.1500 1512, Jul. 2007. [32] F. Kocer, M.P. Flynn, A long -range RFID IC with on RFIC pp 361 364, June 2005. [33] W. Jeon, J. Melngailis, R.W. Newcomb, CMOS passive RFID transponder with readonly memory for low cost fabrication, IEEE International SOC Conference, 2005. Proceedings, pp 181 184. [34] Wooi Gan Yeoh, Yeung Bun Choi, et.al. A CMOS 2.45 GHz radio frequency identification tag IC with read/write memory Radio Frequency integrated Circuits Symposium Jun. 2005, pp. 365 368. [35] Yao Huang Kao, Chia Chuan Liu, Hung -Chang Kuo, Study of Front End of CMOS RFID Tag with Inductively -coupled Broadband Antenna Asia Pacific Microwave Conference, Dec. 2007, pp.1 4. [36] A. Facen, A. Boni, CMOS power retriever for UHF RFID tags Electronics Letters vol.23, Issue 25, pp.1424 1425, Dec. 2007. [37] Ashry, A., Sharaf, K, Ultra low power UHF RFID t ag in 0.13um CMOS, International Conference on Microelectronics 2007, pp 283 286.

PAGE 127

127 [38] S. Mandal, R. Sarpeshkar, Low -Power CMOS Rectifier Design for RFID Applications IEEE Transactions on Circuits and Systems I Vol 54, Issue 6 pp.1177 1188, June 2007. [39] Yang Hong, Chi Fat C han, et.al. Design of passive UHF RFID tag in 130nm CMOS technology, IEEE Asia Pacific Conference on Circuits and Systems Nov.2008, pp. 1371 1374. [40] C.T. Rodenbeck, Planar Miniature RFID Antennas Suitable for Integration With Batteries IEEE Transactions on Antennas and Propagation Vol. 54, Issue 12, pp.3700 3706, Dec. 2006. [41] A. Galehdar, D.V. Thiel, S.G. O'Keefe, Antenna Efficiency Calculations for Electrically Small, RFID Antennas IEEE Antennas and Wireless Propagation Letters Vol. 6, pp.156 159, 2007. [42] G. Marrocco, The art of UHF RFID antenna design: impedance -matching and size reduction techniques IEEE Antennas and Propagation Magazine Vol.50, Issue 1 pp.66 79, Feb. 2008. [43] K.V.S. Rao, P.V. Nikitin, S.F. Lam, Antenna design for UHF RFID tags: a review and a practical application IEEE Transactions on Antennas and Propagation, Vol. 53, Issue 12, pp.3870 3876, Dec. 2005. [44] J. Xi, N. Yan, W. Che, X. Wang, H. Jian, H. Min, On chip antenna design for UHF RFID Electronics Letters Vol. 45, Issue 1 pp.14 16, Jan. 2009. [45] Wonkyu Choi, H.W. Son, Chansoo Shin, Ji Hoon Bae, Gilyoung Choi, RFID tag antenna with a meandered dipole and inductively coupled feed IEEE Antennas and Propagation Symposium Jul. 2006, pp.619 622. [46] Yuri Tikhov, Yongjin Kim, YoungHoon Min, Compact low cost antenna for passive RFID transponde r, IEEE Antennas and Propagation Symposium Jul. 2006, pp. 1015 1018. [47] Y. Tikhov, Yongjin Kim, YoungHoon Min, A novel small antenna for passive RFID transponder European Microwave Conference Oct. 2005, Vol. 1, pp.4 6. [48] Glen Patrick Flores, Binary Molecular Weight Polyvinyl Pyrrolidone -controlled Growth of Silver Nanoparticles for Application in Low -curing temperature Conductive Biomaterials, Ph.D. Dissertation, University of Florida, Gainesville, FL, 2008. [49] Keith Gurnett and Tom Adams, Ultra thin semiconductor wafer applications and processes III Vs Review Vol. 19, Issue 4, pp. 38 40, May 2006.

PAGE 128

128 [50] P.L. Drake and K.J. Hazelwood, Exposure related health effects of silve r and silver compounds: A review, Ann Occup. Hyg.vol.49, pp. 575 585, 2005. [51] R. Malcolm, et al., Silver acetate gum as a deterrent to smoking, Chest. v ol.90, pp. 107 111, 1986. [52] Allen Taflove and Susan C. Hagness Computational Elect rodynamics: The Finite Difference Time -Domain Method, 3rd ed. Artech House, Inc., 2005. [53] Z P Liao H L Wong, B P Yang, Y F Yuan, A transmitting boundary for transient wave analyses Sci. Sin. A 27, pp. 1063 1076, 1984. [54] E. H. Grant, The physical interaction of electromagnetic fields with people, in Proc. Conf. Electromag. Fields Human Health, pp. 1.2.1 1.2.5 1991. [55] M.A. Stuchly, S.S. Stuchly, "Dielectric properties of biological substances tabulated", Journal of Microwave Power V ol.15, No.1, pp.19 26, 1980. [56] James C. Boylan, Encyclopedia of Pharmaceutical Technology, 3rd ed. Marcel Dekker, Inc., 1990, Volume 2. [57] Lei Wang, T.D. Drysdale, D.R.S. Cumming In Situ Characterization of Two Wireless Transmission Schemes for Ingestible Capsules, IEEE Trans. On Biomedical Engineering, Vol. 54, Issue 11 pp. 2020 2027, Nov., 2007. [58] T. Ogura, Y. Furuya, S. Matuura HPMC capsules: an alternative to gelatin, Pharm. Tech Europe V ol.10, Issue 11, pp.32 42, 1998. [59] Remington, The Science and Practice of Pharmacy, 20th ed. Volume 1 Lipp incott Williams & Wilkins, 2000 [60] Gilbert S. Bankers, Modern Pharmaceutics, 4th ed. Dekker series, 2002. [61] Lawrence C. Chirwa, et al. Electromagnetic Radiation from Ingested Sources in the Human Intestine between 150 MHz and 1.2 GHz, IEEE Trans. on Biomedical Engineering, vol. 50, No. 4, pp.484 492, Apr. 2003. [62] C.A. Balanis, Antenna Theory: Analysis and Design, 2nd ed. John Wiley & Sons, 1997, Ch.2. [63] K. Schneider, A re look at antenna in band RCSR via load mismatching, Proc. IEEE Antennas and Propagation Society Int. Symp. pp. 1398 1401, June 1996. [64] R. Bancroft, Fundament al dimension limits of antennas Centuri on Wireless Technologies June 2002.

PAGE 129

129 [65] G. Marrocco, Gain optimized self -resonant meand er line antennas for RFID applications IEEE Antennas and Wireless Propagation Letters, vol.2 pp. 302305, 2003. [66] Design document of Alien H 2 generation chip [ Internet ]. Alien Technology (US); [updated 2010, cited 2008 March] Available: http://www.alientechnology.com/products/ [67] K. Kim, Design and Characterization of RF Components for Inter and Intra Chip Wireless communication, Ph.D Dissertation University of Florida, Gainesville, FL, 2000. [68] RFID in livestock [Internet ]. Essenrfid (US); [updated 2008, cited 2009 March] Available : http:// www.essenrfid.com/technologylivestock.htm [69] RFID -based systems [Internet] SurgiCount Medical (US); [updated 2009, cited 2009 March]. Available : http://www.surgicountmedical.com/rfid_based_system.html [70] N.M. Nathan, and H.R. Reid, Micropower Circuits for Bidirectional Wireless Telemetry in Neural Recording Applications, IEEE Trans on Biomedical Engineering, Vol. 52, No. 11, pp. 19501959, Nov. 2005. [71] K.D. Wise, G.E. Perlin, Y. Yao, K. Najafi, and A.M. Sodagar, An Implantable 64 Channel Wireless Microsystem for Single -Unit Neural Recording, IEEE J. of Solid state Circuits Vol. 4 4, No. 9, pp. 25912604, Sep. 2009. [72] J. R. Rao and G. Richter, Implantable bio -electrochemical power sources, Naturwissenschaften, Vol. 61, No.5, pp. 200 206, May, 1974. [73] Masaaki Fukumoto, and Yoshinobu Tonomura, Body coupled fingering: Wireless wearable keyboard, CHI7 pp.147 154, 1997. [74] Takashi Handa, etc., A very low -power consumption wireless ECG monitoring system using body as a signal transmission medium, Proceeding of the International Conference on Transducers, Solid-state Sensors and Actuators pp.1003 1007, 1997. [75] Masaaki Fuko moto, etc., A broad -band intra -body communication system with electrooptic probe, First International Conference on Application Design, pp.108 109, 2003. [76] IEEE C95.1 2005, IEEE Standard for Safety Levels with Respect to Human Exposure to Radio Frequency Electromagnetic Fields, 3 kHz to 300 GHz The Institute of Electrical and Electronics Engineers Inc., New York 2006. [77] S. Gabriel, R.W. Lau, and C. Gabriel, The dielectric properties of biological tissues: II. Measurements in the frequency range 10 Hz to 20 GHz Phys. Med. Biology Vol. 41, pp. 2251 2269, 1996.

PAGE 130

130 [78] H.P. Schwan, Electrical properties of tissue and cell suspensions, Advances in Biological and Medical Physics No.5, pp.147209, 1957. [79] D.C. Barber and B.H. Brown, Applied potential tomography, Journal of Physics E: Scientific Instruments vol.17, pp.723733, 1984. [80] K.R. Foster and H.P. Schwan, Dielectric properties of tissues and biological materials: A critical review Critical reviews in biomedical Engineering No. 1, pp.25 104, 1989. [81] Thomas Gurthrie Zimmerman, Personal area network (PAN) Master thesis Massachusetts Institute of Technology, 1995. [82] K. Hachisuka, A. Nakata, T. Takeda, Y. Terauchi, K. Shiba, K. Sasaki, H. Hosaka, and K. Itao, Development and performance analysis of an intra body communication device, International Conference on Transducers, Solid-State Sensors, Actuators and Microsystems, Vol. 2, pp. 1722 1725, 2003. [83] Michael Oberle, Low power system on-chip for biomedical application, PhD dissertation, ETH No.14509, IIS/ETH Zurich, 2002. [84] Derek P. Lindsey, Eric L. McKee, Maury L. Hull, and Stephen M. Howell, A new technique for transmission of signals from implantable transducers IEEE Transactions on Biomedical Engineering, Vol.45, No. 5 pp.614 619, 1998. [85] Keiuke Hachisuka, Teruhito Takeda, Yusuke Terauchi, Ken Sasaki, Hiroshi Hosaka, and Kiyoshi Itao, Intra body data transmission for the personal area network, Microsyst. Technol. Vol.11, pp.10201027, 2005. [86] Tomohiro Yamada, Takumi Uezono, Hirotaka Sugawara,, Kenichi Okada, Kazuya Masu, Akio Oki, and Yasuhiro Horiike, Battery-less Wireless Communication System through Hunan Body for in -vivo Healthcare Chip, Topical Meeting on Silicon Monolithic Integrated Ci rcuits in RF systems pp. 322325, 2004. [87] Keisuke Hachisuka, Yusuke Terauchi, Yoshinori Kishi, Ken Sasaki, Terunao Hirota, Hiroshi Hosaka, Katsuyuki Fujii, Masaharu Takahashi and Koichi Ito, Simplified circuit modeling and fabrication of intrabody communi cation devices Sensors and Actuators A: Physical, Vol. 130131, pp.322330, 2006. [88] Marc Simon Wegmuller, Intra body Communication for Biomedical Sensor Networks, PhD dissertation, ETH No.17323, ETH Zurich, 2007. [89] International Commission on Non Ionizing Radiation Protection (ICNIRP), Guidelines for limiting exposure to time -varying electric, magnetic, and electromagnetic fields (up to300GHz) 1997. [90] John G, Polaris, Digital Communications 4th edition, McGraw Hill, Inc., 2000, Ch.4

PAGE 131

131 [91] A. Mazzanti and F. Svelto, A 1.8 GHz injection locked quadrature CMOS VCO with low phase noise and high phase accuracy, IEEE Transactions on Circuits and Systems I, Vol. 53, N o. 3 pp. 554 560, Mar. 2006. [92] J. Hou, C. Chang, J. Su, T. Liou, S. Wong, and Y. Wang, A low supply voltage VCO implemented by a single common source 90nm CMOS transistor, IEEE Microwave and Wireless Components Letters, V ol. 1 7, N o. 1, pp. 1314 1319, Jan. 2007. [93] J. P. Curty, N. Joehl A model for -power rectifier analysis and design, IEEE Trans. on circuits and systems I, Vol. 52, No. 12, pp. 27712779, Dec. 2005. [94] Bergeret, E., Gaubert, J., Pannier, P., Gaultier, J.M., Modeling and Design of CMOS UHF Voltage Multiplier for RFID in an EEPROM Compatible Process IEEE Trans. On Circuits and Systems II Vol. 54, No. 10, pp.833837, Oct. 2007. [95] Jun Yi, Wing Hung Ki, and Chi Ying Tsui, Analysis and Design Strategy of UHF Micro -Power CMOS Rectifiers for Micro Sensor and RFID Applicati ons, IEEE Trans. on circuits and systems I Vol. 54, No 1, pp. 153166, Jan., 2007. [96] Giuseppe De Vita and Giuseppe Iannaccone, Design Criteria for the RF Section of UHF and Microwave Passive RFID Transponders, IEEE Trans. on Microwave Theory and Techno logy Vol. 53, No. 9, pp. 29782990, Sep. 2005. [97] Udo Karthaus and Martin Fischer, Fully Integrated Passive UHF RFID Transponder IC With 16.7-state Circuits,Vol.38, No.10, pp. 16021608, 2003. [98] Ray Barnett, Ganesh Balachandran, etc., A Passive UHF RFID Transponder for EPC Gen 2 with 2007 IEEE International Solid-State Circuits Conference pp.582583. [99] Internal documents for UMC130nm CMOS process. [100] Phillip E. Allen CMOS Analog Circuit Design, 2nd ed., Oxford University Press, 2002, Ch.8 [101] Behzad Razavi Design of Analog CMOS Integrated Circuits McGraw Hill, Inc., 2001, Ch.11 [102] Mihai Albulet, RF Power Amplifiers Noble Publishing, 2001, Ch.3 [103] E. Juarez Hernandez, A. Diaz -Sanchez, A novel CMOS charge -pump circuit with positive feedback for PLL applications The 8th IEEE International Conference on Electronics, Circuits and S ystems, pp. 349 352, 2001. [104] E xtremely conductive ink 12446, Creative Materials Incorporated.

PAGE 132

132 BIOGRAPHICAL SKETCH Hong Yu was born in Beijing, China. He attended the University of Science and Technology of China, Heifei, and earned his bachelors degree in electrical and computer engineering in 2001. After obtaining his bachelors degree, he came to the United States, and worked toward his Master of Science (M.S) degree at the University of Massachusetts, Amherst, where he designed a two -frequency, dual -polarized feed antenna for weather profiling radar In August, 2004, he came to the University of Florida to pursue his Ph.D. degree in electrical engineering. For his Ph.D. research, he worked under the supervision of Dr. Rizwan Bashirullah in the Integrated Circuit Research (ICR) group. He spent the first two years with his colleagues developing an implantable neural r ecording system which can amplify, process and wirelessly transmit recorded neural signals. His Ph.D topic is to design an electronic pill for medication compliance monitoring, which allows the capsule to be detected after being in gested into the human gas trointestinal tract. He has more than 10 publications in technical journals and conferences.