Table of Contents

Cover

Series Title

Title Page

List of Contributors

Chapter 1: Introduction

1.1 Current Methods of Physiological Monitoring
1.2 Need for Noncontact Physiological Monitoring
1.3 Doppler Radar Potential for Physiological Monitoring
References

Chapter 2: Radar Principles

2.1 Brief History of Radar
2.2 Radar Principle of Operation
2.3 Doppler Radar
2.4 Monostatic and Bistatic Radar
2.5 Radar Applications
References

Chapter 3: Physiological Motion and Measurement

3.1 Respiratory System Motion
3.2 Heart System Motion
3.3 Circulatory System Motion
3.4 Interaction of Respiratory, Heart, and Circulatory Motion at the Skin Surface
3.5 Measurement of Heart and Respiratory Surface Motion
References

Chapter 4: Physiological Doppler Radar Overview

4.1 RF Front End
4.2 Baseband Module
4.3 Signal Processing
4.4 Noise Sources
4.5 Conclusions
References

Chapter 5: CW Homodyne Transceiver Challenges

5.1 RF Front End
5.2 Baseband Module
5.3 Signal Demodulation
References

Chapter 6: Sources of Noise and Signal-to-Noise Ratio

6.1 Signal Power, Radar Equation, and Radar Cross Section
6.2 Oscillator Phase Noise, Range Correlation and Residual Phase Noise
6.3 Contributions of Various Noise Sources
6.4 Signal-to-Noise Ratio
6.5 Validation of Range Correlation
6.6 Human Testing Validation
References

Chapter 7: Doppler Radar Physiological Assessments

7.1 Actigraphy
7.2 Respiratory Rate
7.3 Tidal Volume
7.4 Heart Rates
7.5 Heart Rate Variability
7.6 Respiratory Sinus Arrhythmia
7.7 RCs and Subject Orientation
References

Chapter 8: Advanced Performance Architectures

8.1 DC Offset and Spectrum Folding
8.2 Motion Interference Suppression
8.3 Range Detection
References

Chapter 9: Applications and Future Research

9.1 Commercial Development
9.2 Recent Research Areas
9.3 Conclusion
References

Index

End User License Agreement

List of Illustrations

Chapter 2: RADAR PRINCIPLES

Figure 2.1 Basic principle of radar. A target will reflect an echo signal; the echo signal's power, phase delay, and frequency depends on the target's distance, radar cross section, and velocity.
Figure 2.2 Propagation direction of electromagnetic wave.
Figure 2.3 Incident wave transmission and reflection at a planar boundary between two materials.
Figure 2.4 Geometry of a radar and a target in deriving the Doppler frequency shift.
Figure 2.5 CW Doppler radar block diagram.
Figure 2.6 (Top) Frequency–time relation in an FMCW radar with linear triangular frequency modulation. Solid lines represent the transmitted signal, dashed lines represent the received signal delayed by a time $c02-math-0074$ ; $c02-math-0075$ excursion, $c02-math-0076$ frequency. (Bottom) Absolute value of the frequency difference between the transmitted and received signals.
Figure 2.7 Block diagram of FMCW radar with homodyne receiver [Komemou, 2009].
Figure 2.8 Block diagram of pulse Doppler radar.
Figure 2.9 (a) Baseband demodulated signal when the Doppler frequency $c02-math-0087$ ; (b) baseband signal for the Doppler frequency $c02-math-0088$ .
Figure 2.10 (a) Monostatic and (b) bistatic radar configuration.

Chapter 3: PHYSIOLOGICAL MOTION AND MEASUREMENT

Figure 3.1 The thoracic wall, body cavities, and muscles of respiration.
Figure 3.2 Motion of upper, lower, and lowest ribs.
Figure 3.3 The location of the heart in the rib cage. The intercostal spaces are indicated by the numbers 1–5. The heart is beneath the sternum and the cartilage of the third, fourth, and fifth ribs. After Flint [1859].
Figure 3.4 Diagrammatic section of the heart. The arrows indicate the direction of blood flow.
Figure 3.5 Example of an electrocardiogram. Atrial depolarization causes the P wave, ventricular depolarization causes the QRS complex, and ventricular repolarization causes the T wave.
Figure 3.6 Motion of the heart throughout the cardiac cycle.
Figure 3.7 During the beginning of systole, the ventricles are contracting, but all the valves in the heart are closed; this is known as the isovolumetric ventricular contraction (1). The pressure in the ventricle increases, and when it is greater than the pressure in the aorta, the aortic valve opens, and ventricular ejection (2) begins. The pressure in the ventricle decreases and blood flows out of it, and when the pressure drops below that of the aortic valve, the aortic valve closes and diastole begins. Since all the valves in the heart are closed and the ventricle is relaxing, this is known as the isovolumetric ventricular relaxation period (3). When the left ventricular pressure drops below that of the atria, the mitral valve opens, and ventricular filling (4) begins.
Figure 3.8 Model of the arterial system, showing major arteries.
Figure 3.9 Diagram of arterial pressure in systole and diastole. During systole, the artery distends, storing blood; during diastole, the artery contracts so that blood continues flowing into the arterioles after the aortic valve is closed.
Figure 3.10 Lee's model of an artery in tissue for analyzing surface motion with radial motion of the vessel wall.

Chapter 4: PHYSIOLOGICAL DOPPLER RADAR OVERVIEW

Figure 4.1 Simplified block diagram of a physiological Doppler radar system.
Figure 4.2 A simple Doppler transceiver architecture denoting transmit and receive antennas, signal source, and a frequency mixer.
Figure 4.3 Illustration of null and optimum points in a single-channel receiver system.
Figure 4.4 Spectra of real cosine signal and an imaginary sine signal when represented in complex notation.
Figure 4.5 For the real signal, $c04-math-0037$ , to be multiplied by a complex exponential with only a negative frequency component, $c04-math-0038$ , the signal must be split and mixed with local oscillator signals to determine the in-phase component, $c04-math-0039$ , and the quadrature component, $c04-math-0040$ . The LO signal on the Q channel is delayed by 90° before mixing. The two components can be summed to create the output: $c04-math-0041$ .
Figure 4.6 Self-image problem with a direct-conversion receiver. If a quadrature receiver is not used, both the positive and negative frequency components are down-converted to baseband, where they can interfere with each other.
Figure 4.7 Avoiding the self-image problem with a quadrature direct-conversion. When the RF signal is mixed with a complex exponential, only the positive or negative band is converted into baseband, avoiding the interference problem.
Figure 4.8 A quadrature receiver system (a) and a sample of received I and Q signals (b).
Figure 4.9 Simulated plots showing the effect of amplitude imbalance between the I and Q channels (a), and the effect of phase imbalance between the I and Q channels (b).
Figure 4.10 Simulated IQ plot of a motion as seen by two different frequencies. The arc length is about four times longer at 10 GHz compared with $c04-math-0060$ .
Figure 4.11 Power budget of 2.4 GHz radio wave at each internal layer in the body. Most of reflected signal comes from air/skin interface, which is about 51% of the incident power.
Figure 4.12 Baseband signal amplifiers, low-pass filtering, and data acquisition for the radar system.
Figure 4.13 Illustration of I and Q signals in the complex plane with DC and AC coupling. The dotted circle is a circle fitted on the actual data (arc). DC coupling (a) and exaggerated effects of AC coupling (b) is shown.
Figure 4.14 Illustration of the quantization noise in the data acquisition system.
Figure 4.15 Plots illustrating the principle for linear demodulation.
Figure 4.16 Plots illustrating the principle for nonlinear (arctangent) demodulation technique.
Figure 4.17 Peak detection for a sample ECG signal. $c04-math-0084$ is the peak to peak distance that can be used to estimate the heart rate.
Figure 4.18 Illustration of the short-time FT frequency analysis: (1) a short-time moving window FFT slides through the time domain data, (2) spectrum of the window of the time domain data is calculated, (3) peak of interest in the frequency spectrum is mapped to a time plot, which shows frequency versus time analysis.
Figure 4.19 Equivalent circuit for thermal noise in conductors.
Figure 4.20 Flicker noise versus thermal noise and the concept of corner frequency.
Figure 4.21 Simulated baseband plots showing the effect of distributed random noise on data. Arc “A” shows the IQ plot with very little noise while arc “B” shows the IQ plot with a lot of noise.

Chapter 5: CW HOMODYNE TRANSCEIVER CHALLENGES

Figure 5.1 Physiological radar system block diagram.
Figure 5.2 Block diagram of Doppler radar transceiver.
Figure 5.3 Photograph of Doppler radar transceiver board.
Figure 5.4 Measurement setup.
Figure 5.5 Measured respiration signals at optimum (dashed lines) and null (solid lines) positions. When at the null point and displacement of the target is much smaller than $c05-math-0069$ (a), sensitivity decreases significantly making accurate rate measurements difficult. Even with exaggerated deep-breathing displacement (b), error still occurs with the frequency of the output signal double that of actual motion as measured at the optimum position.
Figure 5.6 Measurement history data for both respiration and heart rate with quadrature receivers, at either optimum (a) or null points (b, c). At the optimum point (a), the Doppler-measured heart rate corresponds closely to the reference for all $c05-math-0070$ . At the null point during continuous breathing (b), the Doppler measured heart rate and reference differ by the respiration reference frequency ( $c05-math-0071$ ), while with breath-holding (c) it jumps between either double (case II) or equal to the actual frequency (case III). The “(ref)” signal in (b) is that measured from the quadrature (not null) channel.
Figure 5.7 Block diagrams of direct conversion systems. The LO leakage from the RF port of the mixer in the receiver configuration, (a), is separated into two components. One is external LO leakage, which affects other receivers, and the other is an LO self-mixing signal which induces DC offset. In a transceiver configuration (b), there is Tx leakage to the receiver chain and a self-mixing component, in addition to receiver leakage problems.
Figure 5.8 Block diagram of DC offset canceller.
Figure 5.9 Measured DC offset with and without compensation. The DC offset could be set to zero using the LO leakage cancellation technique.
Figure 5.10 Measured receiver Flicker noise reduction.
Figure 5.11 Measured DC offset and LO leakage power variation.
Figure 5.12 Measured DC offset and Flicker noise levels at 3 Hz.
Figure 5.13 Measured DC offset.
Figure 5.14 Measured Flicker noise.
Figure 5.15 Measurement setup (a) and measured control voltage and imbalance factors (b). Using phase shifters, I and Q imbalance factors for a homodyne radar system can be measured without circuit board modification. An object moving with constant velocity is simulated by using a sawtooth wave to linearly sweep a set of phase shifters through 360° (3.1 V). The resultant I and Q baseband output signals are sinusoidal, with a single frequency that corresponds to the velocity simulated by the slope of the sawtooth wave. Amplitude and phase imbalance factor were measured here as 4.7° and 18.5°, respectively.
Figure 5.16 (a) DC cancellation using a sample-and-hold and (b) using an ADC–DAC pair as an infinite S/H in the two-stage system.
Figure 5.17 Block diagram of two-stage system utilizing digital feedback system.
Figure 5.18 Block diagram of DC offset compensation system.
Figure 5.19 Time plots of a subject 1 m away with arm movement. (a) I plot of the magnified raw signal, (b) I plot of the AC coupling response, (c) I and Q plot of the DC cancelled output.
Figure 5.20 Measurement setup for DC compensation. Overall radar setup is shown (a), with data acquisition (dashed region in (a)) details provided for the I channel (the Q channel is exactly the same) (b). The clutter- and circuit-based DC offset measured with no target present is reproduced (DC supply) and subtracted from the response for a human subject, so that the heart motion signal (which includes a DC component) can be digitized with maximum resolution.
Figure 5.21 Polar plot of I/Q data. The I/Q data with DC preserved forms a portion of a circle centered at the origin, verifying preservation of all phase information, while the I/Q signals without DC information form a line near the center for which phase information cannot be accurately recovered.
Figure 5.22 Heart rate measurements for both channels in an intermediate position. Band-pass-filtered (0.9–2 Hz) Doppler radar I and Q signals are shown along with the combined arctangent demodulated output (AT), and a wired finger pulse reference (a). Heart rate history (using autocorrelation) is also shown (b), where the Q channel data are at times off by the respiration rate value, as predicted. Standard deviation is less than 1 beat over the full 40-s interval for the AT data, while it is 3.9 and 9.8 beats for the I and Q channels, respectively.
Figure 5.24 I, Q, and arctangent demodulated signals (a) measured for the I channel close to a null position. Data dropout regions occur for both I (23% of the interval) and Q (5%) channels. Standard deviation is 7.5 or 1.7 beats for the I and Q channels, respectively, and only 0.6 for the arctangent output.
Figure 5.23 I, Q, and arctangent (AT) demodulated signals (a) measured for a position where the Q channel is close to a null condition. The Q channel rate (b) shows drop-out regions (in 35% of the interval) when the SNR is insufficient for digitization, as occurs with the squaring effect when in the null position. Excluding drop-outs, the I and Q channels have errors of 4.8 or 5.2 beats, respectively, over the same 40-s interval where the AT data has an error of only 0.9 beats.
Figure 5.25 Block diagram of a quadrature direct conversion Doppler radar system in a measurement setup for heart rate extraction (a), with data acquisition (dashed region in (a)) details provided for the I channel (the Q channel is exactly the same) (b). Two stages of preamplifiers are used to obtain high power of heart signal without losing DC information. First preamplifier is for obtaining DC signal as well as antialiasing filtering, then second amplifier is AC coupled, thus only the chest motion signal is amplified and digitized with maximum ADC resolution. © 2007 IEEE, Reprinted, with permission, from Park et al. [2007a].
Figure 5.26 I, Q, and arctangent-demodulated signals measured for a position where the Q channel is close to a null condition. Arc formed by respiration motion of the chest tracked back to the origin in order to eliminate DC offset (a). Digitally band-pass-filtered data extracts heart signal from a raw data (b). The Q channel rate data (c) shows dropout regions when the SNR is insufficient for digitization, excluding dropouts, the I and Q channel data have an error of 1.7 or 5.1 beats, respectively, over the same 60-s interval where the arctangent output has an error of only 1.3 beats. © 2007 IEEE, Reprinted, with permission, from Park et al. [2007a].
Figure 5.27 Measurement result of a target's movement. Since the movement deviation of 200 cm is much bigger than wavelength, baseband I and Q outputs are frequency modulated according to speed of the target as well as amplitude modulated due to the receiving signal power variation (a). Complex plot of I and Q outputs forms complete circle with different radius but same center point, and center offset is brought back to the origin in order to remove DC offset (b). Arctangent demodulation output can restore actual movement of a target by simply unwrapping output to compensate $c05-math-0156$ singular effect (c).
Figure 5.28 Block diagram of adaptive DC offset compensation. DC cancellation includes DAC and voltage divider to scale voltage to ±50 mV range. DSP includes center find function for DC information preservation.
Figure 5.29 Diagram of test setup measuring respiration and pulse using quadrature direct conversion Doppler radar, airflow rate, and finger pressure transducers.
Figure 5.30 Time (a) and power spectral density (b) of Doppler radar off a human subject.
Figure 5.31 Extracted heart rate from AC-coupled and DC offset compensation signals with comparison to a wired finger pulse. Both utilized arctangent demodulation.

Chapter 6: SOURCES OF NOISE AND SIGNAL-TO-NOISE RATIO

Figure 6.1 Illustration of power, effective radiated power, and power density at various points in the Doppler radar system. $c06-math-0008$ is the transmitted power, G is the antenna gain, R is the distance between the target and the antenna, $c06-math-0009$ is the attenuation, $c06-math-0010$ is the RCS, and $c06-math-0011$ is the wavelength of the RF signal. These equations assume that the target and antenna are sized such that they are in the far-field at the range of measurement.
Figure 6.2 RF sideband spectrum, including phase noise and spurious noise. The phase noise spectrum is symmetrical about the oscillation frequency, indicating that phase noise, and not amplitude noise, is dominant in this oscillator. The peaks in the spectrum are spurious noise, indicating modulation by other signals.
Figure 6.3 Exaggerated depiction of phase noise in the time domain. The solid line is the perfect sinusoid in Equation 6.24 and the dotted line is the sinusoid with phase noise in Equation 6.26.
Figure 6.4 Measurement of single-sideband phase noise, $c06-math-0048$ .
Figure 6.5 Example phase noise spectrum: a typical phase noise spectrum will have a $c06-math-0066$ dependence close to the carrier, a $c06-math-0067$ dependence beyond that, and be flat farther from the carrier.
Figure 6.6 Illustration of the range correlation phase noise filtering effect. Since the transmitted signal is derived from the same source as the received signal, the phase noise on the LO, $c06-math-0070$ , and the RF input, $c06-math-0071$ , are correlated. When the two signals are mixed, most of the phase noise at baseband is effectively cancelled, leaving only the residual phase noise, $c06-math-0072$ .
Figure 6.7 Setup for the range-correlation verification experiment. The baseband noise spectrum was measured with the VSA. Cables of various lengths were connected in the place of the cable marked t to change the time delay between the RF and LO signals. © 2004 IEEE, Reprinted, with permission, from Droitcour et al. [2004].
Figure 6.8 (a) Measured RF phase noise with −30 dB/dec fit line used to predict baseband noise and (b) measured and predicted baseband residual phase noise with time delays of 28.0, 12.6, and 6.2 ns. © 2004 IEEE, Reprinted, with permission, from Droitcour et al. [2004].
Figure 6.9 Predicted signal-to-noise ratio for (a) heart and (b) respiration with each noise source and all noise sources, using the parameters in Table 6.1. © 2009 IEEE, Reprinted, with permission, from Droitcour et al. [2009].
Figure 6.10 Data from subject 4062 at 0.5 m. (a) The top trace is the combined heart signal from the radar; the second trace is the combined respiration signal from the radar, the third trace is the ECG, and the fourth trace is the combined respiration signal from the straps. (b) Heart and respiratory rates calculated from the Doppler radar and the reference. © 2009 IEEE, Reprinted, with permission, from Droitcour et al. [2009].
Figure 6.11 Data from subject 4062 at 1.5 m. (a) The top trace is the combined heart signal from the radar; the second trace is the combined respiration signal from the radar, the third trace is the ECG, and the fourth trace is the combined respiration signal from the straps. (b) Heart and respiratory rates calculated from the Doppler radar and the reference. © 2009 IEEE, Reprinted, with permission, from Droitcour et al. [2009].
Figure 6.12 Measured and theoretical SNR for (a) heartbeat and (b) respiratory rate. The theoretical radar-cross-section–mean-squared ratio product was set at a value of $c06-math-0184$ for respiratory motion and $c06-math-0185$ for heartbeat to provide the best possible fit. © 2009 IEEE, Reprinted, with permission, from Droitcour et al. [2009].
Figure 6.13 Scattergram of error versus signal-to-noise ratio for heart and respiratory rate measurement with the Doppler radar. The error is defined as the standard deviation of the difference between the radar-based measurement and the reference, and the signal-to-noise ratio is measured as described in this chapter. A linear regression is performed on the data; the model for the heart rate is $c06-math-0186$ with $c06-math-0187$ of 0.59. The model for respiration is $c06-math-0188$ with $c06-math-0189$ of 0.42. © 2009 IEEE, Reprinted, with permission, from Droitcour et al. [2009].

Chapter 7: DOPPLER RADAR PHYSIOLOGICAL ASSESSMENTS

Figure 7.1 A block diagram of test protocol setup. The reference signals, ECG, finger pulse, upper chest belt, lower chest belt, and spirometer, were completely isolated from the radar system. The sync signal provided the marker for when the BIOPAC reference system starts to line up the data from the two systems.
Figure 7.2 An example of 30-min data from the Doppler radar output after linear demodulation, taken for a seated subject. Gray cross markers indicate detected peaks, and dotted gray line shows detected periods of motion. © 2011 IEEE. Reprinted, with permission, from Massagram et al. [2011].
Figure 7.3 The behavior of human subjects varied from (a) remained still at the beginning of the measurement then started fidgeting, (b) started fidgeting at first then remained still in the rest of the measurement, and (c) remained still at first, started to fidget, and then back to still during the measurement. © 2011 IEEE. Reprinted, with permission, from Massagram et al. [2011].
Figure 7.4 The linear regression of the respiratory rate provided by the Doppler radar system and that provided by the Embla system. © 2009 IEEE. Reprinted, with permission, from Droitcour et al. [2009].
Figure 7.5 Bland–Altman plot: the difference versus the mean of measurement of respiratory rates provided by the Doppler radar and by the Embla system. © 2009 IEEE. Reprinted, with permission, from Droitcour et al. [2009].
Figure 7.6 Example of DC-corrected operation. The signal, reconstruct 1, represents the output from the first integration, and the signal, reconstruct 2, represents the output from the second integration. © 2009 IEEE. Reprinted, with permission, from Massagram et al. [2009].
Figure 7.7 Relative volume displacement of the radar, upper chest belt, lower chest belt, and spirometer respiration signals. © 2009 IEEE. Reprinted, with permission, from Massagram et al. [2009].
Figure 7.8 Instantaneous respiration rates for all signals. © 2009 IEEE. Reprinted, with permission, from Massagram et al. [2009].
Figure 7.9 Tidal volumes of the radar, upper chest belt, lower chest belt, and spirometer respiration signal. © 2009 IEEE. Reprinted, with permission, from Massagram et al. [2009].
Figure 7.10 Statistical analysis of the tidal volume: (a) correlation plot and (b) Bland–Altman analysis. © 2009 IEEE. Reprinted, with permission, from Massagram et al. [2009].
Figure 7.11 Heart rates from 5-min window measurements in (a) seated and (b) supine positions of subject 2205. The radar output from supine position shows a better accuracy than from the seated position. © 2009 IEEE. Reprinted, with permission, from Massagram et al. [2009].
Figure 7.12 Bland–Altman plots from 5-min window measurements in (a) seated and (b) supine positions of subject 2205. All data points in supine measurement were with 95% confidence intervals unlike for seated measurement. The Bland–Altman bias magnitude in seated position is much greater than in supine position (seated −0.733 mL vs supine −0.037 mL). The BA plot in (a) displays a cloud shape, a sign of bad correlation, whereas (b) is almost a line.
Figure 7.13 Breathing signals (a) subject 2303 with irregular period and nonsinusoidal pattern and (b) subject 2205 with regular period and sinusoidal pattern.
Figure 7.14 PSD of the breathing signals (a) subject 2303 with wilder spectrum spread and (b) subject 2205 well-defined peak at 0.217 Hz and narrower spectrum spread.
Figure 7.15 RR interval histogram plots of subject 2205 in (a, b) seated positions and (c, d) supine positions, which have wider spread of distribution in RR intervals. © 2009 IEEE. Reprinted, with permission, from Massagram et al. [2009].
Figure 7.16 Two-axis plots representing the respiratory signal and the heart period for (a) seated position and (b) supine position, for subject 2205. © 2009 IEEE. Reprinted, with permission, from Massagram et al. [2009].
Figure 7.17 RSA peak–valley amplitude estimation for (a) seated position and (b) supine position, for subject 2205. © 2009 IEEE. Reprinted, with permission, from Massagram et al. [2009].
Figure 7.18 Radar cross section of metallic sphere of radius $c07-math-0119$ as a function of wavelength.
Figure 7.19 Two types of waves scattering off a metallic sphere.
Figure 7.20 A metallic half-cylinder.
Figure 7.21 Body shape in three sleeping positions: (a) supine, (b) prone, and (c) side.
Figure 7.22 Center-tracked arcs for the subject in the supine, prone, and side positions at 2-m range with (a) 2.4 GHz and (b) 5.8 GHz carriers. © 2011 IEEE. Reprinted, with permission, from Kiriazi et al. [2012].

Chapter 8: ADVANCED PERFORMANCE ARCHITECTURES

Figure 8.1 Measurement setup showing the single-channel mixer together with the DC cancellation path utilizing RF.
Figure 8.2 Measurement setup showing the antennas, phase shifters, mixers, and splitters.
Figure 8.3 Artificial target used in the experiment. The arrow depicts the trajectory of the motion of the target.
Figure 8.4 DC voltage output of the single-channel mixer, while phase 1 and phase 2 have been swept over a 180° phase shift. © 2011 IEEE. Reprinted, with permission, from Mostafanezhad and Boric-Lubecke [2011].
Figure 8.5 Baseband signal strength from the single-channel mixer, while phase 1 and phase 2 have been swept over a 180° phase shift. © 2011 IEEE. Reprinted, with permission, from Mostafanezhad and Boric-Lubecke [2011].
Figure 8.6 Respiration signal from radar (top), respiration reference, heart signal filtered from the radar output (middle), and a finger pressure pulse (bottom) recorded simultaneously as a reference heart signal. © 2011 IEEE. Reprinted, with permission, from Mostafanezhad and Boric-Lubecke [2011].
Figure 8.7 Estimated heart rate from the traditional radar IQ channels, reference finger pulse outputs, and the proposed single-channel radar, and the absolute error of heart rate. © 2011 IEEE. Reprinted, with permission, from Mostafanezhad and Boric-Lubecke [2011].
Figure 8.8 Block diagram of the Ka-band remote monitoring system. © 2006 IEEE. Reprinted, with permission, from Xiao et al. [2006].
Figure 8.9 The output spectrum of the transmitter, measured at the antenna connector. The resolution bandwidth and the video bandwidth were both set at 3 MHz. © 2006 IEEE. Reprinted, with permission, from Xiao et al. [2006].
Figure 8.10 Heartbeat detection at null point (a) and optimum point (b). The heart-rate accuracy is 54.5% at the null point while 94% at the optimum point. The frequency difference between them is only 56 MHz. © 2006 IEEE. Reprinted, with permission, from Xiao et al. [2006].
Figure 8.11 Simple diagram of a low-IF receiver (a) and a coherent low-IF receiver (b). © 2010 IEEE, Reprinted, with permission, from Mostafanezhad et al. [2010].
Figure 8.12 Signal spectrum in low IF receiver. © 2010 IEEE, Reprinted, with permission, from Mostafanezhad et al. [2010].
Figure 8.13 Measurement setup. Note the coherent low-IF generation path. © 2010 IEEE, Reprinted, with permission, from Mostafanezhad et al. [2010].
Figure 8.14 Baseband I and Q signals from direct conversion and coherent low-IF receiver paths. © 2010 IEEE, Reprinted, with permission, from Mostafanezhad et al. [2010].
Figure 8.15 Baseband signal spectrum for the two receiver paths. © 2010 IEEE, Reprinted, with permission, from Mostafanezhad et al. [2010].
Figure 8.16 Location of antenna and the subject. © 2008 IEEE, Reprinted, with permission, from Mostafanezhad et al. [2008].
Figure 8.17 RF configuration of the radar.
Figure 8.18 Radar channel outputs (top) and the demodulated signal (bottom). © 2008 IEEE, Reprinted, with permission, from Mostafanezhad et al. [2008].
Figure 8.19 Demodulated motion (top) and antenna's recorded mechanical motion in x and y direction (bottom). © 2008 IEEE, Reprinted, with permission, from Mostafanezhad et al. [2008].
Figure 8.20 Radar signal and motion-cancelled signal, zoomed in the area of antenna motion. © 2008 IEEE, Reprinted, with permission, from Mostafanezhad et al. [2008].
Figure 8.21 Time-variant heart rate calculated from the radar, reference, and motion-cancelled signals. The motion-cancelled radar signal is in good agreement with the reference finger pulse signal, while the radar signal before motion cancellation cannot be used to retrieve the heart rate. © 2008 IEEE, Reprinted, with permission, from Mostafanezhad et al. [2008].
Figure 8.22 A monostatic radar (a), and a sensor node configuration (b). © 2007 IEEE, Reprinted, with permission, from Mostafanezhad et al. [2007].
Figure 8.23 Simulated outputs of the sensor node and monostatic radar for subject displacement of 5 mm. The monostatic antenna begins shaking with the amplitude of 10 mm after 10 s. While sensor node output (top) remains unchanged, antenna's shake greatly alters monostatic radar output. The bottom trace is the reference displacement. © 2007 IEEE, Reprinted, with permission, from Mostafanezhad et al. [2007].
Figure 8.24 Simulation: human subject fixed in a location next to sensor node, while Tx antenna (circle) is moved to various positions to simulate how effective sensor node will be. The transmit antenna is shaking in the x direction.
Figure 8.25 Received signal pattern for various locations of the Tx antenna. I, Q, reference, and node signals are depicted in this plot.
Figure 8.26 Block diagrams of the monostatic transceiver (a) and sensor node (b). © 2007 IEEE, Reprinted, with permission, from Mostafanezhad et al. [2007].
Figure 8.27 The sensor node used in measurements.
Figure 8.28 Measured signals from the radar node, monostatic radar, chest band reference, and x–y displacement. The monostatic antenna begins shaking after 31 s. © 2007 IEEE, Reprinted, with permission, from Mostafanezhad et al. [2007].
Figure 8.29 Detected respiration rate from the monostatic radar (dotted line), sensor node (dashed line), and reference (solid line) outputs. It can be seen that the monostatic output, once the antenna is physically shaking, loses track of the reference signal while the sensor node output remains in good agreement with the reference. © 2007 IEEE, Reprinted, with permission, from Mostafanezhad et al. [2007].
Figure 8.30 Proposed RF tagging for Doppler radar respiratory monitoring using harmonic tags. © 2012 IEEE. Reprinted, with permission, from Singh and Lubecke [2012].
Figure 8.31 RF tag used as an on-body sensor (not to scale). © 2011 IEEE. Reprinted, with permission, from Singh and Lubecke [2011].
Figure 8.32 (a) Fourier transform of data from Experiment I(a) and I(b) (tag and mechanical target motion, respectively) and (b) their I–Q plots showing the relative amplitudes of the motion. © 2012 IEEE. Reprinted, with permission, from Singh and Lubecke [2012].
Figure 8.33 (a) FFT data for Experiment I(c) (tag and mechanical target moving together) showing an increase in detected target motion compared with case I and (b) the I–Q plot showing the presence of two frequency components and the phase relation between the two. © 2012 IEEE. Reprinted, with permission, from Singh and Lubecke [2012].
Figure 8.34 Schematic of the dual-channel 2.45-GHz Doppler radar. © 2012 IEEE. Reprinted, with permission, from Singh and Lubecke [2012].
Figure 8.35 Response of 2.45-GHz CW Doppler radar to two simultaneous moving objects in its view. (a) Raw data showing the amplitude changes due to EM interaction between the two targets and (b) the rate indicating the inability of the radar to clearly isolate the motion of any of the two moving objects. © 2012 IEEE. Reprinted, with permission, from Singh and Lubecke [2012].
Figure 8.36 Respiration rate of a human subject with an untagged moving scattering object in front of harmonic radar. © 2012 IEEE. Reprinted, with permission, from Singh and Lubecke [2012].
Figure 8.37 The response of 2.45 and 4.9 GHz radar to (a) a tagged human subject in front of radar when the target is not moving (Experiment III(a)), (b) tagged human with the target moving at 0.2 Hz (Experiment III(b)), and (c) the error rate in the detected respiration rate for the two radars. As expected, both the radar can track respiration accurately when mechanical target is stationary (53 (a)) but 2.45 GHz radar cannot track the respiration accurately in (53 (b)) when the mechanical target starts moving. © 2012 IEEE. Reprinted, with permission, from Singh and Lubecke [2012].
Figure 8.38 Response of 2.45 and 4.9 GHz radar to a tagged human and a mechanical untagged object. The mechanical target is moving 2 cm at a frequency of approximately 0.15 Hz. This condition represents the worst-case scenario where 2.45-GHz radar would completely detect the undesired motion. © 2012 IEEE. Reprinted, with permission, from Singh and Lubecke [2012].
Figure 8.39 Spectrums of UWB and narrow-band systems.
Figure 8.40 UWB radar block diagram.
Figure 8.41 UWB quadrature Doppler radar block diagram.
Figure 8.42 Detailed block diagram of the reconfigurable UWB pulse radar system. © 2010 IEEE, Reprinted, with permission, from Wang et al. [2010].
Figure 8.43 Fabricated 2–10 GHz Vivaldi Array. © 2010 IEEE, Reprinted, with permission, from Wang et al. [2010]. (a) 1 × 8 linear Vivaldi full array; (b) prototype of single Vivaldi subarray; (c) measured return loss of the Vivaldi subarray.
Figure 8.44 Data acquisition and transfer module including two MAX104 ADCs, a Xilinx Virtex-4 FPGA evaluation board, a USB cable, and a laptop. © 2011 IEEE, Reprinted, with permission, from Wang and Fathy [2011].
Figure 8.45 Experimental setup for stationary target detection.
Figure 8.46 Real-time image of multiple stationary targets.
Figure 8.47 Experimental detection of a moving person.
Figure 8.48 Real-time images of a moving person.
Figure 8.49 Human marching on the spot with one-arm swing. © 2011 IEEE, Reprinted, with permission, from Wang and Fathy [2011]. (a) Range of the object versus time and (b) Doppler spectrogram (TF signature).
Figure 8.50 Human marching on the spot with two-arm swing. © 2011 IEEE, Reprinted, with permission, from Wang and Fathy [2011]. (a) Range of the object versus time and (b) Doppler spectrogram (TF signature).
Figure 8.51 Human walking with one-arm swing. © 2011 IEEE, Reprinted, with permission, from Wang and Fathy [2011]. (a) Range of the object versus time and (b) Doppler spectrogram (TF signature).
Figure 8.52 Human walking with two-arm swing. © 2011 IEEE, Reprinted, with permission, from Wang and Fathy [2011]. (a) Range of the object versus time and (b) Doppler spectrogram (TF signature).

Chapter 9: APPLICATIONS AND FUTURE RESEARCH

Figure 9.1 Sleep monitoring approaches. Polysomnography (PSG) involves several physiological sensors attached to the body, which can be connected to a body-worn wireless transponder (b). (c) Less comprehensive sensing can be performed without any bodily attachments.
Figure 9.2 Experiment setup at night in a hallway (a). Note that the building is not completely isolated from variations in the weather outside. FFT of linear demodulated data showing detected respiration rate at approximately 0.27 Hz for a 69-m distance (b).
Figure 9.3 Figure depicting a rescue situation where in addition to tracking the rescuer, it is critical to find any victim (conscious or unconscious). f_m1 and f_m2 refer to the physiological signal coming from persons m1 and m2, respectively. Since m2 is wearing a tag, he/she is also sending a signal 2f_m2.
Figure 9.4 A block diagram showing two-frequency radar setup where tag subtraction algorithms could be used to separate sources of motion.
Figure 9.5 Experiment setup showing the relative positions of mechanical target and human subject with respect to radar (a). Detected motion rate for different signals obtained from experiment III (b). The 4.9 GHz trace shows the successful detection of the tag motion at 0.4 Hz (24 BPM) exclusively. The trace from 2.45 GHz radar initially fails to track the respiration rate of human subject. After the application of ANC technique, the data from 2.45 GHz radar tracks the respiration of the human subject exclusively as verified by the reference chest belt.
Figure 9.6 Classification algorithm used for each radar to characterize motion.
Figure 9.7 Photograph showing the setup for monitoring chameleon activity.
Figure 9.8 (a) Raw data from front and side radar showing changes in amplitude due to motion and (b) result of the detection algorithm for front and side radar. The swaying of the body is detected as locomotion by the side radar and fidgeting by the front radar as expected. A few spurious alerts were generated by the eigen vector algorithm but were revealed as no motion by phase analysis and video reference.
Figure 9.9 A plot of 0–50 s taken from Figure 9.8 showing a new class of activity (large fidgeting motion) that cannot be considered as locomotion. But the radar has the capability to differentiate between small fidgeting and large fidgeting.

List of Tables

Chapter 1: INTRODUCTION

Table 1.1 Doppler Radar Physiological Monitoring from 1975 to 2014

Chapter 3: PHYSIOLOGICAL MOTION AND MEASUREMENT

Table 3.1 Mechanical Events of the Heart
Table 3.2 Techniques for Surface Measurement of Respiration Rate
Table 3.3 Techniques for Surface Measurement of Pulse Rate

Chapter 6: SOURCES OF NOISE AND SIGNAL-TO-NOISE RATIO

Table 6.1 System Parameters Used for SNR Calculation
Table 6.2 Measured and Collected Subject Data
Table 6.3 Bland–Altman Data for Heart Rate and Respiratory Rate Measurements at Each Range
Table 6.4 SNR Data for Heart and Respiration Measurements at Each Range
Table 6.5 Correlation Between Respiration SNR and Body Measurements at Each Range

Chapter 7: DOPPLER RADAR PHYSIOLOGICAL ASSESSMENTS

Table 7.1 Percentage of Data Without Motion Artifacts
Table 7.2 Patient Demographic Information
Table 7.3 Summary of Agreement of Doppler Radar with Reference Measurements
Table 7.4 Average Heart Rate and Bland–Altman Analysis for Seated Position Measurements
Table 7.5 Average Heart Rate and Bland–Altman Analysis for Supine Position Measurements
Table 7.6 HRV Indexes from Doppler Radar and ECG Signals of Subjects in Seated Positions
Table 7.7 HRV Indexes from Doppler Radar and ECG Signals of Subjects in Supine Positions
Table 7.8 RSA Peak–Valley Estimation of Seated Subjects
Table 7.9 RSA Peak–Valley Estimation of Supine Subjects

Chapter 8: ADVANCED PERFORMANCE ARCHITECTURES

Table 8.1 Ka-Band Radio Building Blocks and Their Specifications

Chapter 9: APPLICATIONS AND FUTURE RESEARCH

Table 9.1 Types of Sleep Monitoring Systems

Wiley Series in Biomedical Engineering and Multi-Disciplinary Integrated Systems

KAI CHANG, SERIES EDITOR

Advances in Optical Imaging for Clinical Medicine

Nicusor Iftimia, William R. Brugge, and Daniel X. Hammer (Editors)

Antigen Retrieval Immunohistochemistry Based Research and Diagnostics

Shan-Rong Shi and Clive R. Taylor

Introduction to Nanomedicine and Nanobioengineering

Paras N. Prasad

Biomedical Image Understanding

Joo-Hwee Lim, Sim-Heng Ong, and Wei Xiong (Editors)

Doppler Radar Physiological Sensing

Olga Boric-Lubecke, Victor M. Lubecke, Amy D. Droitcour, Byung-Kwon Park, and Aditya Singh (Editors)

DOPPLER RADAR PHYSIOLOGICAL SENSING

Edited by

OLGA BORIC-LUBECKE

VICTOR M. LUBECKE

AMY D. DROITCOUR

BYUNG-KWON PARK

ADITYA SINGH

No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning, or otherwise, except as permitted under Section 107 or 108 of the 1976 United States Copyright Act, without either the prior written permission of the Publisher, or authorization through payment of the appropriate per-copy fee to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, (978) 750-8400, fax (978) 750-4470, or on the web at www.copyright.com. Requests to the Publisher for permission should be addressed to the Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, (201) 748-6011, fax (201) 748-6008, or online at http://www.wiley.com/go/permission.

Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in preparing this book, they make no representations or warranties with respect to the accuracy or completeness of the contents of this book and specifically disclaim any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives or written sales materials. The advice and strategies contained herein may not be suitable for your situation. You should consult with a professional where appropriate. Neither the publisher nor author shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages.

For general information on our other products and services or for technical support, please contact our Customer Care Department within the United States at (800) 762-2974, outside the United States at (317) 572-3993 or fax (317) 572-4002.

Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic formats. For more information about Wiley products, visit our web site at www.wiley.com.

Doppler radar physiological sensing / edited by Olga Boric-Lubecke, Victor M. Lubecke, Amy D. Droitcour, Byung-Kwon Park, Aditya Singh.

I. Boric-Lubecke, Olga, 1966-, editor. II. Lubecke, Victor M., editor. III. Droitcour, Amy D., editor. IV. Park, Byung-Kwon, editor. V. Singh, Aditya, 1984-, editor.

[DNLM: 1. Heart Rate. 2. Monitoring, Physiologic– methods. 3. Respiratory Rate. 4. Signal Processing, Computer-Assisted. 5. Ultrasonography, Doppler– methods. WG 140]

List of Contributors

Olga Boric-Lubecke, Department of Electrical Engineering, University of Hawaii at Manoa, Honolulu, Hawaii, United States
Amy D. Droitcour, Wave 80 Biosciences, Inc., San Francisco, California, United States
Aly Fathy, Department of Electrical Engineering and Computer Science, University of Tennessee, Knoxville, Tennessee, United States
John Kiriazi, QCT RF Systems, Qualcomm Inc., San Diego, California, United States
Jenshan Lin, Department of Electrical and Computer Engineering, University of Florida, Gainesville, Florida, United States
Victor M. Lubecke, Department of Electrical Engineering, University of Hawaii at Manoa, Honolulu, Hawaii, United States
Wansuree Massagram, Department of Computer Science and Information Technology, Naresuan University, Phitsanulok, Thailand
Isar Mostafanezhad, Department of Physics and Astronomy, University of Hawaii at Manoa, Honolulu, Hawaii, United States
Byung-Kwon Park, DAS Sensor SW Engineering Team, Hyundai Mobis Mechatronics R&D Center, Gyeonggi-Do, South Korea
Aditya Singh, University of Hawaii Neuro-science and MRI research Program, John A. Burns School of Medicine, Honolulu, Hawaii, United States
Alex Vergara, Theranova LLC, San Francisco, California, United States
Yazhou Wang, Boston Design Center, RF Micro Devices, Inc., Billerica, Massachusetts, United States
Shuhei Yamada, Department of Electrical Engineering, University of Hawaii at Manoa, Honolulu, Hawaii, United States
Ehsan Yavari, Department of Electrical Engineering, University of Hawaii at Manoa, Honolulu, Hawaii, United States