<%BANNER%>

Multiple Sensor Platforms for Hydrogen and Human Physiological Movement Sensing

Permanent Link: http://ufdc.ufl.edu/UFE0043697/00001

Material Information

Title: Multiple Sensor Platforms for Hydrogen and Human Physiological Movement Sensing
Physical Description: 1 online resource (101 p.)
Language: english
Creator: Yu, Xiaogang
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: doppler -- heartbeat -- hydrogen -- noncontact -- radar -- respiration -- sensor
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: This dissertation begins with a demonstration of the integration of multiple sensor techniques with state-of-the-art hydrogen sensing devices. The proposed multiple sensor system uses six Zigbee transceivers to collect hydrogen density information from the dispersedly deployed AlGaN/GaN high electron mobility transistor (HEMTs) differential sensing diodes. The collected hydrogen density information is transmitted wirelessly to the base station for data logging and tracking of each individual sensor. The software at the base station defines and implements the monitoring states, transitions, and actions of the hydrogen sensing system. The software also is able to warn the user of potential sensor failure, power outages, and network failures through cell phone network and Internet. Real-time responses of the sensors are displayed through a web site on the Internet. The sensing system has shown good stability for more than 18 months in an outdoor field test. After that, Chapter 3 is devoted to a presentation of the integration of multiple sensor techniques with noncontact vital sign detection. Using the multiple vital sign sensor platform, two-dimensional random body movement cancellation is achieved. The multiple sensor system includes four detectors, an 8-channel data acquisition module, and a computer for spectrum analysis. Each of the detectors consists of a radio frequency transceiver, a baseband analog circuit, and a power management circuit. The multiple sensor platforms also strengthen the detecting sensitivity on the respiration and heartbeat. A DC offset compensation algorithm is introduced to free the body movement cancellation from disturbance of unwanted DC offset. Experiments were performed with a human subject in laboratory environment. Results were analyzed to verify the improved detection performance at the presence of 2-D human body movement. The limitation of sensitivity improvement and body movement cancellation are demonstrated with simulation results. Chapter 4 details the hardware design of the individual portable Doppler radar for noncontact vital sign detection. Topics including RF transceiver board design, baseband signal processor design, sampling frequency selection guideline, and noise analysis for the receiver chain of the detector will be discussed. Finally, a system integration of noncontact vital sign detector with antennas on-board will be presented in Chapter 5.
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 Xiaogang Yu.
Thesis: Thesis (Ph.D.)--University of Florida, 2011.
Local: Adviser: Lin, Jenshan.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2013-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0043697:00001

Permanent Link: http://ufdc.ufl.edu/UFE0043697/00001

Material Information

Title: Multiple Sensor Platforms for Hydrogen and Human Physiological Movement Sensing
Physical Description: 1 online resource (101 p.)
Language: english
Creator: Yu, Xiaogang
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: doppler -- heartbeat -- hydrogen -- noncontact -- radar -- respiration -- sensor
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: This dissertation begins with a demonstration of the integration of multiple sensor techniques with state-of-the-art hydrogen sensing devices. The proposed multiple sensor system uses six Zigbee transceivers to collect hydrogen density information from the dispersedly deployed AlGaN/GaN high electron mobility transistor (HEMTs) differential sensing diodes. The collected hydrogen density information is transmitted wirelessly to the base station for data logging and tracking of each individual sensor. The software at the base station defines and implements the monitoring states, transitions, and actions of the hydrogen sensing system. The software also is able to warn the user of potential sensor failure, power outages, and network failures through cell phone network and Internet. Real-time responses of the sensors are displayed through a web site on the Internet. The sensing system has shown good stability for more than 18 months in an outdoor field test. After that, Chapter 3 is devoted to a presentation of the integration of multiple sensor techniques with noncontact vital sign detection. Using the multiple vital sign sensor platform, two-dimensional random body movement cancellation is achieved. The multiple sensor system includes four detectors, an 8-channel data acquisition module, and a computer for spectrum analysis. Each of the detectors consists of a radio frequency transceiver, a baseband analog circuit, and a power management circuit. The multiple sensor platforms also strengthen the detecting sensitivity on the respiration and heartbeat. A DC offset compensation algorithm is introduced to free the body movement cancellation from disturbance of unwanted DC offset. Experiments were performed with a human subject in laboratory environment. Results were analyzed to verify the improved detection performance at the presence of 2-D human body movement. The limitation of sensitivity improvement and body movement cancellation are demonstrated with simulation results. Chapter 4 details the hardware design of the individual portable Doppler radar for noncontact vital sign detection. Topics including RF transceiver board design, baseband signal processor design, sampling frequency selection guideline, and noise analysis for the receiver chain of the detector will be discussed. Finally, a system integration of noncontact vital sign detector with antennas on-board will be presented in Chapter 5.
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 Xiaogang Yu.
Thesis: Thesis (Ph.D.)--University of Florida, 2011.
Local: Adviser: Lin, Jenshan.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2013-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0043697:00001


This item has the following downloads:


Full Text

PAGE 1

1 MULTIPLE SENSOR PLATFORMS FOR HYDROGEN AND HUMAN PHYSIOLOGICAL MOVEMENT SENSING By XIAOGANG YU A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR TH E DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2011

PAGE 2

2 2011 Xiaogang Yu

PAGE 3

3 To my parents

PAGE 4

4 ACKNOWLEDGMENTS I would like to express my sincere gratitude to my advisor Dr. Jenshan Lin for his advice, encouragemen t, and mentoring throughout my PhD study. I have truly enjoyed patience and kindness to other people are things I admire greatly I would also like to thank my committee members, Dr. Fan Ren Dr. Huikai Xie and Dr. Eric McLamore for their time and precious comments. I am also thankful to my colleagues ( Changzhi Li, Yan Yan, Mingqi Chen, Zivin Park, Raul Chinga) in the Radio Frequency Circuits and Systems Research Group, for all the h elp and happiness they offered. I would like to thank my parents for their encourageme nt and unconditional support. I dedicate this d issertation to my family, whose love gives me the courage to my life.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 8 LIST OF ABBREVIATIONS ................................ ................................ ........................... 11 ABSTRACT ................................ ................................ ................................ ................... 12 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 14 1.1 Backg round ................................ ................................ ................................ ....... 14 1.2 Recent Progresses on Hydrogen Sensing ................................ ........................ 15 1.3 Recent Progresses on Physiological Movement Sensing ................................ 17 1.3.1 Theoretical Breakthroughs ................................ ................................ ...... 17 1.3.2 RF Front end Architectures ................................ ................................ ..... 17 1.3.3 Ad vances in Signal Processing Techniques ................................ ............ 21 1.3.4 Miniaturization and System on chip ................................ ......................... 23 2 MULTIPLE WIRELESS SENSOR PLATFORM USING ALG A N/G a N HIGH ELECTRON MOBILITY TRANSISTOR DIFFERENTIAL DIODE SENSORS ......... 24 2.1 Hydrogen Sensors with Different Fabrication Technologies ............................. 24 2.2 Experiments with Differential Sensor Pairs ................................ ....................... 25 2.3 Wireless Multiple Sensor System ................................ ................................ ...... 27 2.3.1 System Overview ................................ ................................ ..................... 27 2.3.2 Detection Circuits ................................ ................................ .................... 29 2.3.3 Zigbee Wireless Network ................................ ................................ ......... 30 2.3.4 W ireless Sensor Network Monitoring Software ................................ ....... 31 2.3.5 Monitoring States, Transitions, and Actions ................................ ............ 32 2.3.6 Packages ................................ ................................ ................................ 32 2.4 Field Test ................................ ................................ ................................ .......... 36 2.5 Summary ................................ ................................ ................................ .......... 37 3 MULTIPLE DOPPLER RADAR SENSOR PLATFORM F OR TWO DIMENSIONAL HIGH SENSITIVITY HUMAN PHYSIOLOGICAL MOVEMENT DETECTION ................................ ................................ ................................ ........... 39 3.1 Challenges of Body Movements ................................ ................................ ....... 39 3.2 Principle of Noncontact Vital Sign Detection ................................ ..................... 40 3.3 Two Dimensional Random Body Movement Cancellation ................................ 44

PAGE 6

6 3.4 Sensitivity Improvement Using Doppler Radar Array ................................ ........ 47 3.5 DC Offset Compensation ................................ ................................ .................. 50 3.6 Experiments ................................ ................................ ................................ ...... 52 3.7 Limitation of Sensitivity Improvement ................................ ................................ 54 3.8 Limitation of Real time Large Body Movement Cancellation ............................. 57 3.9 Summar y ................................ ................................ ................................ .......... 57 4 SYSTEM LEVEL INTEGRATION OF HANDHELD WIRELESS NONCONTACT VITAL SIGN SENSOR RADAR ................................ ................................ .............. 59 4.1 Challenges of Portable Applicatio ns ................................ ................................ 59 4.2 Vital Sign Detection System Architecture ................................ .......................... 60 4.3 Baseband Signal Processor Design ................................ ................................ .. 64 4.4 Receiver Chain Noise Analysis ................................ ................................ ......... 67 4.4.1 LNA and Gain Block ................................ ................................ ................ 67 4.4.2 Mixer with LO Input ................................ ................................ .................. 67 4.4.3 Baseband Amplifier ................................ ................................ ................. 69 4.4.4 Complete Noise Performance Evaluation Model ................................ ..... 69 4.5 Experiments ................................ ................................ ................................ ...... 71 4.5.1 Two tone Actuator Movement ................................ ................................ 72 4.5.2 Human Respiration and Heart Beat Measurement ................................ .. 73 4.5.3 Guideline for Selecting the Sampling Frequency ................................ ..... 75 4.5.4 The Effect of Output SNR on Detection Accuracy ................................ ... 75 4.5.5 The Trade off between Output SNR and Detection Accuracy ................. 77 4.6 Summary ................................ ................................ ................................ .......... 79 5 INTE GRATED VITAL SIGN RADAR SENSOR WITH ON BOARD ANTENNA ...... 80 5.1 Integration of Vital Sign Radar and Antennas ................................ ................... 80 5.2 Transmitting and Receiving Antenna Arrays Design ................................ ......... 80 5.3 Orientation of the TX and RX Antennas ................................ ............................ 83 5.4 Simulation of the Coupling between TX and R X Antennas ............................... 84 5.5 System Integration of the Vital Sign Detector with On board Antenna .............. 86 5.6 Low power Design, Link budget, and Emission Safety ................................ ..... 88 5.7 Summary ................................ ................................ ................................ .......... 91 6 CONCLUSIONS ................................ ................................ ................................ ..... 92 LIST OF REFER ENCES ................................ ................................ ............................... 9 3 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 101

PAGE 7

7 LIST OF TABLES Table page 2 1 Wireless hydrogen sen sor board bill of material ................................ ................. 33 4 1 RF transceiver board bill of material ................................ ................................ ... 63 4 2 Receiver chain components noise specification ................................ ................. 67 5 1 Dimensions of the patch antenna array. ................................ ............................. 83 5 2 Received RF power estimate for 5.8 GHz integrated vital sign sensor. .............. 89

PAGE 8

8 LIST OF FIGURES Figure page 1 1 Topology of the hydrogen sensor network reported in Sensors [24]. .................. 16 1 2 Quadrature homodyne vital sign radar architecture. ................................ ........... 18 1 3 Double sideband heterodyne vital sign radar architecture. ................................ 19 1 4 Direct IF sampling heterodyne vital sign radar architecture. ............................... 20 1 5 Self injection locking vital sign radar architecture [78]. ................................ ....... 21 2 1 Microscopic images of differential sensing diodes.. ................................ ............ 26 2 2 Absolute and differential current of HEMT diodes.. ................................ ............ 27 2 3 Star network layout. ................................ ................................ ............................ 28 2 4 Block diagram of wireless multiple hydrogen sensor system. ............................. 29 2 5 Sequence of transceiver modu le operation. ................................ ....................... 31 2 6 Images of wireless sensor network monitoring software ................................ .... 34 2 7 An image of the hydrogen sensing website showi ng the real time responses of the hydrogen sensors. ................................ ................................ .................... 35 2 8 State flow diagram of the hydrogen sensor network software monitoring mechanism. ................................ ................................ ................................ ........ 35 2 9 Individual hydrogen sensor package.. ................................ ................................ 36 2 10 A photograph of base station including wireless receiver and computer. ........... 36 3 1 Block diagram and setup of the vital sign detection system. .............................. 41 3 2 Baseband I/Q signals: time domain signal and frequency domain spectrum.. .... 43 3 3 Block diagram of the vital sign detection system with Doppler radar array. ........ 45 3 4 Simulation of 2 D random body movement cancellation.. ................................ ... 46 3 5 Amplitude of Bessel functions ................................ ................................ ............. 48 3 6 Respiration and heartbeat sensitivity improves as the number of detectors increases. ................................ ................................ ................................ ........... 49

PAGE 9

9 3 7 Heartbeat spectra when DC offset is present in various detector settings .......... 51 3 8 Illustration of DC offset compensation algorithm. ................................ ............... 51 3 9 Photograph of the RF radar array and TX/RX antennas. ................................ .... 52 3 10 Two dimensional random body movement cancellation using multiple detector s array. ................................ ................................ ................................ ... 53 3 11 Amplitude of Bessel functions. ................................ ................................ ............ 55 3 12 Respiration and heartbeat sensitivity peaks at 12 sensors and 8 sensors, respectively. ................................ ................................ ................................ ........ 56 4 1 Block diagram of the vital sign detection system. ................................ ............... 61 4 2 Photograph of the RF transceiver board and signal processor board ................. 62 4 3 Block Diagram of the RF transceiver board ................................ ........................ 63 4 4 Flow diagram of the spectrum analysis algorithm ................................ ............... 65 4 5 Photo of the digital signal processor board ................................ ......................... 66 4 6 Noise figure of active mixer and passive mixer in 0.13 um CMOS. .................... 68 4 7 Two tone actuator movement experiment setup ................................ ................. 72 4 8 Theoretical results vs. experimental results of t he two tone actuator experiment ................................ ................................ ................................ .......... 73 4 9 Human respiration and heart beat measurement setup. ................................ ..... 74 4 10 Detected baseband signal and spectra in non contact vital sign detection. ........ 74 4 11 Simulated baseband signal and spectrum in non contact vital sign detection. ... 76 4 12 Detected baseband signal and spectrum in non contact vital sign detection. ..... 78 4 13 Simulated receiver output SNR. ................................ ................................ ......... 78 5 1 Patch antenna array model used i n on board antenna design. .......................... 81 5 2 Patch antenna array radiation pattern. Maximum gain 11.5 dB is achieved. ...... 82 5 3 S11 of patch a ntenna array. The antenna resonates at 5.8 GHz. ....................... 82 5 4 H plane patch antenna array orientation. ................................ ........................... 84

PAGE 10

10 5 5 Mutual coupling simulatio n model in Ansoft HFSS. ................................ ............ 85 5 6 S12 of the on board TX and RX patch antenna array. ................................ ........ 86 5 7 Photograph of the integrated vital sign radar sensors with on board antennas. ................................ ................................ ................................ ............ 87 5 8 Photograph of the real time integrated vital sign radar software. ........................ 88 5 9 IEEE RF s afety standard C95.1 2005. ................................ ............................... 90 5 10 Power density of the integrated noncontact vital sign detector. .......................... 91

PAGE 11

11 LIST OF ABBREVIATION S HEMT High electron mobility transi stor; VCO Voltage controlled oscillator; LCD Liquid crystal display; PIO Parallel Input/Output; LED Light emitting diode; RAM Random access memory; CMOS Complementary metal oxide semiconductor; RMS Root mean square; f Frequency; f c Carrier frequency; VCO V oltage controlled oscillator; CW Continuous wave; TX Transmitter; RX Receiver; RF Radio frequency; IF Intermediate frequency; LO Local oscillator; LNA Low noise amplifier; BPF Band pass filter; FFT Fast Fourier transform;

PAGE 12

12 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy MULTIPLE SENSOR PLATFORMS FOR HYDROGEN AND HUMAN PHYSIOLOGICAL MOVEMENT SENSING By Xiaogang Yu Dec ember 2011 Chair: Jenshan Lin Major: Electrical and Computer Engineering This dissertation begins with a demonstration of the integration of multiple sensor techniques with state of the art hydrogen sensing devices T he proposed multiple sensor system us es six Zigbee transceivers to collect hydrogen density information from the dispersedly deployed AlGaN/GaN high electron mobility transistor (HEMTs) differential sensing diodes. The collected hydrogen density information is transmitted wirelessly to t he ba se station for data logging and tracking of each individual sensor The software at the base station defines and implements the monitoring states, transitions, and actions of the hydrogen sensing system T he software also is able to warn the user of potent ial sensor failure, power outages and network failures through cell phone network and Internet. Real time responses of the sensors are displayed through a web site on the Internet. The sensing system has shown good stability for more than 18 months in an outdoor field test. After that, Chapter 3 is devoted to a presentation of the integration of multiple sensor techniques with noncontact vital sign detection. Using the multiple vital sign sensor platform, two dimensional random body movement cancellation i s achieved. The multiple sensor system includes four detectors, an 8 channel data acquisition module,

PAGE 13

13 and a computer for spectrum analysis. Each of the detectors consists of a radio frequency transceiver, a baseband analog circuit, and a power man agement c ircuit. The multiple sensor platform s also strengthen the detecting sensitivity on the respiration and heartbeat. A DC offset compensation algorithm is introduced to free the body movement cancellation from disturbance of unwanted DC offset. Experiments we re performed with a human subject in lab oratory environment. Results were analyzed to verify the improved detection performance at the presence of 2 D human body movement. The limitation of sensitivity improvement and body movement cancellation are demonst rated with simulation results. Chapter 4 details the hardware design of the individual portable Doppler radar for noncontact vital sign detection Topics includ ing RF transceiver board design, ba seband signal processor design, sampling frequency selection guideline, and noise analysis for the receiver chain of the detector will be discussed. Finally, a system integration of noncontact vital sign detector with antennas on board will be presented in Chapter 5

PAGE 14

14 CHAPTER 1 INTRODUCTION 1.1 Background Multiple se nsor technology is a key component in the science and applications of sensing physical, chemical, and biological phenomena. The relative low cost of sensors, the availability of high speed communication networks, and the increased computational capability have enabled great research interests and advances in this area In recent years, new techniques for sensing two particular phenomena, hydrogen and human physiological movements, ha ve enjoyed great advances. The integration of multiple sensor technology wi th these new techniques is the next logical step in the evolution of hydrogen and physiological movement sensing. In the area of hydrogen sensing, the sensors are required to detect hydrogen near room temperature with minimal power consumption and weight a nd with a low rate of false alarms. Due to their low intrinsic carrier concentrations, GaN and SiC based wide band gap semiconductor sensors are developed to operate at lower current levels than conventional Si based devices and offer the capability of de tection to 600 C [1 ] [ 23]. The ability of electronic devices fabricated in these materials to function in high temperature, high power and high flux/energy radiation conditions enable performance enhancements in a wide variety of spacecraft, satellite, homeland def ense, mining, automobile, nuclear power, and radar applications. In the area of human physiological sensing, the concept of noncontact vital sign detection has been demonstrated in various publications before 2000 [26] [ 30 ]. After 2000, more microwave sens ing systems [ 3 1 ] [ 81 ] with lower power, smaller package, improved sensitivity, and longer detection range have been developed to detect the

PAGE 15

15 physiological movements, i.e. heart beat and respiration The microwave sensing system transmits a radio frequency, single tone continuous wave (CW) signal, which is reflected off of a target and then demodulated in the receiver. CW radar with the human body as the target will receive a signal with its phase modulated by the time varying chest wall position. Demodulatin g th is phase will then give a signal proportional to the chest wall position that contains information about movement due to heartbeat and respirat ion. This technique enabled non contact detection of vital signs of humans or animals from a distance away, wi thout any sensor attached to the body. The non intrusive nature and penetration capability through the building materials bring unique property to home healthcare mon itoring, search and rescue for earthquake or fire victims, security, and military applicat ions. 1.2 Recent Progresses on Hydrogen Sensing In the field of hydrogen sensing, r ecent developments in the early 2000s ha ve shown the promising performance of AlGaN/GaN high electron mobility transistors (HEMTs) for use in hydrogen sensing [ 4 ] [ 23 ] The high electron sheet carrier concentration of nitride HEMTs provides an increased sensitivity relative to simple Schottky diodes fabricated on GaN layers T his dissertation will present the work of a multiple sensor platform using a differential pair of Al GaN/GaN HEMT diodes for hydrogen sensing near room temperature [2 5 ] This multiple sensor configuration provides a built in control mechanism to reduce false alarms due to temperature swings or voltage transients. The design and optimization of the detecti on circuitry, digital signal processing, wireless network, and monitoring states to maintain an accurate and reliable system were investigated.

PAGE 16

16 Fig ure 1 1. Topology of the hydrogen sensor network reported in Sensors [24] In terms of wireless network de sign for hydrogen sensors, a wireless sensor network [24] was report ed for in situ monitoring of atmospheric hydrogen concentration in 2003. In that network design, the system consists of multiple sensor nodes, equipped with titania nanotube hydrogen senso rs, distributed throughout the area of interest; each node is both sensor, and data relay station Figure 1 1 shows the experimental setup of the one way peer to peer sensor network. Node 2 transmits the sensor information to Node 3 since it is the only no de within the transmission range of Node 2. Similarly, Node 4 is the preceding node of Node 3 due to its proximity, and Node 1 i s the preceding node of Node 4. This peer to peer setup enables extended wide area monitoring However, the potential failure of any preceding sensor node will break the afterward data relay path and will result in the malfunction of the sensor network. This

PAGE 17

17 dissertation will present a hydrogen sensor network using star topology. I n the star topology the sensor nodes are individua lly connected to the base station. The failure of one sensor will not affect the functioning of the whole sensor network. 1.3 Recent Progresses on Physiological Movement Sensing 1.3.1 Theoretical Breakthrough s In the field of noncontact vital sign detectio n, researchers working on noncontact vital sign detection have spent great efforts to achieve a ccurate and robust performance while solving many technical challenges, especially in the years from 2008 to 2010 [ 42] [80] As one of the main challenges, the i nfluence of clutter noise and phase noise has been solved by the range correlation effect by applying the same transmitted signal to the receiver as the reference signal [ 33 ]. Another challenge, the null detection point problem, was solved by frequency tun ing in the double sideband transmission system [ 37 ] and complex signal/arctangent demodulation in the quadrature direct conversion system [ 40 ][4 2 ] In addition to the experimental efforts to improve the system performance, theoretical analyses have been pe rformed to study the Doppler non contact vital sign detection and provide guidelines for the designs. Achievements include the analysis on the range correlation effect and I/Q performance benefits [ 33 ], the modeling and analysis of the double sideband tran smission to eliminate the null detection point [ 3 7 ], the spectral analysis of the non linear phase modulation effect [ 4 2 ], the analysis of the arctangent demodulation in quadrature receivers [ 4 0 ], and the comparative study on different radio architectures for vital sign detection [ 39]. 1.3.2 RF Front end Architectures There are various RF front end architectures designed to achieve the theoretical concepts outlined in the aforementioned publications. Five kinds of architectures have

PAGE 18

18 been reported: homodyne, heterodyne, double sideband architectures, direct intermediate frequency (IF) sampling, and self injection locking. Homodyne transceiver for vital sign detection is originally implemented using single channel direction conversion architecture. Although t he detected signal of the single channel transceiver contains the vital sign information, it is very weak at certain detection distances, i.e. null detection points. Quadrature direction conversion Doppler radar is designed to eliminate the null detection point problem [33]. It is also found that the quadrature baseband signals can be combined in software to perform complex signal demodulation [42] or arctangent demodulation [40]. Figure 1 2 shows a block diagram of the quadrature homodyne vital sign radar. Fig ure 1 2 Quadrature homodyne vital sign radar architecture Before the debut of homodyne transceivers in 2001 [31][32], the h eterodyne transceiver ha d been the dominant design architecture for vital sign detection [ 29]. Since the heterodyne tra nsceivers suffer the same null detection problem in single

PAGE 19

19 channel homodyne transceivers, they have to be designed in quadrature architecture. In 2005, a double sideband heterodyne architecture was proposed to eliminate the need of generating quadrature LO signals [37] Figure 1 3 shows a block diagram of the double sideband vital sign radar architecture. The heterodyne transceiver transmits both the upper and lower sidebands in double sideband configuration. The double sideband signal is reflected on the s ubject and received by the heterodyne receiver. By combining the baseband signal of both the sidebands, the distance between optimal and null detection points is changed to IF /16 Since IF is the wavelength at the IF stage, the double sideband configurat ion results in a much longer separation than the distance ( R F /8) in conventional heterodyne transceiver. Fig ure 1 3 Double sideband heterodyne vital sign radar architecture In 2008, a direct IF sampling heterodyne transceiver for vital sign detection was reported [50]. Figure 1 4 shows a simplified block diagram of the direct IF sampling

PAGE 20

20 transceiver. In this architecture, the output of the RF mixer is sampled and digitized by a high speed ADC. The digital IF signal is demodulated by a digital quadratur e demodulator. The subsequent DSP is performed directly on the digital quadrature signals. The d irect IF sampling is free from the I/Q imbalance in an analog IF quadrature demodulator and eliminates the DC offset calibration. Fig ure 1 4 Direct IF sampli ng heterodyne vital sign radar architecture In 2010, a new self injection locking approach was introduced to implement the detection of vital signs [7 8 ]. A differential LC voltage controlled oscillator (VCO) with injection port is used in the new architec ture. The output of the VCO is amplified by a power amplifier (PA) and transmitted toward the subject. The reflected signal is received by the receiving antenna (RX) and sent to the injection port of the VCO as the injection signal. The vital sign informat ion modulated in the injection port signal is demodulated by the self injection locking mechanism. The self injection locking architecture provides higher signal gain at low modulation frequency and improved noise attenuation at long

PAGE 21

21 distance detection. A s uccessful experiment has been achieved with a subject seat ed at a distance of 50 cm. Figure 1 5 shows a block diagram of the self injection locking architecture of vital sign detection. Fig ure 1 5 Self injection locking vital sign radar architecture [7 8] 1.3.3 Advances in Signal Processing Technique s The basic signal processing methods for vital sign detection are complex signal demodulation [42] and arctangent demodulation [38]. In complex signal demodulation, the baseband I/Q signals are multiplied t ogether so that the complex signal is free from residual phase and optimum/null detection problem. In arctangent demodulation, the algorithm calculates the Doppler phase shift as = arctan( Q / I ) ; therefore, the optimum/null problem is also eliminated. Multiple input, multiple output (MIMO) and single input, multiple output (SIMO) techniques have been introduced to detect vitals sign from multiple subjects [35][38] Using these mult iple output algorithms, it is prove n by the generalized likelihood ratio test (GLRT) that the distinguish ing among 2, 1, or 0 subjects can be achieved MIMO

PAGE 22

22 techniques are also used to cancel the random body movement of the subject and improve the sensitiv ity of the vital sign detection system [42][81]. Since most of the human body under test has random body movement, e.g. a seated person randomly moving in two horizontal dimensions, the body m ovement presents a challenge to detect successfully the vital si gn movements. It has been reported recently that the difference in phase characteristics of the vital sign movements and body movement creates an opportunity for random body movement cancellation in single direction [51] and in two dimensions and above [81 ]. Aside from increasing the number of the detectors, the improvement in signal processing is also taking place in the increasing of the number of carrier frequencies. In a multiple frequency Doppler radar system RF signals with different carrier frequenc ies are transmitted toward the subject in very small beam angles so that the reflection point of the RF signals are different. The differential measurement can be used to cancel random body motions. A dual helical antenna and simple direct conversion radar are reported to use this differential measurement approach [58]. Two frequency radar [ 43] and multiple frequency interferometric radar [56] are reported. In the spectrum of the vital sign signals, the third and fourth harmonics of the respiration signal i s close to the heart beat frequency, leading to difficulty for extraction of the correct heart beat signal. A parametric and cyclic optimization approach, referred to as the RELAX algorithm, is designed to mitigate these difficulties. The implementation of the RELAX algorithm in vital sign detection wa s reported in 2010 [73]. Other signal processing methods for vital sign detection include adaptive filtering [75], Kalman filtering and principal component combining of quadrature channels [52], fast

PAGE 23

23 clutter c ancellation [77], DC information preservation [ 46 ], and blind source separation [ 44 ]. 1.3.4 Miniaturization and System on chip Recently, the realization of the detection in the compact portable system has become a new focus of interest. M any of the applica tions such as sleep apnea monitoring and earthquake search and find rescue require integration of the entire system in small portable packages. An integrated noncontact vital sign detector was developed for handheld applications [ 53 ]. Noise performance of the integrated detector was investigated to guide the hardware design [ 74 ]. In addition, three reports of vital sign sensor integrated circuits chip have been published [45][59][76]. This dissertation will present a multiple sensor platform for two dimensi onal random body movement cancellation. A portable non contact vital sign detector for handheld applications will be presented Details of hardware design and noise analysis of the individual sensor will be discussed. It will also introduce the new nonconta ct vital sign detector with on board antennas and real time noncontact vital sign detection software.

PAGE 24

24 CHAPTER 2 MULTIPLE WIRELESS SE NSOR PLATFORM USING ALG A N/G A N HIGH ELECTRON MOBILITY TRANSISTOR DIFFERENTIAL DIODE S ENSORS 2.1 Hydrogen Sensors with Differe nt Fabrication Technologies There is great interest in detection of hydrogen sensors for use in hydrogen fuelled automobiles and with proton exchange membrane (PEM) and solid oxide fuel cells for space craft and other long term sensing applicati ons. These sensors are required to detect hydrogen near room temperature with minimal power consumption and weight and with a low rate of false alarms. Due to their low in trinsic carrier concentrations, GaN and SiC based wide band gap semiconductor sensors can be op erated at lower current levels than conventional Si based devices and offer the capability of detection to 600 C [1 23]. The ability of electronic devices fabricated in these materials to function in high temperature, high power and high flux/energy radiation conditions enable performance enhancements in a wide variety of spacecraft, satellite, homeland defen se, mining, automobile, nuclear power, and radar applications. AlGaN/GaN high electron mobility transistors (HEMTs) show promising performance for use in broadband power amplifiers in base station applications due to the high sheet carrier concentration, electron mobility in the two dimensional electron gas (2DEG) channel and high saturation velocity. The high electron sheet carrier concentration of nitride HEMTs is induced by piezoelectric polarization of the strained AlGaN layer and spontaneous polariza tion is very large in wurtzite III nitrides. This provides an increased sensitivity relative to simple Schottky diodes fabricated on GaN layers [4 23]. An additional attractive attribute of AlGaN/GaN diodes is the fact that gas sensors based on this materi al could be integrated with high temperature electronic devices on the same chip. The advantages of GaN over SiC for sensing include the

PAGE 25

25 presence of the polarization induced charge, the availability of a heterostructure and more rapid pace of device techn ology development for GaN which borrows from the commercialized light emitting diode and laser diode businesses. In this dissertation we report on a demonstration of a hydrogen sensing system using a differential pair of AlGaN/GaN HEMT diodes for hydroge n sensing near room temperature. This configuration provides a built in control diode to reduce false alarms due to temperature swings or voltage transients. The design and optimization of the detection circuitry, digital signal processing, wireless networ k, and monitoring states to maintain an accurate and reliable system were investigated. 2.2 Experiments with Differential Sensor Pairs Al GaN/GaN HEMT layer structures were grown on C plane Al 2 O 3 substrates by a molecular beam epitaxy (MBE) system. The laye r structure included an initial 2 m thick undoped GaN buffer followed by a 35 nm thick unintentionally doped Al 0.28 Ga 0.72 N layer. The sheet carrier concentration was 110 13 cm with a mobility of 980 cm 2 /(V s) at room temperature. We designed a mask that employed a differential diode c onfiguration, with a Pt contact device as the active member of the pair and a Ti/Au contact device as the control. Mesa isolation (the electrical components of an integrated circuit are isolated, using P N junction or dielectric isolation) was achieved wit h 2000 plasma enhanced chemical vapor deposited SiNx. The Ohmic contacts was formed by lift off of ebeam deposited Ti (200 )/Al (1000 )/Pt (400 )/Au (800 ). Th e contacts were annealed at 850 C for 45 s under a flowing N 2 ambient in a Heat pulse 610T system. Schottky contacts of 100 Pt for the active diode and 200 Ti/1200 Au for the reference diodes were deposited by e beam evaporation. Final metal of e beam

PAGE 26

26 deposited Ti/Au (300 /1200) in terconnection contacts was employed on the HEMT diodes. Fig ure 2 1 shows an optical microscope image of the completed devices. Fig ure 2 1. Microscopic images of differential sensing diodes. The opening of the active diode was deposited with 10nm Pt, and the reference diode was deposited with Ti/Au. Fig ure 2 2 shows the absolute and differential forward current voltage (I V) characteristics of the HEMT active (top) and reference ( bottom) diodes, both in air and in a 1% H 2 in air atmosphere. For the activ e di ode, the current increases upon introduction of the H 2 through a lowering of the e ffective barrier height. The H 2 catalytically decomposes on the Pt metallization and di ffuses rapidly to the interface where it forms a dipole layer [23]. The differenti al change in forward current upon introduction of the hydrogen into the ambient is 1 4mA over the voltage range examined and peaks at low bias. This is roughly double the detection sensitivity of comparable GaN Schottky gas sensors tested under the same conditions, confirming

PAGE 27

27 that the HEMT based diode has advantages for applications req uiring the ability to detect hydrogen even at room temperature. Fig ure 2 2. Absolute and differential current of HEMT diode s. (a) Absolute and differential current of HEMT sensor diode. (b) Absolute and differential current of HEMT reference diode. 2.3 W ireless Multiple Sensor System 2.3.1 System Overview The wireless sensing system consists of six wireless sensor nodes and a base station including a wireless receiver and a computer equipped with monitoring software. The topology of the sensor network is star topology as shown in Figure 2 3. The star topology reduces the chance of network failure by connecting all of the sensor nodes to a central node. Each sensor node consists of a differential sensor pair, detection circuits, microcontroller, wireless tr ansceiver, and power management circuits. The main part of the detection circuits is an instrumentation amplifier used to sense the change of current in the device. The current variation, embodied as a change in the output voltage of the detection circuit, is fed into the microcontroller. The microcontroller calculates the corresponding current change and controls the transceiver to transmit the data to the

PAGE 28

28 wireless network base station. The block diagram of the sensor node and the wireless network base sta tion are illustrated in Fig ure 2 3a. The user friendly hydrogen sensor network monitoring software on the base station computer performs functions such as communication port setting, emergency alarm, data collection and data plot. The monitoring states, t ransition, and actions are defined in this software. The software also sends the data to a remote web site through the Internet. Internet users around the world can access the web page at any time and see the plotted data exactly the same as those on the l ocal sensor base station. The block diagram of the remote sensing system is shown in Figure 2 3b. Figure 2 3. Star network layout. The wireless sensing system consists of six wireless sensor nodes and a base station The wireless network is enabled by IE EE 802.15.4 WPAN (Zigbee) technology. The Zigbee wireless communication nodes are operating at 2.4 GHz.

PAGE 29

29 Figure 2 4 Blo ck diagram of wireless multiple hydrogen sensor system. (a) The wireless sensing system consisting of differe ntial sensor pair, diff erential sensor pair, detection circuits, microcontroller, wireless transceiver, power management circuits, receiver and computer. (b) The remote sensing system consisting of computer, web server, ASP.NET program, and hydrogen sensing web site. 2.3.2 Detec tion Circuits In this design, the differential change of the forward current in the hydrogen sensor device causes a voltage variation on the sensor node. This voltage variation is usually very small as demonstrated previously. A differential input single e nded output instrumentation amplifier is used to amplify precisely the variation to a level that will be sampled correctly by a microcontroller. To obtain optimal input characteristics, two

PAGE 30

30 voltage followers buffer the input signal. The input impedance of the buffers is very high and allows the instrumentation amplifier to be used with high source impedances and still have low error. Also the high input impedance tolerates the unbalanced source impedance with no degradation in common mode rejection. The buf fers drive the balanced differential amplifier. The gain of the amplifier is set by the feedback voltage divider. The bias voltage of the sensors is set to 1.8V to reduce the power consumption. The circuits and specifications are shown in the detection cir cuit part of Fig. 2 3a. The signal from the instrumentation amplifier is in the continuous analog form. In order to transmit the data through digital wireless transceivers, the data should be digitized first. A MSP430 ultra low power microcontroller is use d to perform the analog to digital conversion. The choice of microcontroller is based on the power consumption consideration. It is programmed to operate in low power mode after the analog to digital conversion operation and reduce the power consumed at th e processor core. The on chip analog to digital converter (ADC) features a data transfer controller. This feature allows samples to be converted and stored without CPU intervention. 2.3.3 Zigbee Wireless Network An IEEE 802.15.4 WPAN (Zigbee) compliant 2.4 GHz wireless sensor network has been set up for data transmission, to accommodate a number of hydrogen sensor nodes implemented in the system. The Zigbee compliant wireless network supports the unique needs of low cost, low power sensor networks, and opera tes within the unlicensed 2.4GHz band. The transceiver module is completely turned off for most of the time, and it is turned on to transmit data in extremely short intervals. The timing of the system is shown in Fig ure 2 4. When the sensor module is turne d on, it is programmed to power up for the first 30 s. Following the initialization process, the

PAGE 31

31 detection circuit is periodically powered down for 5 s and powered up again for another 1 s, achieving a 16.67% duty cycle. The ZigBee transceiver is enabled f or only 5.5ms to transmit the data at the end of every cycle. This gives a RF duty cycle of only 0.09% and significantly saves the power consumption. Fig ure 2 5 Sequence of transceiver module operation. 2.3.4 Wireless S ensor N etwork Monitoring S oftware The hydrogen sensor network software was developed using NET Framework v3.5. The software can be installed and launched on any Windows based operating system. It performs the functions of communication port setting, emergency alarm, data collection, and r eal time data plot viewing. The software also defines the monitoring states, transitions, and actions. In addition, a remote hydrogen sensing system was developed to present the data plot to Internet users, regardless of the user locations. The general con trol interface and the graphical data view of the software are presented in Fig ure 2 5a and b. The data channeled from the Zigbee receiver cont ains sensor ID, sensed currents, and sensed voltage. The software uses these data to calculate the density of the hydrogen gas. Based on the calculation the monitoring state will either transit or stay. And the corresponding action will be performed. The data are transferred in the same

PAGE 32

32 format to the remote web site through the Internet. Internet data transfer employ s the data package technology for safety purposes. The data packages are stored and analyzed again at the web server. A web page is constructed for displaying sensor information. The web page is presented in Fig ure 2 6. Users can select different time wind ows from real time to 6 days to display the sensor data. 2. 3. 5 Monitoring States, Transitions, and A ctions The state diagram of the hydrogen sensor network software is illustrated in Fig ure 2 7. The monitoring states include: initialize, collect dat a, anal yze data, emergency, and sleep. The state machine runs through each state unti l a possible emergency hydrogen density is detected and sustained for 20 s. The emergency threshold was set at a level that hydrogen concentration would be high enough to pose an y danger. In case of an emergency, the software will trigger the alarm and make phone calls to the numbers computer is required). The Internet data transfer and storage is performed in the 2.3.6 Packages The sensor module is fully integrated on an FR4 PC board as shown in Fig ure 2 8a. The FR4 PC board has a thickness of 0.062" and is measured 2.75" x 1.52" The circuit board is enclosed in a plastic pa ckage as shown in Fig ure 2 8b, which has a sensor guard to protect the sensor device from being damaged by an external object. The circuit board is powered by the AC power and backed up by a 9 V battery. A power sensing chip is used to sense the voltage fr om the wall plug adapter. In the case of power failure, the power management circuits will switch to 9 V battery. The base

PAGE 33

33 station consisting of a wireless receiver and computer are presented in Fig ure 2 9. A bill of material of the transceiver is listed i n Table 2 1. Table 2 1 Wireless h ydrogen s ensor b oard b ill of m aterial Block Vendor Specification ZigBee RF Module Digi 2.4 GHz operating frequency 92 dBm receiver sensitivity, 90 m outdoor range, 250 Kbps data rate, 0 dBm output power MSP430 Micr ocontroller TI 8 MIPS, 1 .8 3.6 V operat ing voltage up to 60 KB FLASH 12 bit SAR ADC Crystal ABRACON 32.768 KHz operating frequency, 12.5 p F load capacitance, through hole mounting, 20 ppm frequency tolerance, 20C to +70C operating temperature Linear Regulator Maxim 1.8V, 2.5V, 3.3V, and 5V f ixed output voltage, 2.5V to 12V Input Voltage Range 200 mA max Output Current Power Supervisory Circuits Maxim 5.0V, 3.3V, 3.0V, and 2.5V p ower s upply m onitoring 1.2V o perating s upply v oltage Operati onal Amplifier Maxim 1 V to 5.5 V v oltage o peration 9 supply current consumption, r ail to r ail o utput swing Switch Switchcraft DPDT c ontact c onfiguration r aised s lide a ctuator 125 V maximum c ontact v oltage 3 A maximum c ontact c urrent PCB G oldphoenix 2 layers, 0.062" board thickness 1 oz copper thi ckness, FR4 TG130, two side silk 2.75" x 1.52" board size Sensor Enclosure Box Enclosures P lastic, 1.5" x 2.75" x 4.6" box size, 9 V B attery

PAGE 34

34 Fig ure 2 6 Images of w ireless s ensor n etwork m onitoring s oftware (a) An image of general control interfa ce including monitoring status, data file, and emergency call functions. (b) An image of graphic al data view interface including data view and curve view.

PAGE 35

35 Fig ure 2 7 An image of the hydrogen sensing website showing the real time responses of the hydroge n sensors. Fig ure 2 8 State flow diagram of the hydrogen sensor network software monitoring mechanism.

PAGE 36

36 Fig ure 2 9 Individual hydrogen sensor package. (a) A photograph of differential hydrogen sensor PC board including differential hydrogen sensor, detection circuits, microcontroller, and wireless transmitter. (b) A photograph of sensor node package with sensor guard. Fig ure 2 10 A photograph of base station including wireless receiver and computer. The picture is taken at the Greenway Ford dealer ship, Orlando, Florida. 2.4 Field Test Field tests have been conducted both at the University of Florida and at Greenway Ford in Orlando, FL. The outdoor tests at the University of Florida have been conducted

PAGE 37

37 several times for a period of 2 weeks, to test a range of possible real world conditions in a more controlled setting. Hydrogen leakage was successfully detected for hydrogen concentrations in a range from1% to 100% at the point of the leak and heights ranging from 1 to 10 ft in an outdoor environment. The setup at Greenway Ford was aimed to test the stability of the sensor hardware and the server software under the actual operating environment. The test was started on the 30th of August 2006 and has run until the time of this report. Six sensor modules and the server have been functioning since then. A web site was also developed to share the collect ed sensor data via the Internet (http://ren.che.ufl.edu/app/default.aspx), as shown in Fig ure 2 6. This figure illustrates the current level of each sensor on the network and data for real time and the choices of past 85min, 15 h or 6 days can be viewed on the web site. If any of the alarm is triggered. The server program for the wireless sensor network could also report a hydrogen leakage emergency through phone lines using the modem to send a message to cell phones, beepers, fire department, and so forth 2.5 Summary In conclusion, a wireless sensor network whi ch uses the IEEE 802.15.4 standards has been constructed to transmit data from a number of hydrogen sensors to a base station. A use r friendly program has been developed to share the data collected by base station to Internet, so that the data can be analy zed and monitored from anywhere with an Internet connection. A cell phone alarm has been implemented to report any potential hydrogen leakage to responsible personnel. The entire system has been tested for functionality and stability both at the University of Florida and at Greenway

PAGE 38

38 Ford in Orlando. Field tests show that the low power hydrogen sensor can work stably and react quickly to possible hydrogen leakage.

PAGE 39

39 CHAPTER 3 MU LTIPLE DOPPLER RADAR SENSOR PLATFORM FOR TWO DIMENSIONAL HIGH SENSITIVITY HUMAN PH Y SIOLOGICAL MOVEMENT DETECTION 3.1 Challenges of Body Movements The non contact vital sign detection system [ 2 6 ] [ 27 ] is designed based on the Doppler phase modulation effect in microwave frequency band s. The radar transmits an ultra low power un modulated el ectromagnetic wave toward the human body, where it is reflected and phase modulated by the periodic physiological movement, i.e. the respiration and heartbeat. By down conversion and proper signal processing of the reflected signal, the vital signs can be extracted. To achieve accurate and robust performance, researchers working on noncontact vital sign detection have spent great efforts for more than two decades on several technical challenges. Among these challenges, noise has always been a main concern. One of the main challenges, the influence of clutter noise and phase noise has been solved by the range correlation effect by means of applying the same transmitted signal to the receiver as the reference signal [3 3 ]. In order to understand the overall n oise performance of vital sign detectors, the investigation of the signal to noise ratio (SNR) of the detectors in quadrature direct conversion architecture [3 3 ][ 42 ] clarifies the effect of SNR on vital sign detections [ 74 ]. In addition to the inherent noi ses from the electronic circuits, the noise from the random movement of the human body presents even severer distortion to the vital sign information. To solve the random body movement challenge, researchers introduced the double detector technique to can cel the body mo vement in single direction [ 42 ]. As the above efforts kept on pushing the non contact vital sign detection closer to daily applications, we needed to give special attention to a few challenges before we

PAGE 40

40 c ould accomplish a practical vital sig n detection system. Since most of the human body under test conditions has random body movement in at least two dimensions, e.g. a seated person randomly moving in two horizontal dimensions, the cancellation techniques need to be expanded to multiple dimen sions. In addition, the resultant vital sign signal trajectories from the multiple detectors should be compared carefully in the constellation graph and compensated with certain DC offset in real time to ensure a correct recovering of vital sign informatio n, i.e. the respiration rate and heart rate. In this chapter we report a two dimensional noncontact vital sign detection system with Doppler radar array for the accurate and body movement calibrated operation. The system consists of four noncontact vital s ign detectors placed at the four sides of the human body. Each noncontact vital sign detector includes a radio frequency transceiver, a baseband analog circuit, and a power management circuit. The baseband signals from the multiple detectors are channeled to a computer for spectrum analysis. Details of sensitivity enhancement and DC offset compensation algorithm will be discussed. 3.2 Principle of Noncontact Vital Sign Detection Figure 3 1 (a) shows a block diagram of the quadrature direct conversion vital s ign detection system. Figure 3 2(b) shows a continuous wave (CW) Doppler radar vital sign detection experiment setup. The vital sign detector consists of an RF transceiver front end, a baseband amplifier, and a built in analog to digital converter (ADC) of the digital signal processor. For vital sign detection, the radar transmits a continu ous wave un modulated RF signal toward the human subject. The transmitted signal can be represented as (3 1)

PAGE 41

41 (a) (b) Figure 3 1. (a) Block diagram of the vital sign detection system. (b) S etup of the vital sign detection experiment The RF signal is reflected on the surface of the human body. The reflected signal is modulated by the physiological movement x ( t ) and received at Node 1. The received RF signal can be represented as (3 2 )

PAGE 42

42 where f is the carrier frequency, = c / f is the wavelength, d 0 is the distance between the vital sign detec tor and the subject, x ( t ) is the time varying displacement of the subject, and is the phase noise of the received signal. The received signal is then amplified and fed to the mixer a t node 2. When the signal at node 2 is mixed with the LO signal derived from the transmitted signal, the down converted signal at node 3 can be represented as (3 3) w he re x h ( t ) = m h sin ( h t ) and x r ( t )= m r sin( r t ) represent the heartbeat and respiration is the residual phase noise. The down converted signal is amplified by a baseband amplifier and the amplified baseband signal at node 4 is sampled and digitized by the ADC. A digital signal processor or a computer can be used to analyze and calculate the magnitude of each frequency component within the digitized signal. Figure 3 2 shows an example of the baseband time domain signal and frequency domain spectrum using a Dopple r radar on a CMOS chip. The radar chip uses the homodyne quadrature architecture and has two baseband output channels ( I/Q ). Since the same transmitted signal is used as the LO to down convert the received signal which is phase modulated by the physiologic al movement, there is no frequency offset in the baseband. The timing delay does not affect the detection either. Therefore, no synchronization mechanism is needed for the system. With the transmitted signal as LO for down conversion, the range correlation effect [33] minimizes the distortion of the baseband from LO phase noise. Without the range

PAGE 43

43 correlation effect, the term in the baseband signal B(t) will change over time and distort the detection of the phase modulation by the physiological movement. In order to study the spectrum of the baseband signal, the sinusoidal function in Equation 3 3 can be expanded using Fourier series. The Fourier series representation of the phase modulated signal in Equation 3 3 is: (3 4) Figure 3 2 Baseband I/Q signals: time domain signal and frequency domain spe ctrum. From [57].

PAGE 44

44 From Equation 3 4 we can observe that the signal strength of the vital sign signals are determined by the harmonics as well as the intermodulation tones of the heart beat and respiration signals. For example, the detected heart beat sig nal is determined by the ( l = 1, k = 0) terms and its signal strength J 1 h 0 r is dependent on both m r and m h 3.3 Two D imensional Random Body Movement Cancellation The block diagram of the two dimensional vital sign detection system with D oppler radar array is shown in Fig ure 3 3 The measurement is performed from the four sides of the human body. When the human body roams randomly in the horizontal plane, the body movement generates a significant noise spectrum component in the frequency d omain of the output signal in every individual detector. By combining the random frequency shifts caused by body movement in the multiple detectors, the noise can be extracted and canceled in spectrum analysis. With the random body movement in presence, th e baseband signal detected by I and Q channel of the radar array can be modeled by complex time series as: (3 5.a) (3 5 b ) (3 5 c ) (3 5 d )

PAGE 45

45 Fig ure 3 3 Block diagram of the vital sign detection system with Doppler radar array. where x rk = m rk sin( r t ) and x hk = m hk sin( h t ), k = 1,2,3,4 are the respiration induced and heartbeat induced physiological movement amplitudes on the front chest wall, back, left side and right side of the human body ; r and h are the angular frequency of the respiration an d heartbeat; k for k = 1,2,3,4 are the wavelengths of the radar carrier signa ls (near 5.8 GHz in this paper); x b ( t ) = V x t and y b ( t ) = V y t are the x and y axis components of the planar body displacement D b ( t ) = x b ( t ) x + y b ( t ) y T he body movement resembles a random walk in a two dimensional space. The random variable V x = V x x + V y y that is the speed during a movement period, is approximated by the discrete

PAGE 46

46 random variable series uniformly distributed betw een 4 mm/s and 4 mm/s. The modeled random walk of the human body is shown in Fig ure 3 4 (a) as the inset. Fig ure 3 4 Simulation of 2 D random body movement cancellation. (a) Baseband spectra obtained from individual detectors when planar random walk of human body is present. The planar random walk of body is shown in the inset. (b) Baseband spectra recovered by two dimensional random body movement cancellation using radar array, showing respiration at 21 beats/min and heartbeat at 72 beats/min. T he pairs of physiological movements on opposite sides of the body, e.g. x r 1 and x r2 move in the same direction relative to their respective detecting radar. On the other hand, when the body is drifting toward one of the radars, it is moving away from the opposite one. The signs of the body displacements in each dimension are opposite because the movement directions are opposite relative to the pair of detectors. Since the baseband output signals in the radar array are in phase but the body movement

PAGE 47

47 components are 180 degree out of phase, by multiplying the four vectors S k ( t ), the noise from the two dimensional random walk of human body can be eliminated. Note that the different amplitudes m rk and m hk of respiration induced movement x rk and heart beat induced moveme nt x h k are summarized together in Equation 3 6, thus will not affect the cancellation technique. The processed time series of the baseband signal with pure respiration and heartbeat information is: (3 6 ) The simulated baseband spectra from the multiple detectors are shown in Fig ure 3 4 (a), and the recovered baseband spectrum is shown in Fig ure 3 4 (b). The simulation result verified the theory. 3.4 Sensitivity Improvement Using Do ppler Radar Array In addition to cancelling the planar random walk of the human body, the Doppler radar array approach of vital sign detection also effectively strengthens the vital sign components within the frequency domain of the signal. Harmonic analys is using Fourier expansion [ 82 ] on the recovered baseband signal in E quation 3 6 gives (3 7 )

PAGE 48

48 a nd (3 8 ) wh ere J n is the Bessel function of the first kind. Since the magnitude of e is 1 and is independent of the value of the sensitivity on respiration and heartbeat detection are determined by the value of the 1st Fourier coefficients at frequency r and h in Equation 3 7 i.e. 2 C 10 and 2 C 01 (a) (b) (c) (d) Fig ure 3 5 Amplitude of Bessel functions: (a) J 1 k m rk / ) ; (b) J 0 k m hk / ); (c) J 0 (4 k m rk / ) ; (d) J 1 (4 k m hk / )

PAGE 49

49 The behavior of the function J 0 k m hk / ) and J 1 k m rk / ) is presented in Figure 3 5 (a) and (b). The magnitude of human heartbeat induced movement m h is in the range of 0.1 mm, which k m hk / k m hk k m hk / = 0.15), thus J 0 k m hk / ) is close to 1. In the case of respiration detection, therefore, the coefficient C 10 can b e approximated by J 1 k m rk / k m rk / is small, J 1 k m rk / k m rk / increases. The combination of the baseband signals of multiple detectors described in Section II increases the value of k m rk / and thus improves the r espiration sensitivity which depends on C 10 In our simulation (for f k m rk goes from 1.2 mm to 2.6 mm), the multiple detectors approach versus the single detection from the front chest wall almost doubles the value of C 10 from 0.14 to 0.30. T he increase of C 10 as the number of detectors increases is shown in Fig ure 3 6 Fig ure 3 6 Respiration and heartbeat sensitivity improves as the number of detectors increases.

PAGE 50

50 In the case of heartbeat detection, the detection sensitivity is related to t he coefficient C 01 The behavior of the function J 0 (4 k m rk / ) and J 1 (4 k m hk / ) is presented in Figure 3 5 ( c ) and ( d ). In the range of the vital sign movement amplitude, typically from 0.1 mm to 3 mm, as m r and m h increase, J 0 (4 k m rk / ) decreases and J 1 (4 k m hk / ) increases. Since the change in J 1 is fast er than J 0 C 01 increases as m r and m h increase. In effect, the signal strength at heartbeat frequency is also improved by the multiple detector approach. In our simulation (for f k m rk goes from 1.2 k m hk goes from 0.1 mm to 0 .6 mm), the combination of multiple baseband signals increases the value of C 01 from 0.029 to 0.066. The increase of C 01 as the number of detectors increases is shown in Fig ure 3 6 3.5 DC Offset Compensation The Doppler radar array app roach for noncontact vital sign detection is not immune from the disturbance of DC offset. Fig ure 3 7 shows the distortion of heart beat information while DC offset with amplitudes of 20%, 40%, and 60% of the signal amplitude is present in different number of detectors. In fa ct, based on the simulation, as the number of detectors increases, DC offset will introduce more noise to the spectrum. In order to guarantee an accurate recovery of the vital sign spectrum, a DC offset compensation algorithm is developed to cancel the unw anted DC offset. Fig ure 3 8 (a) shows the baseband signal trajectory with unwanted DC offset in the constellation graph. Based on Equations 3 5.a 3 5.b the signal trajectory of the baseband signal without DC offset will be an arc on the unit circle. When unwanted DC offset is introduced by down conversion of the radar carrier wave reflected on a still object and the leakage from transmitting antenna to receiving antenna, the arc is shifted in the constellation graph. The compensation algorithm calculates the shift of the arc

PAGE 51

51 center and adds corresponding DC values to I and Q signals to move the arc back to the unit circle. Fig ure 3 8 (a) also shows the calibrated signal trajectory. Fig ure 3 8 (b) shows the baseband spectrum before and after the DC offset com pensation. Fig ure 3 7 Heartbeat spectra when DC offset is present in (a) one detector, (b) two detectors, (c) three detectors, and (d) four detectors. Vertical dash line marks the correct heartbeat frequency. Fig ure 3 8 Illustration of DC offset comp ensation algorithm. (a) Baseband signal trajectory before and after the DC offset compensation. (b) Recovered baseband spectrum before and after the DC offset compensation. Vertical dash line marks the correct heartbeat frequency.

PAGE 52

52 3. 6 Experiments The two d imensional noncontact vital sign detection system with Doppler radar array was tested in the lab oratory environment. The system consists of four individual noncontact detectors. Each detector consists of a transceiver as the radio front end, a baseband amp lifier as an interface to amplify and level shift the transceiver output, a data acquisition module to sample and digitize the baseband output signal, and a computer to perform the spectral analysis. All of the detectors are operating at the 5.8 GHz ISM ba nd. Note that the gains of the detectors are not necessary to be the same. If assuming the receivers are operating in the linear region, the difference in receiver gains will only introduce a scalar factor to the time series of the processed signal, thus w ill not affect the normalized spectrum of the recovered vital sign signal. A photograph of the noncontact vital sign detection system is shown in Fig ure 3 9 Fig ure 3 9 Photograph of the RF radar array and TX/RX antennas. The antennas of opposite facing transceivers use orthogonal polarization to prevent one unit from saturating and interfering of the other unit.

PAGE 53

53 In the experiment, the human subject was seated in the middle of the detection system setup, 0.5 m away from each of the vital sign detectors. A sampling frequency of 20 Hz is used. The spectra obtained by the individual detectors are shown in Fig ure 3 9 (a). The spectrum of the recovered vital sign signal is shown in Fig ure 3 9 (b). Fig ure 3 10 Two dimensional random body movement cancellation using multiple detectors array: (a) spectra measured from the front, back, left and right side detectors; (b) Recovered spectrum by the Doppler radar array, the heartbeat information is successfully recovered. In the figures, the magnitude of the spectrum was normalized for reading convenience. The unit of the horizontal axis is beats per minute. The spectrum of the rate at around 80 beats/min. From Fig ure 3 10 (a), the main Doppler frequency shift of the random body moveme nt can be seen at 17 beats/min. By applying the proposed

PAGE 54

54 algorithms, the main spectral component of the planar body movement is eliminated by the radar array approach, as shown in Fig ure 3 10 (b). 3.7 Li mitation of Sensitivity Improvement In Section 3.4 the sensitivity improvement feature of the four sensor Doppler radar array is demonstrated. By using four radar sensors, the detection sensitivit ies of both the respiration and heart beat are at least dou bled compared to the sensitivity when using only one sensor However, t he increase in sensitivity using multiple sensors is not limitless. By using more Doppler radar sensors than the four sensors in this dissertation the value of Bessel function J 0 and J 1 will be pushed to a point to produce diminishing detection sensitivity. In the case of respiration detection, the sensitivity is dependent on t he behavior of Bessel function J 1 k m rk / ) and J 0 k m hk / ) which is presented in Figure 3 1 1 (a). The Bessel function J 0 k m hk / ) at the frequency of 5.8 GHz is very close to 1 even with many operating sensors since the heart beat movement amplitude is very small (e.g., with 20 senso rs, k m hk = 2.4 J 0 k m hk / ) = 0.92) T herefore, the respiration signal strength C 10 can still be approximated by J 1 k m rk / ). When the number of sensors is less than a marginal number J 1 k m rk / ) increases k m rk / in creases. This Bessel function behavior produces the improvement of sensitivity discussed in Section 3.4 However, w hen the number of sensors is larger than a marginal number, J 1 k m rk / ) starts to decrease, thus the respiration sensitivity starts to dete riorate In our simulation (for f k m rk goes from 0.6 mm to 12 mm), the respiration amplitude peak s when there are 12 sensors in operat ion As a result, t he maximum sensitivity that a multiple radar system

PAGE 55

55 can achieve is 7.7 times that of the single detection from the front chest The saturation of C 10 as the number of detectors increases is shown in Fig ure 3 1 2 (a) (b) Fig ure 3 11 Amplitude of Bessel functions: (a) J 1 k m rk / ) and J 0 k m hk / ); (b) J 0 (4 k m rk / ) and J 1 (4 k m hk / )

PAGE 56

56 Fig ure 3 12 Respiration and heartbeat sensitivity peaks at 12 sensors and 8 sensors, respectively. In the case of heartbeat detection, the detectio n sensitivity is related to the behav ior of the function J 0 (4 k m rk / ) and J 1 (4 k m hk / ) which is presented in Figure 3 1 1 ( b) In the range of the vital sign movement amplitude, typically from 0.1 mm to 3 mm, as m r and m h increase, J 0 (4 k m rk / ) decreases and J 1 (4 k m hk / ) increases. When the number of detectors is less than a marginal number, the change in J 1 is faster than J 0 When the number of detector s is larger than a marginal number, the change in J 1 is slow er than J 0 I n effect, the signal strength at heartbeat frequency is peaked at a marginal number of detectors In our simulation (for f k m rk goes from 0.6 mm to 12 k m hk goes from 0.1 2 mm to 2.4 mm), the combination of multiple

PAGE 57

57 baseband signals results in a maximum sensitivity improvement at 8 detectors. T he maximum value of C 01 of the multiple detector approach is 5.5 times that of the single detector approach The increase of C 01 as the number of detectors increases is shown in Fig ure 3 1 2 3.8 Limitation of Real time Large Body Movement Cancellation By using the multiple Doppler radar array discussed in this disserta tion, cancellation of body movements with amplitude less than 1 cm and speed less than 12 mm/s is achieved. However, real time cancellation of large body movement still remains a challenge. When the large body movement is present, the baseband amplifier of the high sensitivity vital sign detectors is usually saturated. The saturation of the baseband signal results in the failure of the body movement cancellation algorithm. In order to prevent the saturation, the rec eiver gain and sensitivity need to be decr eased in real time. Therefore, there is a trade off between the detection sensitivity of the vital signs and the cancellation capabilities of the Doppler radar array. An intelligent software controlled amplifier should be implemented to balance these two f actors in real time and accomplish a real time large body movement cancellation system. 3. 9 Summary A Doppler radar array is proposed in this chapter to cancel the two dimensional human body movement in vital sign detections. By using the radar array, the detecting sensitivity for respiration and heartbeat is improved. An algorithm for compensating DC offset is introduced to ensure the proper operation of the two dimensional body movement cancellation. Experiments on human subject are performed to verify t hese techniques for two dimensional vital sign detection. The limitation of sensitivity

PAGE 58

58 improvement and body movement cancellation are demonstrated with simulation results.

PAGE 59

59 CHAPTER 4 SYSTEM LEVEL INTEGRA TION OF HANDHELD WIR ELESS NONCONTACT VIT AL SIGN SENSO R RADAR 4.1 Challenges of Portable Applications In daily applications of the multiple sensor platforms for noncontact vital sign detection, m any of the applications such as sleep apnea monitoring and baby monitor require integration of the entire system in small portable packages. T he realization of the detection in compact portable system becomes a new focus of interest. Although the integrations of the radio frequency front end have been reported in both board level [ 3 9 ] and chip level [ 4 5 ], most of the p reviously reported systems rel y on computers for real time processing or post processing of the signals. In this chapter we report a n on contact vital sign detector for handheld applications without the need of computers The radio frequency transceiver, th e baseband analog circuit, and the power management circuit are integrated on a single printed circuit board. The baseband signal processing board includes an ARM7TDMI microprocessor and its peripherals. The spectrum of the baseband signal can be channeled through the PIO controller to a commercial LCD display. All of the above components can be potentially integrated together in an easy to carry package. A method to evaluate the overall noise performance of the handheld system will be reported in this chap ter The overall noise performance evaluation will provide quantitative guidelines on system archite cture choice, components design and detection accuracy optimization. We will perform a thorough noise analysis on the quadrature direct conversion vital si gn detector. A key design choice of mixer is deriv ed based on this noise analysis The requirement of output SNR is added as a design

PAGE 60

60 guideline for vital sign detectors. Simulations and experiments related to the guideline are performed and the results wil l be discussed. 4. 2 Vital Sign Detection System Architecture The block diagram of the vital sign detection system is shown in Fig ure 4 1. Typically, a noncontact vital sign detection system consists of a transceiver as the radio frequency front end, a base band amplifier as an interface to amplify and level shift the transceiver output, a digital signal processor for spectrum analysis, and a display unit. The quadrature transceiver, the two stage baseband amplifier, and the power management circuit are integ rated on a single Rogers printed circuit board as the vital sign detection radar. The size of the radar is 6.8 cm 7.5 cm, which is suitable for handheld applications. Due to the range correlation effect [3 3 ], a free running voltage controlled oscillator (VCO) can be used to generate the radio frequency signal. As demonstrated in [ 39 ], due to the non linear phase modulation effect, there is an optimal carrier frequency for a subject with certain physiological movement amplitude. This optima l frequency varies from several GHz to the lower region of Ka band. Considering the cost for the handheld radar, the system was designed to have a carrier frequency from 4 7 GHz. Four VCOs covering different frequency ranges within the same package can be implemented onto the board. The VCOs guarantee the phase noise to be always lower than 101 dBc/Hz at 100 kHz offset, and the maximum output power is more than 2 dBm over the entire frequency tuning range. After the Wilkinson power divider, one half of the power is transmitted through the transmitting antenna (TA) and the other half of the power is further amplified and used to drive the mixer in the receiver chain. The receiver chain contains the receiving antenna (RA), a 3.5 7 GHz low noise amplifier (L NA), two stages of adjustable gain block, and the down conversion

PAGE 61

61 mixer, which is a compact I/Q mixer utilizing two standard double balanced mixer cells. The radio frequency part of the receiver chain has an adjustable 30 dB gain control range. The down co nverted baseband quadrature signals are amplified by a two channel two stage amplifier, which is realized in a space saving package with four unit gain stable operational amplifiers. Fig ure 4 1. Block diagram of the vital sign detection system. Except for the VCO and the passive I/Q mixer, all the other components have a single supply voltage of 5 V. The VCO is 3 V supplied and requires a 0 to 10 V tuning voltage. Therefore, 5 V and 3 V fixed output voltage regulators are implem ented, and an adjustable output regulator with up to 11 V output voltage is used to tune the carrier frequency. Either a 6 9 V wall plug or a 9 V battery can be used to power up the radar. A photograph of the complete RF transceiver board and signal proc essor board is shown in Fig ure 4 2. A detailed block diagram of the transceiver board is shown in Figure 4 3. A bill of material of the transceiver is listed in Table 4 1. The amplified baseband IQ signals are sampled by the AT91SAM7S64 microprocessor. The on chip

PAGE 62

62 10 bit Successive Approximation Register (SAR) Analog to Digital Converter (ADC) converts the sampled baseband signal to a digital format. The sampling rate can be set to 2 32 Hz to guarantee sufficient headroom over the Nyquist frequency of com mon vital signs (respiration and heart beat). A 256 point radix 2 fixed point Fast Fourier Transform (FFT) is implemented on the AT91SAM7S64 microprocessor to analyze the magnitude of each frequency within the vital sign signal. The choice of window size i s optimized to provide maximum frequency resolution and minimum execution time. Spectrum results can be channeled through Parallel Input/Output (PIO) Controller to a LCD display such as DisplayTech 64128H LCD to show the measurement result. The detail of t his baseband signal processor design and spectrum analysis algorithm will be presented in the Section 4.3 Fig ure 4 2. Photograph of the RF transceiver board and signal processor board

PAGE 63

63 Figure 4 3 Block Diagram of the RF transceiver board Table 4 1 R F t ransceiver b oard b ill of m aterial Block Vendor Specification VCO1 Hittite 4.46 5.0 GHz, 105dBc/Hz @100 kHz phase noise, 4 dBm output power VCO2 Hittite 5.0 5.5 GHz, 103dBc/Hz @100 kHz phase noise, 2 dBm output power VCO3 Hittite 5.6 6.1 GHz, 102dBc/Hz @100 kHz phase noise, 2 dBm output power VCO4 Hittite 6.1 6.72 GHz, 101dBc/Hz @100 kHz phase noise, 4.5 dBm output power Switch Hittite DC 8 GHz, 40 dB isolation @6 GHz, 1.8 dB insertion loss @6 GHz, SP4T Gain Block RFMD DC 8 GHz, 15. 5 dB maximum gain, 14.5 dBm P1dB @6Ghz Mixer Hittite 4 8.5 GHz, 50 dB LO to RF isolation, 40 dB image rejection LNA Hittite 3.6 7.0 GHz, 16 dB gain, 2.5 dB NF

PAGE 64

64 4.3 Baseband Signal Processor Design The baseband signal processor is i mplemented mainly wit h an Atmel AT91SAM7S64 microprocessor. The embedded FLASH holds the spectrum analysis code and the on chip RAM stores the data before and after the signal processing. The d own converted baseband signal from the preceding radar receiver stage is fed into th e analog input channels AD6 and AD7 of the ADC. This input signal is in the range of 0.1 3 V. Therefore, the reference voltage of the ADC is set to 3.3 V to cover the dynamic range of the signal. The ADC sample and hold time is set to 600 ns which is min imal and necessary for the ADC to guarantee the best converted final value between two I/Q channels selection. The conversion resolution is 10 bit which provides 1024 quantization levels. Conversions of the active analog channels are initiated with a hard ware trigger from the Time Counter channels in the microprocessor. The interval between two successive triggers is the sampling period. The sampling rate can be set accurately by configuring the Time Counter. In this application the sampling frequency is w ithin the range of 2 32 Hz. The four most significant digits of the conversion result are shown with the LEDs for testing purposes. The digitized baseband signal is windowed and processed by the spectrum analysis code in FLASH. The resultant spectrum is stored in the RAM and can be channeled to the LCD display through PIO controller of the microprocessor. A simple power supply circuit is used to stabilize and adjust the voltage from 5 V to 3.3 V, the input VDD of the AT91SAM7S64 microprocessor. The signal processing ability of the microprocessor is mostly realized by the spectrum analysis code and is described in details below.

PAGE 65

65 Fast Fourier Transform (FFT) is the core of the spectrum analysis algorithms in our application. A standard 256 point radix 2 fix ed point FFT is utilized. The algorithm includes three sub blocks: sine/cosine lookup table generation, bit reverse of the input windowed signal, and iterations of butterfly computations. The flow diagram of the algorithm is shown in Fig ure 4 4 Fig ure 4 4 Flow diagram of the spectrum analysis algorithm

PAGE 66

66 The AT91SAM7S64 microprocessor core is running at 40 MHz. This speed is significantly slower than that of personal computers previously used for signal processing. Therefore, two adjustments are needed to speed up the FFT calculation for displaying measurement results real time. First, the bit reverse algorithm is designed on the bit manipulation level and takes advantage of the bit wise operation offered by the microprocessor Second, the coefficients in the butterfly computation have a repeated pattern. Therefore, the coefficients are calculated ahead of time and stored in a lookup table in the RAM. This will speed up the real time FFT computation significantly. A photograph of the signal processor board is shown in Figure 4 5 Figure 4 5 Photo of the digital signal processor board

PAGE 67

67 4.4 Receiver Chain Noise Analysis The detector is divided into three sub systems including RF front end amplifiers, mixer with LO, and baseban d amplifiers. Table 4 2 lists an example of receiver chain components noise specifications. Table 4 2 Receiver c hain c omponents n oise s pecification LNA Gain Block Mixer BB Amplifier Component Hittite 318MS8G RFMD NBB 400 Hittite 525LC4 Maxim MAX 4478 Gain [dB] 16 15.5 7.5 43 F [dB] 2.5 4.3 7.5 / Cumulative F [dB] 2.5 2.6 2.61 / V n / / / 21 Cumulative Vn 9.51 9.62 9.63 3257 4.4.1 LNA and Gain Block LNA and gain block make up a cascaded RF system. Their function is to amplify and scale the received signal to a level that will be acceptable by the mixer. Stationary noise propagates in these two 50 ohm terminated components and can be measured by noise figure. The cumulative noise figure of these two components can be cal culated based on the cascaded noise figure equation and is listed in Table 4 2 4.4.2 Mixer with LO Input In Doppler radar detection of human vital signs, the baseband signal bandwidth is typically less than a few Hertz. Conventional Gilbert type active mi xers contain several noise sources: the transconductor noise, the LO noise, and the noise from the switching transistors. These noise sources establish an unacceptable noise figure in the

PAGE 68

68 interested baseband spectrum. Therefore, Gilbert type active mixers are not suitable for the vital sign detection receiver. Passive mixers avoid the transconductor stage in the active mixers and have no dc bias current. This feature minimizes the flicker noise at the mixer output. Therefore, a passive mixer was chosen for direct conversion vital sign detection. Figure 4 6 is a noise figure comparison between a Gilbert active mixer and a passive mixer designed for a 5.8 GHz radar receiv er chip in 0.13 m CMOS [4 3 ] [4 8 ] Figure 4 6 Noise figure of active mixer and passive m ixer in 0.13 um CMOS. The difference between active mixer and passive mixer noise figure at 1 Hz is 64.5 dB. In order to minimize the flic ker noise of the passive mixer, the gate source voltage of the switching transistor should be close to V th This break before make bias technique will minimize the dc bias current in the switching transistors. For the switching transistor size, there is a tradeoff between noise figure and capacitive load to the preamplifier stage. To provide appropriate noise figure in th e interested vital sign

PAGE 69

69 bandwidth, large switching transistors are used and they produce relatively large capacitive loads to the preamplifier. As a result, there is a large capacitive load to the preamplifier stage. This is the reason that a source follow er buffer was used at the preamplifier to drive th e mixer in voltage driven mode. 4.4.3 Baseband Amplifier The baseband amplifier is an interface that amplifies the transceiver output to a level that will be acceptable by the ADC. Similar to mixer, an impo rtant noise source disturbing the vital sign information in this sub system is the flicker noise from the amplifier and is measured by noise voltage spectral density V n,BB In order to minimize the noise, low noise operational amplifiers should be used. In the example used for study, an op amp with input referred noise voltage spectral density V n,BB at 10 Hz is used. Another important noise source is the thermal noise of the external resistor in the feedback loop of the amplifier. The feedback resistor has a value of 140 combined noise voltage spectral density of this sub system is the square root of the sum of the squared values of the two individual noise v oltage spectral densities and can 4.4.4 Complete Noise Performance Evaluation Model The purpose of the complete noise evaluation is to develop a single figure of merit that measures the noise in the complete vital sig n detect or. I t is the baseband signal at Node 4 that is sampled and analyzed to generate the spectrum of vital sign information. Therefore, the measure of the complete system noise performance is the signal to noise ratio of the signal at this node. Using this fig ure of merit, one can predict and

PAGE 70

70 optimize the overall system noise performance as well as the vital sign detection accuracy. The noise in RF front end of the vital sign detector is measured by noise figure. The noise in baseband amplifier is measured by n oise voltage spectral density. In order to combine the noise in the two sub systems, the RF input referred noise figure can be converted to noise voltage spectral density at RF output Node 3 using equation (4 1) where G RF is the gain of the RF front end, and V n,ANTENNA is the RMS value of the noise voltage spectral density looking int 4 4 ]. At 25 o C, V n,ANTENNA equals end noise appears as additive noise on the baseband signal and can be summed with the baseband amplifier noise voltage spectral density V n,BB The combined noise voltage is then amplified and added to the signal at sampler input Node 4. The noise voltage spectral density at Node 4 can be described by equation (4 2) w h ere A BB is the baseband amplifier gain of 14 1 The accumulative noise voltage spectral densities at 10 Hz are listed in Table 4 1 For the baseband amplifier, the input referred noise voltage density adjusted by feedback resistor thermal noise is dependent on the frequency, especially in the flicker noise region. The data of this frequency dependent relationship can be found in the data sheet of the baseband amplifier. The total output noise voltage is obtained by integrating the output noise voltage spectr al

PAGE 71

71 density over the baseband amplifier bandwidth. The output signal to noise ratio at Node 4 can be calculated as ( 4 3) where V signal is the baseband signal voltage in RMS, and B is the baseband amplifier bandwidth and equals to 70.4 k Hz. Using the example detector assuming the signal voltage being 0.1 V, the overall output SNR is 41.6 dB. If the ADC noise voltage of 5 mV is included in the noise analysis, the overall output SNR is 26 dB. 4.5 Experiments The integrated vital sign detector was tested in the lab oratory environment. Two experiments have been performed using the integrated system. First, an actuator progra mmed to move in a pattern consisting of a two tone sinusoidal wave was placed 3 meter away from the detector. The spectrum resulted from this experiment shows the accurately the frequency and amplitude of periodic movements. Second, a human subject was seated at 0.5 m away and faced the detector. The subject was breathing normally throughout the duration of the testing. The vital sign detector recorded the vital signal and did the spectrum analysis on the signal. In the second experiment, we discovered a trade off between spectrum sharpness and spectrum response speed as well as stability. This trade off can be used as a guideline in choosing the sampling frequency for different applications. The transceiv er board is configured to run at 5.8 GHz in the experiments.

PAGE 72

72 4.5.1 Two tone Actuator Movement The diagram illustrating the experiment setup is shown in Figure 4 7. In this experiment, the actuator was programmed to move in a pattern determined by the funct ion (4 4) where m 1 = 4 mm, m 2 = 2 mm, f 1 = 0.1 Hz, and f 2 = 0.5 Hz. The baseband signal can be written as (4 5) where is the wa velength of the carr ier (0.0517m in our case) and is the total residue phase noise. The actuator is placed 3 meter s away from the integrated vital sign detector. The sampling frequency of the signal processor is set to be 12.8 Hz. Figure 4 7. Two tone a ctuator movement experiment setup The theoretical and measured baseband signals are shown in Fig ure 4 8 (a) and (b). The theoretical and measured spectrums generated by the baseband signal

PAGE 73

73 processor are shown in Fig ure 4 8 (c) and (d). From the figure, the t wo tones (0.1 Hz and 0.5 Hz) can be identified. (a) (b) (c) (d) Fig ure 4 8 Theoretical results vs. experimental results of the two tone actuator experiment : (a) theoretical baseband signal in two tone experiment; (b) measured baseband signal in two tone experiment; (c) theoretical spectrum in two tone experiment; (d) measur ed spectrum in two tone experiment. 4.5.2 Human Respiration and Heart Beat Measurement The human subject was seated at 0.5 m away from the vital sign detector. In the experiment, two sampling frequencies were used: 25.6 Hz and 6.4 Hz. The diagram illustrat ing the experiment setup is shown in Figure 4 9. The baseband signal and its spectrum obtained by the baseband signal processor are shown in Fig ure 4 10 In the spectrum figure, the magnitude of the spectrum was normalized for reading convenience. The unit of the horizontal axis is beats per minute. The spectrum of the

PAGE 74

74 the heart beat is around 84 beats/min. Fig ure 4 9 Human respiration and heart beat measurement setup. (a) (b) (c) (d) Fig ure 4 10 Detected baseband signal and spectra in non contact vital sign detection. (a) baseband signal detected with a sampling rate of 25.6 Hz; (b) baseband spectrum detected with a sampling rate of 25.6 Hz; (c) baseband signal detected with a sampling rate of 6.4 Hz; (d) baseband spectrum obtained with a sampling rate of 6.4 Hz.

PAGE 75

75 4.5.3 Guideline for Selecting t he Sampling Frequency In previously reported vital sign detectors, the signal processing was h andled by [39], [ 42 ] and [ 4 5 ] used a l arge window size of 10240 and a sampling rate of over 20 Hz to achieve a smooth spectrum. However, the handheld version of a vital sign detector has a limit on the size of the window because the relatively low speed microprocessor cannot calculate large wi ndows very quickly. The handheld vital sign detector in this paper utilizes a 256 point window. Therefore, the sharpness of the spectrum is now dependent on the sampling rate selected. As shown in Fig ure 4 10 the spectrum with the higher sampling rate is not as sharp as the spectrum with the lower sampling rate. Therefore, if the application needs sharper spectrum, a low sampling rate should be selected. However, the lower sampling rate results in a l onger measurement period and prolongs the response of t he spectrum to the change in vital sign. Also, any strong interference in this long period will destroy the spectrum. This dynamic implies that there is a tradeoff between spectrum sharpness and spectrum response speed as well as stability for handheld vi tal sign detectors. The guideline for selecting the sampling rate is: If the application has a relatively stationary subject and needs accurate measurement, a low sampling rate is suitable. An example is the sleep apnea monitoring; if the application needs fast recognition and quick response time, a higher sampling rate should be used. An example is the search and find rescue mission. 4.5.4 The Effect of Output SNR on Detection Accuracy The above model was used to simulate the effect of overall output SNR o n the detection of vital signs. A human subject was seated 1 meter away in the experiment. d an amplitude of m r = 0.8 mm and a frequency of 19

PAGE 76

76 m h = 0.3 mm and a frequency of 72 beats/min. The carrier frequency of the Doppler radar is set at 5.8 GHz. Fig ure 4 11 presents the baseband signals and spectrums without noise and with a low output SNR. (a) (b) (c) (d) Fig ure 4 11 Simulated baseband signal and spectrum in non contact vital sign detection. (a) baseband signal without nois e. (b) baseband spectrum without noise. (c) baseband signal with SNR = 13 dB. (d) baseband signal with SNR = 13 dB. As shown in Fig ure 4 11 with a low SNR, the noise floor of the system overwhelms the heartbeat signal. A low SNR can be caused by either lo w vital sign signal strength or high overall noise voltage at the baseband output. The strength of the

PAGE 77

77 heartbeat signal is normally 5 to 10 times weaker than that of respiration. Therefore, in order to guarantee the detection of heartbeat, the received sig nal level needs to be roughly 14~20 dB higher than the required signal level for respiration detection only. 4.5.5 The Trade off between Output SNR and Detection Accuracy The SNR is related to the detection distance according to a two way Radar range equa tion. A l ong detection distance results in low signal strength and low detector SNR. We can verify this effect by conducting an experiment on a short range low power noncontact vital sign sensor node [ 53 ] and measuring the vital sign signals in different d istances. In order to calculate the SNR of the measured signal, a band pass filter and a band stop filter is used to separate the signal and noise. The SNR is calculated as the ratio of the variance of the signal and the variance of the noise. As shown in Fig ure 4 12 the measured SNR will decrease by 15.3 dB when the detection distance is increased by 60.6%. To raise the SNR of the received vital sign signal, at least one of the following methods should be used: (1) an increase in transmitter output power; (2) closer measurement distance; (3) larger physiological movements; or (4) a lower receiver noise. For a specific measurement setting (human subject and measurement device), the latter two conditions are relatively fixed. Therefore, based on the first tw o conditions, the SNR data should be provided as guidelines for experiment references. Fig ure 4 1 3 is a chart showing the simulated result of SNR at the measurement distance from 10 cm to 40 cm for the short range low power vital sign sensor node. The char t shows three set s of simulation results with three output power settings.

PAGE 78

78 Fig ure 4 12 Detected baseband signal and spectrum in non contact vital sign detection. (a) baseband signal detect ed at 16.5 cm, measured SNR is 26 .2 dB; (b) baseb and spectrum obtained at 16.5 cm; (c) baseband signal detected at 26.5 cm, measured SNR is 10.9 dB; (d) baseband spectrum obtained at 26.5 cm. Fig ure 4 1 3 Simulated receiver output SNR

PAGE 79

79 4.6 Summary An integrated noncontact vital sign detector for handhe ld applications is demonstrated. A low cost, low power, and small size signal processor is developed to perform the spectrum analysis task. N oise analysis on the quadrature direct conversion vital sign detector is demonstrated. The noise characteristics in the detector sub systems are analyzed and are combined to form an overall noise performance evaluation of the vital sign detection system. This integrated system enables the vital sign detection to be integrated in handheld devices. Experiments on both hu man subject and programmed actuator are performed to verify the accuracy of the detection. The guideline on selecting the sampling frequency for different application is described. A key design consideration in selecting mixer for vital sign detection is p resented. The guideline of detector SNR is introduced. The wireless noncontact detection system can be used widely in applications including medical, search and rescue, and military applications.

PAGE 80

80 CHAPTER 5 INTEGRATED VITAL SIGN RADAR SENSOR WITH ON BOARD ANTENNA 5.1 In tegration of Vital Sign Radar and Antenna s Although the radar front end, the baseband analog circuit and digital signal processor (DSP) h ave been integrated on a single printed circuit board and shown satisfactory performance, the integratio n of the antennas on board is of great interest Currently, the antennas used to transmit (TX) and receive (RX) the RF signal are patch antennas with an operating frequency at 5.8 GHz. Integrating the antennas on board requires the designer to solve the ch allenge of the coupling between the TX and RX antennas. The m ethods to minimize the coupling will be discussed in this chapter. Efforts on coupling minimization will help to reduce the DC offset and prevent the leakage from saturating the receiver. The ori entation of both antennas will be further investigated to minimize substrate coupling. 5. 2 Transmitting and Receiving Antenna Array s Design Both t he transmitting (TX) and the receiving (RX) antenna are designed to be patch antenna arrays A patch or micros trip antenna array is a low profile antenna array that has a number of advantages over other antennas. I t is lightweight, inexpensive, and easy to integrate with accompanying electronics thus it makes the perfect candidate to integrate with the Doppler ra dar circuits. Figure 5 1 shows a patch antenna array model designed in Ansoft Designer. The patch antenna array is designed for 5.8 GHz operation on a Rogers RO4350 B substrate with 3.48 dielectric constant and 0.032" thickness. The equation t o determine t he initial setting of the width (W) and length (L) of the microstrip patch antenna are

PAGE 81

81 (5 1) (5 2) Here f r is the resonant frequency r is the dielectric constant of the substrate reff is the effective dielectric constant of the substrate, is the length of fee d line, and v 0 is the sp eed of light. Fig ure 5 1 P atch antenna array model used in on board antenna design. The patch without the feeding network was simulated in Ansoft HFSS to adjust W for resonance at 5.8 GHz. The input impedance of the feed lines to the patches w as simulat ed by placing a at the patch edge. By changing the trench length the input impedance was The power distribution lines are simulated to match the impedance change caused by the branching. Finally, a right

PAGE 82

82 angle quarter wave length transformer was used to match the input impedance of the first power distribution branch terminal The radiation pattern and S11 of the finalized patch antenna is shown in Figure 5 2 and Figure 5 3 The final dimensions of the patch antenna array are listed in Table 5 1. Fig ure 5 2 P atch antenna array radiation pattern. Maximum gain 11.5 dB is achieved. Fig ure 5 3 S11 of patch antenna array. The antenna resonates at 5.8 GHz.

PAGE 83

83 Table 5 1 Dimensions of the patch antenna array. Design Parameter Parameter Value (mm) Note W 16.4 Patch width L 12.9 Patch length WSlot 3.2 Slot width LSlot 4.3 Slot length Dx 34 Patch x axis separation Dy 34 Patch y axis separation Lf1 4 Patch feed line length W4 3.3 Quarter wave length transfo rmer width lambda4 8.2 Quarter wave length transformer length W5 2.3 Right angle Q W length transformer width lambda5 6.7 Right angle Q W length transformer length 5.3 Orientation of the TX and RX A ntennas The coupling between TX and RX patch ante nna arrays is a function of the position of on e array relative to the other and the relative orientation of them [ 50 ]. When the two patch antenna arrays are placed along the H plane, the mutual coupling between the two elements are minimum. An illustration of the H plane orientation is presented in Figure 5 4 At microwave frequencies surface waves along the air dielectric interface contribute mainly to the mutual coupling In the H plane orientation the fields in the space between the elements are primar ily TE and there is not a strong dominant mode surface wave excitation thus produc ing less coupling between the arrays

PAGE 84

84 Fig ure 5 4 H plane p atch antenna array orientation. 5. 4 Simulation of the Coupling between TX and RX A ntennas The mutual coupling be tween two rectangular microstrip patches in H plane orientation can be found to be [50] (5 3) where Z is the cent er to center separation between the slots J 0 is the zero order Bessel function of the first kind L is the separation of patches along E plane, and Z is distance of patches along the H plane. According to the equation, the mutual coupling between the two patches will decrease whe n the separation is increasing.

PAGE 85

85 The mutual coupling between two on board patch antnnas can be calculated by a simulation set up in Ansoft HFSS Figure 5 5 shows the TX and RX antennas model designed in Ansoft HFSS The two antennas are in H plane orientation and separated by a distance of s. The mutual inductance between the two antennas is simulated with a separation from 115 mm to 155 mm The mutual coupling or leakage between the two antennas S 12 over frequency band from 5. 5 GHz to 6 .1 GHz is presented in Figure 5 6 By using the simulation result, an estimate of the on board antenna isolation can be determined. The on board antennas on the fabricated integrated radar board are designed to have a separation of 140 mm. The simulate d isolation between the TX and RX antennas are 3 9 dB at 5.8 GHz Fig ure 5 5 Mutual coupling simulation model in Ansoft HFSS

PAGE 86

86 Fig ure 5 6 S12 of the on board TX and RX patch antenna array 5. 5 System Integration of the Vital Sign Detector with On board Antenna The quadrature transceiver, the baseband amplification circuits, the power management circuit, and the TX and RX antennas are integrated on a single Rogers RO4350B printed circuit board. The size of the integrated portable radar is 20 cm 7cm T he TX and RX antennas are placed at the two sizes of the quadrature transceiver to further reduce the interference between the antennas. The coupling capacitor connected between the RF front end and baseband amplifiers is fine tuned from 10 F to 1 F Thi s coupling capacitor combined with the input impedance of the baseband amplifiers forms a high pass filter. By using the lower capacitance, the cutoff frequency of the filter is increased from 0.13 Hz to 1.3 Hz, thus it rejects the respiration signal and e nhanced the heart beat signal.

PAGE 87

87 The integrated radar draws a current of 0.24 A from the power supply and has a low power consumption of 2.2 W. The output power of the transmitter is 0 dBm which is within the limit set by IEEE RF Safety Guideline. The photo graph of the integrated vital sign radar sensor is show in Figure 5 7. A real time vital sign monitoring software is designed for the integrated radar. The software uses digital filtering to separate the time domain signal of the respiration and heart beat signal. A screen shot of the software is shown in Figure 5 8. Vital sign detection on a human subject is successfully achieved with the integrated hardware and software. The detection results are shown in Figure 5 8. Fig ure 5 7 Photograph of the integ rated vital sign radar sensors with on board antennas.

PAGE 88

88 Fig ure 5 8 Photograph of the real time integrated vital sign radar software. 5.6 Low power Design, Link budget, and Emission Safety As shown in Figure 4 3, the power management circuits supply the p ower consumption of the radar board. It consists of the power chip Maxim MAX603 and Maxim MAX604. The power chips convert the 9 V power from the wall plug to 3.3 V and 5 V. The number of power chips is determined by the total power needed and the maximum p ower each power chip can supply. The most power hungry components on the board are the RF amplifiers. Each RF amplifier (RFMD NBB 400) consumes a current of 50 mA The low power design goal of the vital sign detection radar requires using the minimum numb er of RF amplifier s to achieve su fficient detection sensitivity. In order to determine the minimum number of RF

PAGE 89

89 amplifiers, a link budget analysis is needed to determine the minimum gain needed in the RF receiver chain. Table 5 2 is a list of detailed dat a to calculate the received power using the link budget method The equation for received power estimate is (5 4) where P r is the received pow er, P t is the transmitted power, G t and G r are antenna gain s of the TX and RX antennas, is the signal wavelength in air, is the rada r cross section of the target. The Radar cross section (RCS) is a measure of how detectable an object is with a radar. A larger RCS indicates that an object is more easily detected. The RCS of the human vit al sign is estimated to be 0.01 m 2 It will give a received signal power of 20 dB lower than that using a RCS of 1 Using the equation, the received sign al power is estimated to be 7 7 dBm Table 5 2 Received RF power estimat e for 5.8 GHz integrated vital sign sensor. Value Frequency 5.8 GHz Transmitting Power 0 dBm Antenna Gain 10 dB Radar Cross Section 0.01 m 2 Distance 3 m Received Power 7 7 dB m

PAGE 90

90 In order to guarantee a successful detection of vital sign signals, especially the heart bea t signal, sufficient RF front end gain need s to be present in the receiver chain. Normally the input of the baseband amplifier should be larger than 2 mV ( 41 dBm). From the result of the link budget analysis, the RF gain of the receiver should be larger t han 36 dB. To reach this minimum gain requirement, two receiver RF amplifiers are used in the receiver chain of the integrated vital sign radar sensor. The total power consumed by the individual radar sensor is 2.1 W. Two MAX 604 and one MAX 603 are used t o provide the power. The radiation power of the sensor should meet the IEEE safety standard. Shown in Figure 5 9 is the IEEE RF safety guideline [ 86 ]. The maximum power density at 5.8 GHz in uncontrolled environments should be lower than 1 mW/cm 2 Shown in Figure 5 10 is the power density of the integrated noncontact vital sign detector at distances from 10 cm to 200 cm At any detection distance the radiation of the integrated detector is lower than 1 mW/cm 2 thus it complies with the IEEE RF safety standa rd. Fig ure 5 9 IEEE RF safety standard C95.1 2005.

PAGE 91

91 Fig ure 5 10. Power density of the integrated noncontact vital sign detector. 5 7 Summary An integrated noncontact vital sign sensor with on board antenna is demonstrated. A 5.8 GHz patch antenna array is designed for functioning as the individual antenna on radar board. The theory of mutual coupling between the TX and RX patch antennas are discussed. The consideration of the relative orientation of the antennas is discussed. A pair of H plane aligned p atch antenna arrays is modeled and the simulation results are used as guidelines in designing the on board antennas. This integrated system is fabricated on a single PCB. Accompanying software is programmed in Labview. Experiments on a human subject are pe rformed to verify the functionality of the sensor.

PAGE 92

92 CHAPTER 6 CONCLUSIONS The theory and implementation of multiple sensor platform techniques in hydrogen sensing and two dimensional noncontact vital sign detection are presented in this dissertation The im plemented hydrogen sensing system can detect and display the detected hydrogen density from the six sensors in real time. A use r friendly program has been developed to share the data collected by base station to Internet, so that the data can be analyzed a nd monitored from anywhere with an Internet connection. Field tests show that the low power hydrogen sensor can work stably and react quickly to possible hydrogen leakage. The implemented noncontact vital sign detection system can detect and display the d etected vital sign signals from the four sensors The multiple radar system can cancel the two dimensional human body movement in vital sign detections. By using the radar array, the detecting sensitivity for respiration and heartbeat is improved. An algor ithm for compensating DC offset is introduced to ensure the proper operation of the two dimensional body movement cancellation. An integrated noncontact vital sign detector for handheld applications is demonstrated. The noise characteristics in the detecto r sub systems are analyzed and are co mbined to form an overall noise performance evaluation of the vital sign detection system. The noncontact vital sign detector is further integrated with the transmitting and receiving antennas. The mutual coupling betw een TX and RX antennas are studied. consumption and radiation is studied. A real tim e vital sign detection software programmed in Labview is demonstrated.

PAGE 93

93 LIST OF REFERENCES 1. G.W. Hunter, P.G. Neudeck, R.S. Okojie, G.M. Beheim, V. Thomas, L. Chen, D. 5, pp. 247 257, Oct. 2007. 2. metal oxide no. 21, pp. 3203 3205, Nov. 1995. 3. A. Lloyd Spetz, L. Unus, H. Svenningstorp, P. Tobias, L. G. Ekedahl, O. Larsson, A. Gras, S. Savage, C. Harris, P. Mrtensson, R. Wigren, P. Salomonsson, B. 25, May 2001. 4. chottky diode gas sensors on n 2, pp. 164 168, July 1999. 5. III Nitride Based Gas 45, May 20 01. 6. GaN/AlGaN 430, Dec. 2002. 7. J. Schalwig, G. Mller, U. Karrer, M. Eickhoff, O. Ambacher, M. Stutzmann, L. Grgens, GaN Schottky 1224, Feb. 2002. 8. M. Eickhoff, J. Schalwig, G. Steinhoff, O. Weidmann, L. Gorgens, R. Neuberger, M. Hermann, B. Baur, G.Muller, O. Ambacher,M. based on pyroelectric AlGaN/GaN heterostructures Phys. Status Solidi C, vol. 0, no. 6, pp. 1908 1918, Sep. 2003. 9. elf moving 741, Feb. 2003. 10. H.T.Wang, B.S. Kang, F.Ren, R.C. Fitch, J. Gillespie, N. Moser,G. Jessen, current det ection of hydrogen at room temperature with AlGaN/GaN high electron 172107, Oct. 2005. 11. sensing system for no. 2, pp. 329 333, Mar. 2005.

PAGE 94

94 12. B.S. Kang, R. Mehandru, S. Kim, F. Ren, R. Fitch, J. Gillespie, N. Moser, G. Jessen, T. Jenkins, R. Dettmer, D. Via, A. Crespo, B.P. Gila, C.R. Abernathy, S.J. P earton, induced reversible changes in drain current in Sc2O3/AlGaN/GaN high 4637, Jun. 2004. 13. based metal oxide semiconductor diode vol. 84, no. 7, pp. 1123 1125, Feb. 2004. 14. State Electron., vo l. 47, no. 6, pp. 1069 1073, Jun. 2003. 15. Jun Rui Huang, Wei Chou Hsu, Yeong Jia Chen, Tzong Bin Wang, Kun Wei Lin, Huey Ing Chen, Wen vol. 117, no. 1, pp. 151 158, Sep. 2006. 16. MOS and Schottky W/Pt 104, no. 2, pp. 232 236, Jan. 2005. 17. K. Matsuo, N. N 244, no. 1 4, pp. 273 276, May 2005. 18. en State Electron., vol. 49, no. 8, pp. 1330 1334, Aug. 2005. 19. M. Ali, V. Cimalla, V. Lebedev, H. Romanus, V. Tilak, D. Merfeld, P. Sandvik, O. vol. 113, no. 2, pp. 797 804, Feb. 2006. 20. O. Weidemann, M. Hermann, G. Steinhoff, H. Wingbrant, A. Lloyd Spetz, M. sensitive Pd:GaN Schottky diodes 775, Jul. 2003. 21. Lars Voss, B.P. Gila, S.J. Pearton, Hung 2373 2377, Nov. 2005. 22. J. AlGaN/GaN Schottky diodes operated at 800 133503, Sep. 2005.

PAGE 95

95 23. F. Yun, S. Chevtchenko, Y. re 073509, Aug. 2005. 24. Craig A. Grimes, Keat G. Ong, Oomman K. Varghese, Xiping Yang, G. Mor, Maggie, Paulose, Elizabeth C. Dickey, Chuanmin Ruan, Michael V. Pishko, James W. Kendig, Andr Sensors, vol. 3, no. 3, pp. 69 82, Mar. 2003. 25. X. Yu, C. Li, Z. N. Low, J. Lin, T. J. Anderson, H. T. Wang, F. Ren, Y. L. Wang, C. Y. Chang, S. J. Pearton, C. H. Hsu, A. Osinsky, A. Dabiran, P. Chow, C. Balaban, J. Painter, "Wireless hydrogen sensor network using AlGaN/GaN high electron mobility transistor differential diode sensors," Sens. Actuat. B, vol. 135, no. 1, pp. 188 194, Dec. 2008. 26. J. C. Lin, "Noninvasive Microwave Measurement of Respir ation," Proc. IEEE, vol. 63, no. 10, pp. 1530, Oct. 1975. 27. K. M. Chen, D. Misra, H. Wang, H. R. Chuang, and E. Postow, "An X band microwave life detection system," IEEE Trans. Biomed. Eng., vol. 33, no. 7, pp. 697 702, Jul. 1986. 28. H. R. Chuang, Y. F. Chen, a nd K. M. Chen, "Automatic clutter canceller for microwave life detection system," IEEE Trans. Instrum. Meas., vol. 40, no. 4, pp. 747 750, Aug. 1991. 29. vol. 13, no. 6, pp. 557 565, Oct. 1992. 30. no. 1, pp. 129 137, Jan. 1996. 31. A. D. Droitcour, O. Boric Lubecke, V. Lubecke S Int. Microw Symp. Dig., vol. 1, pp. 175 178, May 2001. 32. A. D. Droitcour, O. Boric CMOS and BiCMOS single chip direct conversion Doppler radars for remote IEEE Int. Solid State Circuits Conf. Dig., pp. 348 349 Feb. 2002. 33. A. D. Droitcour, O. Boric Lubecke, V. M. Lubecke, J. Lin, and G. T. A. Kovac, efits in single chip silicon Doppler Tech., vol. 52, no. 3, pp. 838 848, Mar 2004.

PAGE 96

96 34. Y. Xiao, J. Lin, O. Boric Band Low Power Doppler Radar System for R Proc. IEEE Ann Engineering in Medicine and Biology Soc. pp. 7151 7154, Sep. 2005. 35. O. Boric Lubecke, V. Lubecke, A. Host Madsen, D. Samardzija, and K. Cheung, Proc. Int Conf. Telecom. In Modern Satellite, Cable and Broadcasting Services, pp. 7 11, Sep 2005. 36. A. J. Gatesman, A. Danylov, T. M. Goyette, J. C. Dickinson, R. H. Giles, W. Goodhue, J. Waldman, W. E. Nixon, and W. Hoen, the SPIE, vol. 6212, May 2006. 37. Y. Xiao, J. Lin, O. Boric for Remote Detection of Heartbeat and Respiration Using Low Power Doubl e Sideband Transmission in Ka no. 5, pp. 2023 2032, May, 2006. 38. Q. Zhou, J. Liu, A. Host Madsen, O. Boric multiple heartbeats using Doppl p p. 1160 1163, May 2006. 39. IEEE Radio and Wireless Symp ., pp. 281 284, Jan. 2007. 40. B. Park, O. Boric offset compe Microw Theory Tech vol. 55, no. 5, pp. 1073 1079, May 2007. 41. J. Lin, C. Li, "Wireless Non Contact Detection of Heartbeat and Respiration Using Low Power Microwav e Radar Sensor," Proc. the 19th Asia Pacific Microw Conf pp. 393 396, Dec. 2007. 42. Cancellation Techniques for Non contact Vital Sign Det S Int. Microw Symp Dig pp. 567 570, Jun 2008. 43. J. H. Oum, D. W. Kim, S. Hong, "Two frequency radar sensor for non contact vital signal monitor ," IEEE MTT S Int. Microw. Symp. Dig., pp. 919 922, Jun. 2008. 44. A. Vergara, N. Petrochilos, O. Boric Lubecke, A. Host Madsen, V. Lubecke, "Blind source separation of hu man body motion using direct conversion D oppler radar," IEEE MTT S Int. Microw. Symp. Dig., pp.1321 1324, Jun. 2008. 45. C. Li, Y. Xiao, J. Lin, "A 5 GHz Double Sideband Radar Sensor Chip in 0.18 CMOS for Non Contact Vital Sign Detection," IEEE Microw. Wire less Comp onents Let t. v ol. 18, no. 7, pp. 494 496, Jul. 2008.

PAGE 97

97 46. A. M. Vergara, O. Boric Lubecke, V. M. Lubecke, "DC information preservation for cardiopulmonary monitor utilizing CW Doppler radar," Proc. IEEE Ann Int. Conf. Engineering in Medicine and Biol ogy Soc. pp.1246 1249, Aug. 2008. 47. D. Zito, D. Pepe, M. Mincica, F. Zito, D. De Rossi, A. Lanata, E. P. Scilingo, A. Tognetti, "Wearable system on a chip UWB radar for contact less cardiopulmonary monitoring: Present status," Proc. IEEE Ann Int. Conf. Eng ineerin g in Medicine and Biology Soc. pp. 5274 5277, Aug. 2008. 48. D. Obeid, G. Issa, S. Sadek, G. Zaharia, G. El Zein, "Low power microwave systems for heartbeat rate detection at 2.4, 5.8, 10 and 16 GHz," Proc. Int Symp. Applied Sciences on Biomed and Co mm Tech pp. 1 5, Oct. 2008. 49. I. Immoreev, T. H. Tao, "UWB radar for patient monitoring," IEEE Aerospace and Electronic Systems Magazine, vol. 23, no. 11, pp. 11 18, Nov. 2008. 50. C. Gu, J. Long, J. Huangfu; S. Qiao, W. Z. Cui, W. Ma, L. Ran, "An instruments built Doppler rad ar for sensing vital signs," Proc. Int Symp. on Antennas, Propagation and EM Theory, pp. 1398 1401, Nov. 2008. 51. C. Li, J. Lin, "Random Body Movement Cancellation in Doppler Radar Vital Sign Detection," IEEE Trans Microw. Theory Tech. v o l. 56, n o. 12, pp. 3143 3152, Dec. 2008. 52. M. S. Rahman, B. J. Jang K. D. Kim, "A new digital signal processor for Doppler radar cardiopulmonary monitoring system," Proc. Int Conf Electrical and Computer Engineering, pp. 76 79, Dec. 2008. 53. X. Yu, C. Li, J. Lin, "System Level Integration of Handheld Wireless Non Contact Vital Sign Detectors," IEEE Radio and Wireless Symp Dig. pp. 514 517, Jan 2009. 54. C. Li J. Cummings, J. Lam, E. Graves, W. Wu, "Radar remote monitoring of vital signs," IEEE Microw Magazine, vol. 10, no. 1, pp. 47 56, Feb 2009. 55. D. T. Petkie, C. Benton, E. Bryan, "Millimeter wave radar for remote measurement of vital signs," Proc. IEEE Radar Conf pp. 1 3, May 2009 56. A. Wiesner, "A multifrequency interferometric CW radar for vital signs detect ion," Proc. IE EE Radar Conf. pp. 1 4, May 2009. 57. C. Li, X. Yu, D. Li, L. Ran, J. Lin, "Software Configurable 5.8 GHz Radar Sensor Symp Dig., pp. 97 100, Jun. 2009. 58. R. Fletcher, J. Han, "Low cost differential front end for Doppler radar vital sign monit oring," IEEE MTT S Int. Microw. Symp. Dig., pp. 1325 1328, Jun. 2009.

PAGE 98

98 59. C. J. Li, F. K. Wang, T. S. Horng, and K. C. Peng, "A novel RF sensing circuit using injection locking and freq uency demodulation for cognitive radio applicat MTT S Int. Microw. Symp. Dig., pp. 1165 1168, Jun. 2009. 60. D. Obeid, S. Sadek, G. Zaharia, G. El Zein, "Advances in microwave systems for medical applications," Proc. Int. Conf. Advances in Computati onal Tools for Engineering Applications, pp. 463 468, Jul. 2009. 61. A. Singh, V. Lubecke, "Respiratory monitoring using a Doppler radar with passive harmonic tags to reduce interference from environmental clutter," Proc. IEEE Ann Int. Conf. Engineering in Me dicine and Biology Soc., pp. 3837 3840, Sep 2009 62. J. E. Kiriazi, O. Boric Lubecke, V. M. Lubecke, "Radar cross section of human cardiopulmonary activity for recumbent subject," Proc. IEEE Ann Int. Conf. Engineering in Me dicine and Biology Soc. pp. 4808 4 811, Sep 2009. 63. Y. Yan, C. Li, X. Yu, M. D. Weiss, J. Lin, "Verification of a Non Contact Vital Sign Monitoring System Using an Infant Simulator," Proc. IEEE Ann Int. Conf. Engineering in Medicine and Biolog y Soc., pp. 4836 4839, Sep 2009. 64. M. Loschonsky, C. Feige, O. Rogall, S. Fisun, L. M. Reindl, "Detection technology for trapped and buried people," IEEE MTT S Int Microw Workshop on Wireless Sensing, Local Pos itioning, and RFID, pp. 1 6, Sep 2009. 65. A. Wiesner, "CW radar based vital signs detection," E uropean Microwave Conf pp. 1595 1598, Sep 2009. 66. R. Ichapurapu, S. Jain, M. U. Kakade, D. Y. C. Lie, R. E. Banister, "A 2.4GHz non contact biosensor system for continuous vital signs monitoring on a single PCB," Proc. IEEE Int Conf ASIC, pp. 925 928, O ct. 2009. 67. W. Massagram, V. M. Lubecke, A. Host Madsen, O. Boric Lubecke, "Assessment of Heart Rate Variability and Respiratory Sinus Arrhythmia via Doppler Radar," IEEE Trans on Microw Theory Tech vol. 57, no. 10, pp. 2542 2549, Oct. 2009. 68. D. Obeid, S. Sadek, G. Zaharia, G. El Zein, "Touch less heartbeat detection and cardiopulmonary modeling," Proc. Int Symp Applied Sciences in Biomedical and Comm Tech pp. 1 5, Nov. 2009. 69. B. Gupta, E. Cianca, M. Ruggieri, R. Prasad, "A novel FMUWB system for vital sign monitoring and its comparison with IR UWB," Proc. Int Symp Applied Sciences in Biomedical and Comm Tech pp. 1 4, Nov. 2009. 70. J. S. Kim; M. S. Rahman, B. Y. Kim B. J. Jang K. D. Kim, "DSP embedded hardware for non contact bio radar heart and respi ration rate monitoring system," Proc. Int Conf Future Bio m ed Information Engineering, pp. 560 563, Dec. 2009.

PAGE 99

99 71. L. Chioukh, H. Boutayeb, L. Lin L. Yahia, K. Wu, "Integrated radar systems for precision monitoring of heartbeat and respiratory status," Proc. Asia Pacific Microw Conf pp. 405 408, Dec. 2009. 72. R. R. Fletcher, S. Kulkarni, "Wearable Doppler radar with integrated antenna for patient vital sign monitoring," IEEE Radio and Wireless Symp Dig., pp. 276 279, Jan. 2010. 73. C. Li, J. Ling, J. Li, J. Lin, "Accurate Doppler Radar Non Contact Vital Sign Detection Using the RELAX Algorithm," IEEE Trans Instrum Meas vol. 59, no. 3, pp. 687 695, Mar 2010. 74. X. Yu, C. Li, J. Lin, "Noise Analysis for Noncontact Vital Sign Detectors," IEEE Ann Wireless Microw Tech Conf pp. 1 4, Apr. 2010. 75. G. Lu, F. Yang, X. Jing, J. Wang, "Contact free measurement of heartbeat signal via a doppler radar using adaptive filtering," Proc. Int Conf Image Analysis and Signal Processing pp. 89 92, Apr. 2010. 76. C. Li, X. Yu, C. M Lee, D. Li, L. Ran, J. Lin, "High Sensitivity Software Configurable contact Vital Sign Detection," IEEE Trans Microw Theory Tech RFIC 2009 Special Issue, vol. 58, no. 5, pp. 1410 1419, May 20 10. 77. T. Y. Chin, K. Y. Lin, S. F. Chang and C. C. Chang, "A Fast Clutter Cancellation Method in Quadrature Doppler Radar for Noncontact Vital Si gnal Detection," IEEE MTT S Int. Microw. Symp. Dig. pp. 764 767, May 2010. 78. F. K. Wang, C. J. Li, C. H. Hsiao, T. S. Horng, J. Lin, K. C. Peng, J. K. Jau, J. Y. Li, and C. C. Chen, "An Injection Locked Detector for Concurrent Spectrum and Vital Sign Se nsing," IEEE MTT S Int. Microw. Symp. Dig. pp. 768 771, May 2010. 79. A. Singh, V. Lubecke, "A Heterodyne Receiver for H armonic Doppler Radar Cardiopulmonary Monitoring with Body worn Passive RF Tags," IEEE MTT S Int Microw Symp Dig pp. 1600 1603, May 2010. 80. C. Gu, C. Li, J. Lin, J. Huangfu, L. Ran, "Instrument Based Non Contact Doppler Radar Vital Sign Detection System Using Heterodyne Digital Quadrature Demodulation Architecture," IEEE Trans Instrum Meas vol 59, no. 6, pp. 1580 1588, Jun. 2010. 81. Xiaogang Yu; Changzhi Li; Jenshan Lin, "Two dimensional Noncontact Vital Sign Detection Using Doppler Radar Array Approac h", IEEE MTT S Int. Microw. Symp. Dig., pp. 1 4 Jun. 2011

PAGE 100

100 82. tral Analysis of a Low Power Ka Trans. Microw Theory Tech IMS 2006 Speci al Issue, vol. 54, no. 12, pp. 4464 4470, Dec 2006. 83. Cambridge University Press, 2003. 84. s. Antennas Propagat., vol. 30, no. 6, pp. 1191 1196, Nov. 1982. 85. 535, Jul. 1978. 86. IEEE International Committ ee on Electromagnetic Safety 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. 2005

PAGE 101

101 BIOGRAPHICAL SKETCH Xiaogang Yu received the B.S. degree in physics from Nanjing University, Nanjing, China, in 2004, the M.S. degree in electrical and computer engineering from the University of Florida, Gainesville, in 2007, and the Ph.D. degree in electrical and computer e ngineering at the University of Florida in 2011. His research interests include wireless sensors, biomedical applications of microwave/RF systems, and microwave/millimeter wave circuits. Mr. Yu is a student member of the IEEE Microwave Theory and Techniq ues Society (IEEE MTT S). He was a finalist in the 2009 IEEE Radio and Wireless Symposium Student Paper Competition.