This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS
1
Flexible Non-Constrained RF Wrist Pulse Detection Sensor Based on Array Resonators Yong-Jun An, Student Member, IEEE, Byung-Hyun Kim, Student Member, IEEE, Gi-Ho Yun, Sung-Woo Kim, Seung-Bum Hong, and Jong-Gwan Yook, Senior Member, IEEE
Abstract—This paper presents the development of a non-contact, nonintrusive wrist pulse sensor based on the near-field variation of an array resonator. A compact resonator and its array were designed and fabricated on flexible substrate. The reflection coefficient of the resonator can vary as a function of the distance between the resonator and the walls of the major arteries, and the corresponding variation is utilized to obtain heart rate information at the wrist. To detect very weak pulse signals from the main arteries, a sensitivity enhancement technique was devised using a radio frequency (RF) array resonator. The sensor system was implemented with an RF switch to combine or select appropriate signals from the resonator element and was tested using the 2.4 GHz ISM band. The results demonstrated the sensor system’s excellent performance in both sequential and simultaneous detection schemes. The measurement results showed that a heartbeat pulse can be detected from both radial and ulnar arteries via the array resonators. Considering the high sensitivity and characteristics, the proposed detection system can be utilized as a wearable, long-term health monitoring device. Index Terms—Array biosensor, RF resonator, sensitivity enhancement technique, vital sign sensor, wrist heartbeat pulse.
I. INTRODUCTION
I
N recent years, healthcare services and health monitoring devices have been combined with mobile smart systems to provide continuous monitoring of a user’s health condition and physical statuses, such as heartbeat or respiration rate. For this purpose, wearable devices will be good candidates for unobtrusive sensing [1]. Among various wearable sensors, impedance cardiography utilizes multiple conducting electrodes directly attached on the skin [2], and requires closed loop circuits during heartbeat detection. Of course, this technique can be used for a watch-type band or for two hand holders of a treadmill for Manuscript received October 08, 2014; revised January 09, 2015; accepted February 01, 2015. This research was supported by the Ministry of Science, ICT and Future Planning (MSIP), Korea, under the Convergence Information Technology Research Center (C-ITRC ) support program (NIPA-2014-H040114-1007), supervised by the National IT Industry Promotion Agency (NIPA). This paper was recommended by Associate Editor P. Chiang. Y.-J. An, B.-H. Kim, and J.-G. Yook are with the Departmetn of Electrical and Electronic Engineering, Yonsei University, Seoul 120-749, Korea (e-mail: [email protected]). G.-H. Yun is with the Information and Communications Engineering Department, Sungkyul University, Kyungki-do 430-742, Korea. S.-W. Kim and S.-B. Hong are with LG Electronics, Advanced Research Institute, Future IT R&D Laboratory, Seoul 137-724, Korea. Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TBCAS.2015.2406776
short-term detection or for a chest belt for relatively long-term continuous sensing. In contrast, non-invasive and continuous heartbeat detection methods at a single point, such as on a finger or a wrist, have avoided the attachment of multiple conducting electrodes to the skin. For a decade, piezo-electric sensors have been studied for sensing blood vessel pressure induced by the pulse on the wrist [3], [4]. Recently, several new types of pressure sensors, employing thin film transistors, nanowires or magnetoelastic sensors, have been explored and utilized for wrist pulse sensing [5]–[7]. It has been demonstrated that these pressure sensors can detect a wrist pulse, while additional cardiovascular parameters, such as arterial stiffness, can be extracted [8]. Other wrist pulse sensors have been developed based on infrared (IR) or red light signals with a photo detector, utilizing the fact that a light signal is partly absorbed by hemoglobin in the blood and the reflected signal contains information regarding the saturation rate of oxygen in the blood [9]–[11]. This kind of device is called a photoplethysmography (PPG) sensor. Currently, the abovementioned types of pressure and light-based sensors are popular. However, these sensors should be tightly fastened, and they continuously press on the wrist or finger to acquire pressure information from an artery. Moreover, since light-based sensors are strongly affected by the brightness of the environment, these types of sensors should be blocked from other light sources, as well as tightly fastened. In addition, the abovementioned types of sensors are uncomfortable due to contact to skin. Thus, an electromagnetic fieldbased RF sensor could be an alternative method for detecting vital signals, since such a sensor could detect physical movement without direct contact to a surface of subject. To date, use of RF sensors has been focused on the detection of respiration and heart beat signals near the chest without direct contact with the chest [12]–[16]. To detect a heartbeat signal at a wrist, a much better sensing method is necessary because the pulse movement of the radial artery is much smaller than the chest movement due to breathing or heartbeat. A previous study showed that active isolation techniques can be used to detect small variations in the reflection signal from a resonator [17]. Using this technique, a wrist pulse has been successfully detected with a single resonator at a position near the radial or the ulnar arteries [18]. In actual use, however, if the position of the sensor resonator shifted to a position far from the radial or ulnar arteries, a deteriorated signal quality would result. In this paper, a wrist pulse detection sensor using an array of RF resonators with a sensitive detection scheme is proposed, as illustrated in Fig. 1. The proposed array sensor system can detect a wrist pulse at multiple points on the wrist, such as
1932-4545 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. 2
IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS
Fig. 1. Proposed wrist pulse detection system based on the near field variation of a resonator.
at the positions of radial and ulnar arteries. The compact resonator and its array are designed on a flexible substrate, and the array resonators are connected to a receiver circuit that includes a SP4T (single pole four throw) RF switch. With this switch it is possible to operate in either a sequential or a simultaneous switching mode, and a heartbeat pulse can be detected on 3 points on the wrist. Note that the simultaneous switching ensures reliable signal acquisition. This paper is organized as follows: Section II presents the design and implementation of the compact resonator on a flexible substrate and the near-field effect due to the distances from the resonator to skin is analyzed. In addition, the design and measurements of the array resonator are presented. Section III presents the system configuration and theoretical analysis results. Section IV presents the experimental results obtained using sequential and simultaneous switching, and these results are compared with piezoelectric reference signal data.
Fig. 2. The resonator structure and dimension with the human body model and the preventive layer.
II. RESONATOR DESIGN A. Single Resonator The principal idea behind a non-contact type RF sensor for vital health signal detection is the utilization of the electromagnetic near-field variation due to the physiological movement of a subject. In the present work, an RF resonator is employed as a sensing element. The resonator can be placed (not fixed or attached) on a wrist or embedded in a strap of a smart watch. For wearable applications, the resonators are designed on a flexible PCB with limited space to be able to fit into the straps of wearable devices. In this study, 0.25 mm thick teflon is used and, to minimize the size of the resonator, a patch-type planar inverted-F antenna (PIFA) structure is employed. In general, the resonant frequency of a resonator is down shifted when it is placed close to a human body. To operate near the target frequency (2.4 GHz ISM band), a length of 19.8 mm was chosen for , the length of the resonator, which is smaller than a quarter-wavelength at the target frequency, as shown in Fig. 2. To predict the amount of resonance frequency shift, a closed form model for the resonant frequency of rectangular microstrip antennas with spaced superstrates [19] is used. For a PIFA-type resonator, the equation of the resonant frequency is modified, as in [20] (1)
Fig. 3. Calculation, 3D EM simulation, and measurement results of resonant frequency variation found in the human skin model adjacent to the resonator, . and the calculated effective permittivity
where is the speed of light, is the overall length of the PIFA structure, is the effective open line length, and is the effective permittivity, including the preventive layer as well as the body model, as shown in Fig. 2. For comparison, the permittivity of the preventive layer is set to 3.2 (acrylic), and the human body model is set to 37 [21]. Since the frequency variation of the resonator is the basic mechanism behind the near-field RF sensor, the characteristics of the frequency variations have been carefully modeled, and a three-dimensional, full-wave electromagnetic simulation based on the finite element method (FEM) has been utilized. A box with skin tissue property is positioned over the resonator at distances ranging from 0.25 mm to 2 mm, with a 0.25 mm increment, and the intermediate space is filled with a preventive layer . The resonator design and dimensions are shown in Fig. 2, where is one of the control variables of the resonator which is related to operating frequency as well as impedance match. Before the simulation, the effective permittivity and resonance frequency are calculated by a superstrate method and calculated result is shown in Fig. 3 [19]. It is
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. AN et al.: FLEXIBLE NON-CONSTRAINED RF WRIST PULSE DETECTION SENSOR BASED ON ARRAY RESONATORS
3
TABLE I SIMULATION RESULT OF THE RESONANT FREQUENCY VARIATION WITH ADJACENT SKIN MODEL DEFENDING ON
Fig. 5. Photograph of the flexible array resonator with 5 mm spacing.
Fig. 4. Simulated and measured S-parameter results of the single resonator with varying of the distance (0.25, 0.5, 0.75 mm).
clear that the resonant frequency is downshifted when the resonator is placed close to the human body. For the general rectangular PIFA structure ( in Fig. 2), 3D EM simulation results reveal a very good agreement with calculation results, as shown in Fig. 3. It shows that the PIFA structure can be analyzed using the superstrate method [19]. The frequency variation, as a function of the control variable , is calculated as summarized in Table I. In the table, refers to the resonance frequency with the resonator-to-skin distance of 0.25 mm, while refers to a corresponding distance of 0.75 mm. It should be noted that the resonance frequency is downshifted when becomes large, and the frequency deviation is maximized when is 3.8 mm. As shown in Fig. 2, the proposed resonator is simulated and compared to a rectangular one , as shown in Fig. 3. It was found that the proposed resonator reveals greater frequency variation than that of the rectangular shaped resonator, meaning that the proposed structure has sensitive characteristics for near-field sensing applications. Measurement results are shown in Figs. 3 and 4 as a function of the spacing between the resonator and skin, where layers of acrylic tapes (thickness: 62.5 m) are used to keep the separation constant. The measurement results show excellent correlation with the simulation results. B. Array Resonator On the wrist, a pulse can be easily detected on several locations where the arteries are found. However, since the shape and position of a wrist strap could be slightly different for different types of equipment and since a strap can rotate on the wrist during movement, the position of the resonator cannot be fixed. For a minimized form factor, the size of the proposed resonator is less than 20 mm, and near-field characteristics are utilized; therefore, the detection areas are relatively small with a single resonator. To overcome this issue, array resonator geometry is proposed in this work to extend the detection area to as large as possible. In addition, a multiple detection scheme utilizing the
array resonator are also proposed. As shown in Fig. 5, the proposed array resonators are placed 5 mm apart from one another, extending the detection coverage area to about 35 mm sufficient area to sense at least one artery of the wrist. Fig. 6 shows both the measured and the 3D modeled simulation results for four resonators as a function of resonator-to-skin distances ranging from 0.25 mm to 0.75 mm. Overall, the measured results agree reasonably well with the simulation results, and they reveal a clear distinction with distance and the possibility of versatile wrist pulse detection scheme. C. Switch With the Array Resonator The array resonator is connected to the detection system, with an absorptive single-pole, four-throw (SP4T) RF switch, and the switch is controlled by a 2-bit binary signal. The electrical spacing between the resonators (edge-to-edge) is about (5 mm) to ensure reasonable isolation between the resonators. Note that while such isolation is generally an important parameter in array antenna design, it might be relatively unimportant in the proposed system. It should be emphasized that the proposed detection system can be operated with one resonator at a time and that any leakage to the other resonators is absorbed by the absorptive off-state port of the switch. This absorptive characteristic provides a high isolation level between output ports of the switch and does not disturb impedance matching performance. III. SYSTEM CONFIGURATION In this Section, a sensitivity enhancement technique based on an active isolation technique is proposed. This technique allows detection of the reflected signal from the resonator with much higher sensitivity compared to using a system with a single detector. For the array resonators, proper analysis and specific conditions can be determined as follows. A. Operating Principle The reflected signal from the resonator can be expressed as [17]
(2) where is the reflection coefficient of the resonator, is the transmitted signal, is the constant component as a large signal of the reflection coefficient, and is the time-varying
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. 4
IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS
Fig. 7. Block diagram of the proposed system.
(3) is the initial where is the detector coefficient, and output voltage when the detector input is zero. To obtain the conversion gain slope of the detector, the output voltage in (3) is partially differentiated with
(4) The above (4) shows the sensitivity of the proposed system. should be minimized, thereby, the To increase the sensitivity, active isolator concept can be best suited increasing sensitivity by reducing the constant component of [17]. B. Array Detection System Configuration
Fig. 6. Simulated and measured S-parameter results of the array resonator when varying the distance (0.25, 0.5, 0.75 mm). (a) First resonator. (b) Second resonator. (c) Third resonator. (d) Fourth resonator.
pulse information as the small signal component, respectively. A coupler is utilized to detect the reflected signal. Therefore, the reflected signal is coupled to where is the coupling coefficient of the coupler. If using the RF detector to detect information, the output of the RF detector can be expressed as
A block diagram of the proposed system is shown in Fig. 7. In the figure, is transmitted into port 2 of the coupler and most of the signal reaches the RF switch which is connected to port 1 of the coupler. Then the RF signal is relayed to each resonator, depending on the RF switch status. As mentioned in Section II, the resonator elements of the array have a reflection coefficient dependent on the near-field variation due to physiological movement. The reflected signal can be expressed as “ “ “ “
” ” ” ” (5)
where is the reflection coefficient of the resonator elements, and is the 2 bit binary signal to control the RF switch. Now, the reflected signal from the resonator is coupled to port 3 of the
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. AN et al.: FLEXIBLE NON-CONSTRAINED RF WRIST PULSE DETECTION SENSOR BASED ON ARRAY RESONATORS
5
coupler and becomes . In the same way, the reflected signal is coupled to port 4 and can be expressed as . For phase synchronization a phase shifter is used on the reflected signal, and the signal becomes , where is the shifted phase. In addition, a variable attenuator is utilized to adjust the magnitude of the transmitted signal at port 4 and the adjusted signal becomes . These two signals are combined at the power combiner, and the combined signal can be expressed as follows:
(6) The output of the detector is now differentiated with respect to to obtain the sensitivity of the system, as follows:
(7) To increase the slope, that is, the sensitivity of the combined signal, the denominator in (7) should be as small as possible. For sensitivity enhancement, the phase shifter and the variable attenuator should be properly set as
(8) It is necessary to note that the reflection coefficient of each resonator is slightly different as shown in Fig. 6. As a result, these phase and magnitude differences can degrade the detection sensitivity. Fig. 8, depicts the calculated results based on (7). The figure shows that the sensitivity enhancement curves are dependent on the magnitude and phase differences, and reveals maximum sensitivity when the ideal conditions are satisfied as in (8). It is clear that the sensitivity can be increased about 40 dB as shown in Fig. 8. Note that the amount the sensitivity enhancement varies with respect to both the magnitude ratio of the reflected and the attenuated signal and the phase difference . As shown in Fig. 8(a), if the phase difference is fixed to 180 degrees, the sensitivity enhancement is symmetrically decreased, while is shifted away from the highest sensitivity enhancement value at the center point . However, when the phase difference is shifted from 180 degrees, the center point of the symmetry is shifted to . The center point is determined by using , consequentially the sensitivity enhancement region is largely spreaded on the condition. Fig. 8(b) shows that with a 20% magnitude difference between two signals, the sensitivity can be increased at least 7 dB, where the phase difference is near 180 degree. In a case where
Fig. 8. (a) Calculated result of the sensitivity enhancement depending on the magnitude and phase difference. (b) Calculated results about sensitivity is 0.8, 1, and 1.2. (c) Measured phase of and enhancement where on 2.42 GHz. phase difference with the other resonators
(
in Fig. 8), the sensitivity also increases to near points. therefore, when using a smaller magnitude of than , the sensitivity enhancement range is extended on the phase difference domain in Fig. 8(a). This means that if the condition of the proposed array system is set to , some of the phase mismatch can be accepted for sensitivity enhancement. Also, the measured phase results are shown in Fig. 8(c), it is show that the phase variances of other resonators are less than 60 degrees except in 0.75 mm case. According to Fig. 8(a), at least 10 dB sensitivity could be increased where with measured phase differences. Therefore, the proposed array detection system can provide high sensitivity without controlling the phase shifter and variable attenuator for each resonator. A photograph of the fabricated circuit is shown in Fig. 9. The overall size of the circuit is 30 mm 30 mm. Note that a
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. 6
IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS
Fig. 11. Measurement setup of the proposed wrist pulse detection system.
Fig. 9. Photograph of the proposed wrist pulse detection circuit.
prototype circuit, a power consumption of utilized the RF amplifier and the power detector is not considered. Supplied power is 5 V with 45 mA (RF amp: 20 mA, power detector: 25 mA), it could be reduced with the low-powered components. IV. EXPERIMENTAL RESULTS
Fig. 10. Photograph of the equipment of the array resonator for the wrist pulse detection.
thin film directional coupler (AVX C0402) is utilized to minimize the size of the circuit, and an SP4T absorptive switch (SKY13384) controlled by 2-bit digital signal is employed. For the connection between the RF switch and the array resonator, flexible coaxial cables are employed, as shown in Fig. 10. To control the amplitude , an analog voltage controlled variable attenuator (RF2013A) is utilized, and varactor diodes are used for the voltage controlled phase shifter [22]. A chip power combiner device based on the Wilkinson power divider is used to combine signals, and an RF amplifier (ABA51563) is connected to the output of the power combiner for amplification. In the last stage a power detector (LT5538) detects the output power of the combined signal, containing the wrist pulse information. As a
To detect a heart pulse on the wrist, the array resonator is placed around the wrist. In a previous study [18], since the pulse can be detected directly above the radial artery, the position of the first resonator element (S1 in Fig. 10) is placed close to the radial artery. The remaining resonator elements are spaced at 5 mm. Therefore, the fourth resonator is placed near the ulnar artery. To keep the gap between the resonator and the skin constant, acrylic tape and styrofoam are employed as a preventive layer as well as a spacer. The gap is kept at about 1 mm during the experiment, as shown in Fig. 10. The experiment setup is shown in Fig. 11. A signal generator is used as a signal source and is set to 2.42 GHz with 0 dBm output power. The resonator is connected to the circuit using a flexible coaxial cable, as shown in Fig. 10. As a reference, a piezoelectric pressure sensor is fastened on a finger. The output of the fabricated sensor circuit and the reference signal are recorded by a 24-bit NI DAQ. Note that the RF switch control bit signals are generated by a clock pulse generator, which is controlled by a hard-coded program, and these control bits are recorded by the DAQ board to monitor the switch status. The experimental environments and conditions as follows: First, to demonstrate the capabilities of the proposed sensor, motion artifact is restricted during the experiment. Second, a 2.4 GHz WiFi access point (AP) operates nearby place as a noise source, where the power of the noise signal is measured about dBm. First, a sequential detection is performed; the measurement results of which are shown in Fig. 12. The switch control binary bits are changed every 5 seconds with a detection sequence of S3, S2, S1, and S4, as illustrated in Fig. 12(a). In Fig. 12(b), the detector output signals are divided into each switching state and low pass filtered with 30 Hz cut-off frequency to reduce the 60 Hz noise produced by the power supply. It is clear that the signals from resonators S1 (10–15 sec), S2 (5–10 sec) and S4 (15–20 sec) show reasonably good heartbeat waveforms. As expected, the detected pulses from the radial and the ulnar arteries reveal higher vital signal levels. It is interesting to observe that
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. AN et al.: FLEXIBLE NON-CONSTRAINED RF WRIST PULSE DETECTION SENSOR BASED ON ARRAY RESONATORS
7
Fig. 12. Measurement result of the proposed wrist pulse detection system. (a) Reference signal and detection signal switching the array resonator with every 5 seconds. (b) Low pass filtered and magnified detection signals, each state shows different result.
the pulses from the radial artery clearly reveal second and third peaks that are reflected from the lower body (hand). The time delay and amplitude ratio between the first and second peaks can be utilized to extract various medical and physiological information such as blood pressure and arterial stiffness [8]. In the previous PPG study it was shown that detecting the pulse in the middle of the wrist [11] is possible, and this technique is similar to our S2 position, but with much less signal power. In the second experiment, simultaneous detection is performed with the array resonator. The switch control binary bits are changed every 25 milliseconds so that each resonator is selected 10 times/sec. To reduce the complexity of signal processing, an identical 10 sample/sec sampling rate is used in this work. The measurement results are shown in Fig. 13 with the reference data of Fig. 13(a). It is clear that all 4 resonators can detect a heartbeat pulse with different signal levels, while S1 and S4 reveal excellent pulse waveforms. To demonstrate this in detail, all the sampled signals are interpolated with a sinc function. In accordance with the Nyquist theorem, signals detected via the 10 samples/sec per channel can be recovered up to 5 Hz, and this renders to resolve the second peak in the
Fig. 13. Measurement result with simultaneous switching at 10 samples/sec. (a) Reference pulse signal. (b) Sampled data from the proposed system. Detection signal with interpolation from (c) the first resonator, (d) the second resonator, (e) the third resonator, and (f) the fourth resonator.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. 8
IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS
pulse from the sampled result. Interpolated data sets are shown in Fig. 13(c), (d), (e), and (f). It should be noted that the second peak can be observed in most pulse waves at S4 as shown in Fig. 13(f), and this is very similar to ballistocardiogram (BCG) measurement reading. This information could be utilized as the main diagnosis factor for artery stiffness [8]. If all the waveforms can be combined together, it would be possible to extract clinically important information. For now, it is clear that the array resonator can detect heartbeat pulse on a wrist at multiple locations. V. CONCLUSION This paper has presented a non-constrained radio frequency wrist pulse detection system based on reflection coefficient variation due to blood flow in the radial and ulnar arteries. The proposed system utilizes an array of resonators, which were designed and fabricated on a flexible teflon substrate with a human skin model on top of the resonators. The measured results for a single resonator and an array of resonators were well matched to the simulation results. In addition, an RF switch and a pulse detection circuit were designed to realize the proposed sensitivity enhancement technique, and the phase and magnitude mismatch effect with the array resonator was analyzed. It is clear that the sensitivity of the system is strongly dependent on the value of the phase shifter as well as that of the variable attenuator. However, in some condition, the proposed system can be used with multiple resonators without changing the conditions of the phase shifter or the variable attenuator for each resonator. The sensor system was tested with two different switching modes and demonstrated successful detection of a wrist pulse signal at multiple locations on the wrist. This shows that the proposed system is location insensitive and can be placed on somewhat arbitrary locations on the wrist. It is clear that the proposed non-constrained RF vital sign sensor has a great advantage in that it can be integrated into wearable communication devices, such as a smart watch, with biomedical and healthcare applications. REFERENCES [1] Y.-L. Zheng, X.-R. Ding, C. C. Y. Poon, B. P. L. Lo, H. Zhang, X.-L. Zhou, G.-Z. Yang, N. Zhao, and Y.-T. Zhang, “Unobtrusive sensing and wearable devices for health informatics,” IEEE Trans. Biomed. Eng., vol. 61, no. 5, pp. 1538–1554, May 2014. [2] G. Cotter, Y. Moshkovitz, E. Kaluski, A. J. Cohen, H. Miller, D. Goor, and Z. Vered, “Accurate, noninvasive continuous monitoring of cardiac output by whole-body electrical bioimpedance*,” Chest, vol. 125, pp. 1431–1440, 2004. [3] J. Foo and C. Lim, “Pulse transit time based on piezoelectric technique at the radial artery,” J. Clin. Monit. Comput., vol. 20, pp. 185–192, 2006. [4] V. Almeida, H. C. Pereiraa, T. Pereiraa, E. Figueirasa, E. Borgesa, J. M. R. Cardosoa, and C. Correiaa, “Piezoelectric probe for pressure waveform estimation in flexible tubes and its application to the cardiovascular system,” Sens. Actuators A, Phys., vol. 169, no. 1, pp. 217–226, 2011. [5] G. Schwartz, B. C. K. Tee, J. Mei, A. L. Appleton, D. H. Kim, H. Wang, and Z. Bao, “Flexible polymer transistors with high pressure sensitivity for application in electronic skin and health monitoring,” Nat. Commun., vol. 4, p. 1859, 2013. [6] S. Gong, W. Schwalb, Y. Wang, Y. Chen, Y. Tang, J. Si, B. Shirinzadeh, and W. Cheng, “A wearable and highly sensitive pressure sensor with ultrathin gold nanowires,” Nat. Commun., vol. 5, 2014. [7] C. Hlenschi, S. Corodeanu, and H. Chiriac, “Magnetoelastic sensors for the detections of pulse waves,” IEEE Trans. Magn., vol. 49, pp. 117–119, 2013.
[8] W. W. Nichols, “Clinical measurement of arterial stiffness obtained from noninvasive pressure waveforms,” Amer. J. Hypertens., vol. 18, pp. 3S–10S, Jan. 2005. [9] P. Renevey, R. Vetter, and J. Krauss, “Wrist-located pulse detection using IR signals, activity and nonlinear artefact cancellation,” in Proc. 23rd IEEE Engineering in Medicine and Biology Soc. Conf., Istanbul, Turkey, 2001, pp. 3030–3033. [10] E. Geun, H. Heo, K. C. Nam, and Y. Huh, “Measurement site and applied pressure consideration in wrist photoplethysmography,” in Proc. 23rd Int. Tech. Conf. Circuits/Systems, Computers and Communications, 2008, pp. 1129–1132. [11] K. Li and S. Warren, “A wireless reflectance pulse oximeter with digital baseline control for unfiltered photoplethysmograms,” IEEE Trans. Biomed. Circuits Syst., vol. 6, no. 3, pp. 269–278, Jun. 2012. [12] A. A. Serra, P. Nepa, G. Manara, G. Corsini, and J. L. Volakis, “A single on-body antenna as a sensor for cardiopulmonary monitoring,” IEEE Ant. Wireless Propag. Lett., vol. 9, pp. 930–933, 2010. [13] S. G. Kim, G. H. Yun, and J. G. Yook, “Compact vital signal sensor using oscillation frequency deviation,” IEEE Trans. Microw. Theory Tech., vol. 60, pp. 393–400, 2012. [14] S. G. Kim, G. H. Yun, and J. G. Yook, “Wireless RF sensor structure for non-contact vital sign monitoring,” J. Korean Inst. Electr. Eng. Sci., vol. 12, pp. 37–44, 2012. [15] Y. Hong, S. G. Kim, B. H. Kim, S. J. Ha, H. J. Lee, G. H. Yun, and J. G. Yook, “Noncontact proximity vital sign sensor based on PLL for sensitivity enhancement,” IEEE Trans. Biomed. Circuits Syst., vol. 8, no. 4, pp. 584–593, 2014. [16] Y.-J. An, G.-H. Yun, and J.-G. Yook, “Vital sign detection for handheld communication device using antenna mismatching effect,” in Proc. IEEE MTT-S Int. Microw. Symp. Dig., Jun. 2–7, 2013. [17] Y.-J. An, G.-H. Yun, and J.-G. Yook, “Sensitivity enhanced vital sign detection based on antenna reflection coefficient variation,” IEEE Trans. Biomed. Circuits Syst., 2014, submitted for publication. [18] Y.-J. An, G.-H. Yun, S. W. Kim, and J.-G. Yook, “Wrist pulse detection system based on changes in the near-field reflection coefficient of a resonator,” IEEE Microw. Wireless Compon. Lett., 2014, accepted for publication. [19] J. T. Bernhard and C. J. Tousignant, “Resonant frequencies of rectangular microstrip antennas with flush and spaced dielectric superstrates,” IEEE Trans. Antennas Propag., vol. 47, pp. 302–308, Feb. 1999. [20] I. J. Bahl, P. Bhartia, and S. Stuchly, “Design of microstrip antennas covered with a dielectric layer,” IEEE Trans. Antennas Propag., vol. 30, pp. 314–318, Mar. 1982. [21] S. Gabriel, R. W. Lau, and C. Gabriel, “The dielectric properties of biological tissues: II. Measurements in the frequency range 10 Hz to 20 GHz,” Phys. Med. Biol., vol. 41, pp. 2251–2269, 1996. [22] H. Kim, A. B. Kozyrev, A. Karbassi, and D. W. van der Weide, “Linear tunable phase shifter using a left-handed transmission line,” IEEE Microw. Wireless Compon. Lett., vol. 15, no. 5, pp. 366–368, May 2005.
Yong-Jun An (S’12) was born in Seoul, Korea. He received the B.S. degree in electrical and electronic engineering from Yonsei University, Seoul, Korea, in 2009. Currently, he is working toward the Ph.D. degree in electrical and electronic engineering at Yonsei University. His research interests include remote wireless vital signal monitoring sensors, RF passive circuits, and radars.
Byung-Hyun Kim (S’12) was born in Incheon, South Korea. He received the B.S. degree in electrical and electronics engineering from Yonsei University, Seoul, South Korea, in 2012. Currently, he is working toward the Ph.D. degree in electrical and electronics engineering at Yonsei University. He received the Spring 2012 Undergraduate-Pre-graduate (BS/MS) Scholarship from the IEEE Microwave Theory and Techniques Society (MTT-S). His main research interests are in the microwave/millimeter-wave component systems, remote wireless vital signal monitoring, RF biosensors, RF passive circuits, and radars.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. AN et al.: FLEXIBLE NON-CONSTRAINED RF WRIST PULSE DETECTION SENSOR BASED ON ARRAY RESONATORS
Gi-Ho Yun was born in Jeonju-Si, Korea. He received the B.S., M.S., and Ph.D. degrees in electronics engineering from Yonsei University, Seoul, South Korea, in 1984, 1986, and 1999, respectively. From 1985 to 1997, he worked at Samsung Electronics and Samsung Electro-Mechanics. He also served at Honam University, Gwangju, South Korea, from 1997 to 2008. Currently, he is an Assistant Professor in the School of Information and Communication Engineering, Sungkyul University, Kyeonggi-Do, Korea, and also working with the Advanced Computational Electromagnetic Laboratory of Yonsei University. His research interests include radio frequency (RF) circuits, patch antennas, and biosensors.
Sung-Woo Kim was born in Incheon, South Korea. He received the B.S. degree in biomedical engineering from Yonsei University, Wonju, South Korea, in 2004. From 2004 to 2008, he was a Research Assistant in the Department of Medical Engineering, Yonsei University College of Medicine, Severance Hospital, Seoul, South Korea. Currently, he is working toward the Ph.D. degree in biomedical engineering at Yonsei University. He is a Senior Research Engineer with the Biometrics team, Future IT R&D Laboratory, LG Electronics Advanced Research Institute, Seoul, South Korea. His main research interests include biosignal processing, biosensors such as PPG/IPG, and healthcare using the smart watch.
9
Seung-Bum Hong was born in Seoul, South Korea. He received the B.S. and M.S. degrees in electronics engineering from Yonsei University, Seoul, South Korea, in 1999 and 2001, respectively. Currently, he is a Principal Research Engineer with the Biometrics team, Future IT R&D Laboratory, LG Electronics Advanced Research Institute, Seoul, South Korea, and is an advisory committee member with the KFDA. His main research interests include biosignal processing, biosensors, and the mHealth project.
Jong-Gwan Yook (S’89–M’97–SM’12) was born in Seoul, South Korea. He received the B.S. and M.S. degrees in electronics engineering from Yonsei University, Seoul, South Korea, and the Ph.D. degree from the University of Michigan, Ann Arbor, MI, USA, in 1987, 1989, and 1996, respectively. Currently, he is a Professor with the School of Electrical and Electronic Engineering, Yonsei University. His main research interests are in the areas of theoretical/numerical electromagnetic modeling and characterization of microwave/millimeter-wave circuits and components, design of radio frequency integrated circuits (RFICs) and monolithic microwave integrated-circuits (MMICs), and analysis and optimization of high-frequency high-speed interconnects, including signal/power integrity (EMI/EMC), based on frequency as well as time-domain full-wave methods. Recently, his research team developed various biosensors, such as carbon-nano-tube RF biosensors for nanometer size antigen-antibody detection as well as remote wireless vital signal monitoring sensors.
1
Flexible Non-Constrained RF Wrist Pulse Detection Sensor Based on Array Resonators Yong-Jun An, Student Member, IEEE, Byung-Hyun Kim, Student Member, IEEE, Gi-Ho Yun, Sung-Woo Kim, Seung-Bum Hong, and Jong-Gwan Yook, Senior Member, IEEE
Abstract—This paper presents the development of a non-contact, nonintrusive wrist pulse sensor based on the near-field variation of an array resonator. A compact resonator and its array were designed and fabricated on flexible substrate. The reflection coefficient of the resonator can vary as a function of the distance between the resonator and the walls of the major arteries, and the corresponding variation is utilized to obtain heart rate information at the wrist. To detect very weak pulse signals from the main arteries, a sensitivity enhancement technique was devised using a radio frequency (RF) array resonator. The sensor system was implemented with an RF switch to combine or select appropriate signals from the resonator element and was tested using the 2.4 GHz ISM band. The results demonstrated the sensor system’s excellent performance in both sequential and simultaneous detection schemes. The measurement results showed that a heartbeat pulse can be detected from both radial and ulnar arteries via the array resonators. Considering the high sensitivity and characteristics, the proposed detection system can be utilized as a wearable, long-term health monitoring device. Index Terms—Array biosensor, RF resonator, sensitivity enhancement technique, vital sign sensor, wrist heartbeat pulse.
I. INTRODUCTION
I
N recent years, healthcare services and health monitoring devices have been combined with mobile smart systems to provide continuous monitoring of a user’s health condition and physical statuses, such as heartbeat or respiration rate. For this purpose, wearable devices will be good candidates for unobtrusive sensing [1]. Among various wearable sensors, impedance cardiography utilizes multiple conducting electrodes directly attached on the skin [2], and requires closed loop circuits during heartbeat detection. Of course, this technique can be used for a watch-type band or for two hand holders of a treadmill for Manuscript received October 08, 2014; revised January 09, 2015; accepted February 01, 2015. This research was supported by the Ministry of Science, ICT and Future Planning (MSIP), Korea, under the Convergence Information Technology Research Center (C-ITRC ) support program (NIPA-2014-H040114-1007), supervised by the National IT Industry Promotion Agency (NIPA). This paper was recommended by Associate Editor P. Chiang. Y.-J. An, B.-H. Kim, and J.-G. Yook are with the Departmetn of Electrical and Electronic Engineering, Yonsei University, Seoul 120-749, Korea (e-mail: [email protected]). G.-H. Yun is with the Information and Communications Engineering Department, Sungkyul University, Kyungki-do 430-742, Korea. S.-W. Kim and S.-B. Hong are with LG Electronics, Advanced Research Institute, Future IT R&D Laboratory, Seoul 137-724, Korea. Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TBCAS.2015.2406776
short-term detection or for a chest belt for relatively long-term continuous sensing. In contrast, non-invasive and continuous heartbeat detection methods at a single point, such as on a finger or a wrist, have avoided the attachment of multiple conducting electrodes to the skin. For a decade, piezo-electric sensors have been studied for sensing blood vessel pressure induced by the pulse on the wrist [3], [4]. Recently, several new types of pressure sensors, employing thin film transistors, nanowires or magnetoelastic sensors, have been explored and utilized for wrist pulse sensing [5]–[7]. It has been demonstrated that these pressure sensors can detect a wrist pulse, while additional cardiovascular parameters, such as arterial stiffness, can be extracted [8]. Other wrist pulse sensors have been developed based on infrared (IR) or red light signals with a photo detector, utilizing the fact that a light signal is partly absorbed by hemoglobin in the blood and the reflected signal contains information regarding the saturation rate of oxygen in the blood [9]–[11]. This kind of device is called a photoplethysmography (PPG) sensor. Currently, the abovementioned types of pressure and light-based sensors are popular. However, these sensors should be tightly fastened, and they continuously press on the wrist or finger to acquire pressure information from an artery. Moreover, since light-based sensors are strongly affected by the brightness of the environment, these types of sensors should be blocked from other light sources, as well as tightly fastened. In addition, the abovementioned types of sensors are uncomfortable due to contact to skin. Thus, an electromagnetic fieldbased RF sensor could be an alternative method for detecting vital signals, since such a sensor could detect physical movement without direct contact to a surface of subject. To date, use of RF sensors has been focused on the detection of respiration and heart beat signals near the chest without direct contact with the chest [12]–[16]. To detect a heartbeat signal at a wrist, a much better sensing method is necessary because the pulse movement of the radial artery is much smaller than the chest movement due to breathing or heartbeat. A previous study showed that active isolation techniques can be used to detect small variations in the reflection signal from a resonator [17]. Using this technique, a wrist pulse has been successfully detected with a single resonator at a position near the radial or the ulnar arteries [18]. In actual use, however, if the position of the sensor resonator shifted to a position far from the radial or ulnar arteries, a deteriorated signal quality would result. In this paper, a wrist pulse detection sensor using an array of RF resonators with a sensitive detection scheme is proposed, as illustrated in Fig. 1. The proposed array sensor system can detect a wrist pulse at multiple points on the wrist, such as
1932-4545 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. 2
IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS
Fig. 1. Proposed wrist pulse detection system based on the near field variation of a resonator.
at the positions of radial and ulnar arteries. The compact resonator and its array are designed on a flexible substrate, and the array resonators are connected to a receiver circuit that includes a SP4T (single pole four throw) RF switch. With this switch it is possible to operate in either a sequential or a simultaneous switching mode, and a heartbeat pulse can be detected on 3 points on the wrist. Note that the simultaneous switching ensures reliable signal acquisition. This paper is organized as follows: Section II presents the design and implementation of the compact resonator on a flexible substrate and the near-field effect due to the distances from the resonator to skin is analyzed. In addition, the design and measurements of the array resonator are presented. Section III presents the system configuration and theoretical analysis results. Section IV presents the experimental results obtained using sequential and simultaneous switching, and these results are compared with piezoelectric reference signal data.
Fig. 2. The resonator structure and dimension with the human body model and the preventive layer.
II. RESONATOR DESIGN A. Single Resonator The principal idea behind a non-contact type RF sensor for vital health signal detection is the utilization of the electromagnetic near-field variation due to the physiological movement of a subject. In the present work, an RF resonator is employed as a sensing element. The resonator can be placed (not fixed or attached) on a wrist or embedded in a strap of a smart watch. For wearable applications, the resonators are designed on a flexible PCB with limited space to be able to fit into the straps of wearable devices. In this study, 0.25 mm thick teflon is used and, to minimize the size of the resonator, a patch-type planar inverted-F antenna (PIFA) structure is employed. In general, the resonant frequency of a resonator is down shifted when it is placed close to a human body. To operate near the target frequency (2.4 GHz ISM band), a length of 19.8 mm was chosen for , the length of the resonator, which is smaller than a quarter-wavelength at the target frequency, as shown in Fig. 2. To predict the amount of resonance frequency shift, a closed form model for the resonant frequency of rectangular microstrip antennas with spaced superstrates [19] is used. For a PIFA-type resonator, the equation of the resonant frequency is modified, as in [20] (1)
Fig. 3. Calculation, 3D EM simulation, and measurement results of resonant frequency variation found in the human skin model adjacent to the resonator, . and the calculated effective permittivity
where is the speed of light, is the overall length of the PIFA structure, is the effective open line length, and is the effective permittivity, including the preventive layer as well as the body model, as shown in Fig. 2. For comparison, the permittivity of the preventive layer is set to 3.2 (acrylic), and the human body model is set to 37 [21]. Since the frequency variation of the resonator is the basic mechanism behind the near-field RF sensor, the characteristics of the frequency variations have been carefully modeled, and a three-dimensional, full-wave electromagnetic simulation based on the finite element method (FEM) has been utilized. A box with skin tissue property is positioned over the resonator at distances ranging from 0.25 mm to 2 mm, with a 0.25 mm increment, and the intermediate space is filled with a preventive layer . The resonator design and dimensions are shown in Fig. 2, where is one of the control variables of the resonator which is related to operating frequency as well as impedance match. Before the simulation, the effective permittivity and resonance frequency are calculated by a superstrate method and calculated result is shown in Fig. 3 [19]. It is
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. AN et al.: FLEXIBLE NON-CONSTRAINED RF WRIST PULSE DETECTION SENSOR BASED ON ARRAY RESONATORS
3
TABLE I SIMULATION RESULT OF THE RESONANT FREQUENCY VARIATION WITH ADJACENT SKIN MODEL DEFENDING ON
Fig. 5. Photograph of the flexible array resonator with 5 mm spacing.
Fig. 4. Simulated and measured S-parameter results of the single resonator with varying of the distance (0.25, 0.5, 0.75 mm).
clear that the resonant frequency is downshifted when the resonator is placed close to the human body. For the general rectangular PIFA structure ( in Fig. 2), 3D EM simulation results reveal a very good agreement with calculation results, as shown in Fig. 3. It shows that the PIFA structure can be analyzed using the superstrate method [19]. The frequency variation, as a function of the control variable , is calculated as summarized in Table I. In the table, refers to the resonance frequency with the resonator-to-skin distance of 0.25 mm, while refers to a corresponding distance of 0.75 mm. It should be noted that the resonance frequency is downshifted when becomes large, and the frequency deviation is maximized when is 3.8 mm. As shown in Fig. 2, the proposed resonator is simulated and compared to a rectangular one , as shown in Fig. 3. It was found that the proposed resonator reveals greater frequency variation than that of the rectangular shaped resonator, meaning that the proposed structure has sensitive characteristics for near-field sensing applications. Measurement results are shown in Figs. 3 and 4 as a function of the spacing between the resonator and skin, where layers of acrylic tapes (thickness: 62.5 m) are used to keep the separation constant. The measurement results show excellent correlation with the simulation results. B. Array Resonator On the wrist, a pulse can be easily detected on several locations where the arteries are found. However, since the shape and position of a wrist strap could be slightly different for different types of equipment and since a strap can rotate on the wrist during movement, the position of the resonator cannot be fixed. For a minimized form factor, the size of the proposed resonator is less than 20 mm, and near-field characteristics are utilized; therefore, the detection areas are relatively small with a single resonator. To overcome this issue, array resonator geometry is proposed in this work to extend the detection area to as large as possible. In addition, a multiple detection scheme utilizing the
array resonator are also proposed. As shown in Fig. 5, the proposed array resonators are placed 5 mm apart from one another, extending the detection coverage area to about 35 mm sufficient area to sense at least one artery of the wrist. Fig. 6 shows both the measured and the 3D modeled simulation results for four resonators as a function of resonator-to-skin distances ranging from 0.25 mm to 0.75 mm. Overall, the measured results agree reasonably well with the simulation results, and they reveal a clear distinction with distance and the possibility of versatile wrist pulse detection scheme. C. Switch With the Array Resonator The array resonator is connected to the detection system, with an absorptive single-pole, four-throw (SP4T) RF switch, and the switch is controlled by a 2-bit binary signal. The electrical spacing between the resonators (edge-to-edge) is about (5 mm) to ensure reasonable isolation between the resonators. Note that while such isolation is generally an important parameter in array antenna design, it might be relatively unimportant in the proposed system. It should be emphasized that the proposed detection system can be operated with one resonator at a time and that any leakage to the other resonators is absorbed by the absorptive off-state port of the switch. This absorptive characteristic provides a high isolation level between output ports of the switch and does not disturb impedance matching performance. III. SYSTEM CONFIGURATION In this Section, a sensitivity enhancement technique based on an active isolation technique is proposed. This technique allows detection of the reflected signal from the resonator with much higher sensitivity compared to using a system with a single detector. For the array resonators, proper analysis and specific conditions can be determined as follows. A. Operating Principle The reflected signal from the resonator can be expressed as [17]
(2) where is the reflection coefficient of the resonator, is the transmitted signal, is the constant component as a large signal of the reflection coefficient, and is the time-varying
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. 4
IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS
Fig. 7. Block diagram of the proposed system.
(3) is the initial where is the detector coefficient, and output voltage when the detector input is zero. To obtain the conversion gain slope of the detector, the output voltage in (3) is partially differentiated with
(4) The above (4) shows the sensitivity of the proposed system. should be minimized, thereby, the To increase the sensitivity, active isolator concept can be best suited increasing sensitivity by reducing the constant component of [17]. B. Array Detection System Configuration
Fig. 6. Simulated and measured S-parameter results of the array resonator when varying the distance (0.25, 0.5, 0.75 mm). (a) First resonator. (b) Second resonator. (c) Third resonator. (d) Fourth resonator.
pulse information as the small signal component, respectively. A coupler is utilized to detect the reflected signal. Therefore, the reflected signal is coupled to where is the coupling coefficient of the coupler. If using the RF detector to detect information, the output of the RF detector can be expressed as
A block diagram of the proposed system is shown in Fig. 7. In the figure, is transmitted into port 2 of the coupler and most of the signal reaches the RF switch which is connected to port 1 of the coupler. Then the RF signal is relayed to each resonator, depending on the RF switch status. As mentioned in Section II, the resonator elements of the array have a reflection coefficient dependent on the near-field variation due to physiological movement. The reflected signal can be expressed as “ “ “ “
” ” ” ” (5)
where is the reflection coefficient of the resonator elements, and is the 2 bit binary signal to control the RF switch. Now, the reflected signal from the resonator is coupled to port 3 of the
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. AN et al.: FLEXIBLE NON-CONSTRAINED RF WRIST PULSE DETECTION SENSOR BASED ON ARRAY RESONATORS
5
coupler and becomes . In the same way, the reflected signal is coupled to port 4 and can be expressed as . For phase synchronization a phase shifter is used on the reflected signal, and the signal becomes , where is the shifted phase. In addition, a variable attenuator is utilized to adjust the magnitude of the transmitted signal at port 4 and the adjusted signal becomes . These two signals are combined at the power combiner, and the combined signal can be expressed as follows:
(6) The output of the detector is now differentiated with respect to to obtain the sensitivity of the system, as follows:
(7) To increase the slope, that is, the sensitivity of the combined signal, the denominator in (7) should be as small as possible. For sensitivity enhancement, the phase shifter and the variable attenuator should be properly set as
(8) It is necessary to note that the reflection coefficient of each resonator is slightly different as shown in Fig. 6. As a result, these phase and magnitude differences can degrade the detection sensitivity. Fig. 8, depicts the calculated results based on (7). The figure shows that the sensitivity enhancement curves are dependent on the magnitude and phase differences, and reveals maximum sensitivity when the ideal conditions are satisfied as in (8). It is clear that the sensitivity can be increased about 40 dB as shown in Fig. 8. Note that the amount the sensitivity enhancement varies with respect to both the magnitude ratio of the reflected and the attenuated signal and the phase difference . As shown in Fig. 8(a), if the phase difference is fixed to 180 degrees, the sensitivity enhancement is symmetrically decreased, while is shifted away from the highest sensitivity enhancement value at the center point . However, when the phase difference is shifted from 180 degrees, the center point of the symmetry is shifted to . The center point is determined by using , consequentially the sensitivity enhancement region is largely spreaded on the condition. Fig. 8(b) shows that with a 20% magnitude difference between two signals, the sensitivity can be increased at least 7 dB, where the phase difference is near 180 degree. In a case where
Fig. 8. (a) Calculated result of the sensitivity enhancement depending on the magnitude and phase difference. (b) Calculated results about sensitivity is 0.8, 1, and 1.2. (c) Measured phase of and enhancement where on 2.42 GHz. phase difference with the other resonators
(
in Fig. 8), the sensitivity also increases to near points. therefore, when using a smaller magnitude of than , the sensitivity enhancement range is extended on the phase difference domain in Fig. 8(a). This means that if the condition of the proposed array system is set to , some of the phase mismatch can be accepted for sensitivity enhancement. Also, the measured phase results are shown in Fig. 8(c), it is show that the phase variances of other resonators are less than 60 degrees except in 0.75 mm case. According to Fig. 8(a), at least 10 dB sensitivity could be increased where with measured phase differences. Therefore, the proposed array detection system can provide high sensitivity without controlling the phase shifter and variable attenuator for each resonator. A photograph of the fabricated circuit is shown in Fig. 9. The overall size of the circuit is 30 mm 30 mm. Note that a
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. 6
IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS
Fig. 11. Measurement setup of the proposed wrist pulse detection system.
Fig. 9. Photograph of the proposed wrist pulse detection circuit.
prototype circuit, a power consumption of utilized the RF amplifier and the power detector is not considered. Supplied power is 5 V with 45 mA (RF amp: 20 mA, power detector: 25 mA), it could be reduced with the low-powered components. IV. EXPERIMENTAL RESULTS
Fig. 10. Photograph of the equipment of the array resonator for the wrist pulse detection.
thin film directional coupler (AVX C0402) is utilized to minimize the size of the circuit, and an SP4T absorptive switch (SKY13384) controlled by 2-bit digital signal is employed. For the connection between the RF switch and the array resonator, flexible coaxial cables are employed, as shown in Fig. 10. To control the amplitude , an analog voltage controlled variable attenuator (RF2013A) is utilized, and varactor diodes are used for the voltage controlled phase shifter [22]. A chip power combiner device based on the Wilkinson power divider is used to combine signals, and an RF amplifier (ABA51563) is connected to the output of the power combiner for amplification. In the last stage a power detector (LT5538) detects the output power of the combined signal, containing the wrist pulse information. As a
To detect a heart pulse on the wrist, the array resonator is placed around the wrist. In a previous study [18], since the pulse can be detected directly above the radial artery, the position of the first resonator element (S1 in Fig. 10) is placed close to the radial artery. The remaining resonator elements are spaced at 5 mm. Therefore, the fourth resonator is placed near the ulnar artery. To keep the gap between the resonator and the skin constant, acrylic tape and styrofoam are employed as a preventive layer as well as a spacer. The gap is kept at about 1 mm during the experiment, as shown in Fig. 10. The experiment setup is shown in Fig. 11. A signal generator is used as a signal source and is set to 2.42 GHz with 0 dBm output power. The resonator is connected to the circuit using a flexible coaxial cable, as shown in Fig. 10. As a reference, a piezoelectric pressure sensor is fastened on a finger. The output of the fabricated sensor circuit and the reference signal are recorded by a 24-bit NI DAQ. Note that the RF switch control bit signals are generated by a clock pulse generator, which is controlled by a hard-coded program, and these control bits are recorded by the DAQ board to monitor the switch status. The experimental environments and conditions as follows: First, to demonstrate the capabilities of the proposed sensor, motion artifact is restricted during the experiment. Second, a 2.4 GHz WiFi access point (AP) operates nearby place as a noise source, where the power of the noise signal is measured about dBm. First, a sequential detection is performed; the measurement results of which are shown in Fig. 12. The switch control binary bits are changed every 5 seconds with a detection sequence of S3, S2, S1, and S4, as illustrated in Fig. 12(a). In Fig. 12(b), the detector output signals are divided into each switching state and low pass filtered with 30 Hz cut-off frequency to reduce the 60 Hz noise produced by the power supply. It is clear that the signals from resonators S1 (10–15 sec), S2 (5–10 sec) and S4 (15–20 sec) show reasonably good heartbeat waveforms. As expected, the detected pulses from the radial and the ulnar arteries reveal higher vital signal levels. It is interesting to observe that
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. AN et al.: FLEXIBLE NON-CONSTRAINED RF WRIST PULSE DETECTION SENSOR BASED ON ARRAY RESONATORS
7
Fig. 12. Measurement result of the proposed wrist pulse detection system. (a) Reference signal and detection signal switching the array resonator with every 5 seconds. (b) Low pass filtered and magnified detection signals, each state shows different result.
the pulses from the radial artery clearly reveal second and third peaks that are reflected from the lower body (hand). The time delay and amplitude ratio between the first and second peaks can be utilized to extract various medical and physiological information such as blood pressure and arterial stiffness [8]. In the previous PPG study it was shown that detecting the pulse in the middle of the wrist [11] is possible, and this technique is similar to our S2 position, but with much less signal power. In the second experiment, simultaneous detection is performed with the array resonator. The switch control binary bits are changed every 25 milliseconds so that each resonator is selected 10 times/sec. To reduce the complexity of signal processing, an identical 10 sample/sec sampling rate is used in this work. The measurement results are shown in Fig. 13 with the reference data of Fig. 13(a). It is clear that all 4 resonators can detect a heartbeat pulse with different signal levels, while S1 and S4 reveal excellent pulse waveforms. To demonstrate this in detail, all the sampled signals are interpolated with a sinc function. In accordance with the Nyquist theorem, signals detected via the 10 samples/sec per channel can be recovered up to 5 Hz, and this renders to resolve the second peak in the
Fig. 13. Measurement result with simultaneous switching at 10 samples/sec. (a) Reference pulse signal. (b) Sampled data from the proposed system. Detection signal with interpolation from (c) the first resonator, (d) the second resonator, (e) the third resonator, and (f) the fourth resonator.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. 8
IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS
pulse from the sampled result. Interpolated data sets are shown in Fig. 13(c), (d), (e), and (f). It should be noted that the second peak can be observed in most pulse waves at S4 as shown in Fig. 13(f), and this is very similar to ballistocardiogram (BCG) measurement reading. This information could be utilized as the main diagnosis factor for artery stiffness [8]. If all the waveforms can be combined together, it would be possible to extract clinically important information. For now, it is clear that the array resonator can detect heartbeat pulse on a wrist at multiple locations. V. CONCLUSION This paper has presented a non-constrained radio frequency wrist pulse detection system based on reflection coefficient variation due to blood flow in the radial and ulnar arteries. The proposed system utilizes an array of resonators, which were designed and fabricated on a flexible teflon substrate with a human skin model on top of the resonators. The measured results for a single resonator and an array of resonators were well matched to the simulation results. In addition, an RF switch and a pulse detection circuit were designed to realize the proposed sensitivity enhancement technique, and the phase and magnitude mismatch effect with the array resonator was analyzed. It is clear that the sensitivity of the system is strongly dependent on the value of the phase shifter as well as that of the variable attenuator. However, in some condition, the proposed system can be used with multiple resonators without changing the conditions of the phase shifter or the variable attenuator for each resonator. The sensor system was tested with two different switching modes and demonstrated successful detection of a wrist pulse signal at multiple locations on the wrist. This shows that the proposed system is location insensitive and can be placed on somewhat arbitrary locations on the wrist. It is clear that the proposed non-constrained RF vital sign sensor has a great advantage in that it can be integrated into wearable communication devices, such as a smart watch, with biomedical and healthcare applications. REFERENCES [1] Y.-L. Zheng, X.-R. Ding, C. C. Y. Poon, B. P. L. Lo, H. Zhang, X.-L. Zhou, G.-Z. Yang, N. Zhao, and Y.-T. Zhang, “Unobtrusive sensing and wearable devices for health informatics,” IEEE Trans. Biomed. Eng., vol. 61, no. 5, pp. 1538–1554, May 2014. [2] G. Cotter, Y. Moshkovitz, E. Kaluski, A. J. Cohen, H. Miller, D. Goor, and Z. Vered, “Accurate, noninvasive continuous monitoring of cardiac output by whole-body electrical bioimpedance*,” Chest, vol. 125, pp. 1431–1440, 2004. [3] J. Foo and C. Lim, “Pulse transit time based on piezoelectric technique at the radial artery,” J. Clin. Monit. Comput., vol. 20, pp. 185–192, 2006. [4] V. Almeida, H. C. Pereiraa, T. Pereiraa, E. Figueirasa, E. Borgesa, J. M. R. Cardosoa, and C. Correiaa, “Piezoelectric probe for pressure waveform estimation in flexible tubes and its application to the cardiovascular system,” Sens. Actuators A, Phys., vol. 169, no. 1, pp. 217–226, 2011. [5] G. Schwartz, B. C. K. Tee, J. Mei, A. L. Appleton, D. H. Kim, H. Wang, and Z. Bao, “Flexible polymer transistors with high pressure sensitivity for application in electronic skin and health monitoring,” Nat. Commun., vol. 4, p. 1859, 2013. [6] S. Gong, W. Schwalb, Y. Wang, Y. Chen, Y. Tang, J. Si, B. Shirinzadeh, and W. Cheng, “A wearable and highly sensitive pressure sensor with ultrathin gold nanowires,” Nat. Commun., vol. 5, 2014. [7] C. Hlenschi, S. Corodeanu, and H. Chiriac, “Magnetoelastic sensors for the detections of pulse waves,” IEEE Trans. Magn., vol. 49, pp. 117–119, 2013.
[8] W. W. Nichols, “Clinical measurement of arterial stiffness obtained from noninvasive pressure waveforms,” Amer. J. Hypertens., vol. 18, pp. 3S–10S, Jan. 2005. [9] P. Renevey, R. Vetter, and J. Krauss, “Wrist-located pulse detection using IR signals, activity and nonlinear artefact cancellation,” in Proc. 23rd IEEE Engineering in Medicine and Biology Soc. Conf., Istanbul, Turkey, 2001, pp. 3030–3033. [10] E. Geun, H. Heo, K. C. Nam, and Y. Huh, “Measurement site and applied pressure consideration in wrist photoplethysmography,” in Proc. 23rd Int. Tech. Conf. Circuits/Systems, Computers and Communications, 2008, pp. 1129–1132. [11] K. Li and S. Warren, “A wireless reflectance pulse oximeter with digital baseline control for unfiltered photoplethysmograms,” IEEE Trans. Biomed. Circuits Syst., vol. 6, no. 3, pp. 269–278, Jun. 2012. [12] A. A. Serra, P. Nepa, G. Manara, G. Corsini, and J. L. Volakis, “A single on-body antenna as a sensor for cardiopulmonary monitoring,” IEEE Ant. Wireless Propag. Lett., vol. 9, pp. 930–933, 2010. [13] S. G. Kim, G. H. Yun, and J. G. Yook, “Compact vital signal sensor using oscillation frequency deviation,” IEEE Trans. Microw. Theory Tech., vol. 60, pp. 393–400, 2012. [14] S. G. Kim, G. H. Yun, and J. G. Yook, “Wireless RF sensor structure for non-contact vital sign monitoring,” J. Korean Inst. Electr. Eng. Sci., vol. 12, pp. 37–44, 2012. [15] Y. Hong, S. G. Kim, B. H. Kim, S. J. Ha, H. J. Lee, G. H. Yun, and J. G. Yook, “Noncontact proximity vital sign sensor based on PLL for sensitivity enhancement,” IEEE Trans. Biomed. Circuits Syst., vol. 8, no. 4, pp. 584–593, 2014. [16] Y.-J. An, G.-H. Yun, and J.-G. Yook, “Vital sign detection for handheld communication device using antenna mismatching effect,” in Proc. IEEE MTT-S Int. Microw. Symp. Dig., Jun. 2–7, 2013. [17] Y.-J. An, G.-H. Yun, and J.-G. Yook, “Sensitivity enhanced vital sign detection based on antenna reflection coefficient variation,” IEEE Trans. Biomed. Circuits Syst., 2014, submitted for publication. [18] Y.-J. An, G.-H. Yun, S. W. Kim, and J.-G. Yook, “Wrist pulse detection system based on changes in the near-field reflection coefficient of a resonator,” IEEE Microw. Wireless Compon. Lett., 2014, accepted for publication. [19] J. T. Bernhard and C. J. Tousignant, “Resonant frequencies of rectangular microstrip antennas with flush and spaced dielectric superstrates,” IEEE Trans. Antennas Propag., vol. 47, pp. 302–308, Feb. 1999. [20] I. J. Bahl, P. Bhartia, and S. Stuchly, “Design of microstrip antennas covered with a dielectric layer,” IEEE Trans. Antennas Propag., vol. 30, pp. 314–318, Mar. 1982. [21] S. Gabriel, R. W. Lau, and C. Gabriel, “The dielectric properties of biological tissues: II. Measurements in the frequency range 10 Hz to 20 GHz,” Phys. Med. Biol., vol. 41, pp. 2251–2269, 1996. [22] H. Kim, A. B. Kozyrev, A. Karbassi, and D. W. van der Weide, “Linear tunable phase shifter using a left-handed transmission line,” IEEE Microw. Wireless Compon. Lett., vol. 15, no. 5, pp. 366–368, May 2005.
Yong-Jun An (S’12) was born in Seoul, Korea. He received the B.S. degree in electrical and electronic engineering from Yonsei University, Seoul, Korea, in 2009. Currently, he is working toward the Ph.D. degree in electrical and electronic engineering at Yonsei University. His research interests include remote wireless vital signal monitoring sensors, RF passive circuits, and radars.
Byung-Hyun Kim (S’12) was born in Incheon, South Korea. He received the B.S. degree in electrical and electronics engineering from Yonsei University, Seoul, South Korea, in 2012. Currently, he is working toward the Ph.D. degree in electrical and electronics engineering at Yonsei University. He received the Spring 2012 Undergraduate-Pre-graduate (BS/MS) Scholarship from the IEEE Microwave Theory and Techniques Society (MTT-S). His main research interests are in the microwave/millimeter-wave component systems, remote wireless vital signal monitoring, RF biosensors, RF passive circuits, and radars.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. AN et al.: FLEXIBLE NON-CONSTRAINED RF WRIST PULSE DETECTION SENSOR BASED ON ARRAY RESONATORS
Gi-Ho Yun was born in Jeonju-Si, Korea. He received the B.S., M.S., and Ph.D. degrees in electronics engineering from Yonsei University, Seoul, South Korea, in 1984, 1986, and 1999, respectively. From 1985 to 1997, he worked at Samsung Electronics and Samsung Electro-Mechanics. He also served at Honam University, Gwangju, South Korea, from 1997 to 2008. Currently, he is an Assistant Professor in the School of Information and Communication Engineering, Sungkyul University, Kyeonggi-Do, Korea, and also working with the Advanced Computational Electromagnetic Laboratory of Yonsei University. His research interests include radio frequency (RF) circuits, patch antennas, and biosensors.
Sung-Woo Kim was born in Incheon, South Korea. He received the B.S. degree in biomedical engineering from Yonsei University, Wonju, South Korea, in 2004. From 2004 to 2008, he was a Research Assistant in the Department of Medical Engineering, Yonsei University College of Medicine, Severance Hospital, Seoul, South Korea. Currently, he is working toward the Ph.D. degree in biomedical engineering at Yonsei University. He is a Senior Research Engineer with the Biometrics team, Future IT R&D Laboratory, LG Electronics Advanced Research Institute, Seoul, South Korea. His main research interests include biosignal processing, biosensors such as PPG/IPG, and healthcare using the smart watch.
9
Seung-Bum Hong was born in Seoul, South Korea. He received the B.S. and M.S. degrees in electronics engineering from Yonsei University, Seoul, South Korea, in 1999 and 2001, respectively. Currently, he is a Principal Research Engineer with the Biometrics team, Future IT R&D Laboratory, LG Electronics Advanced Research Institute, Seoul, South Korea, and is an advisory committee member with the KFDA. His main research interests include biosignal processing, biosensors, and the mHealth project.
Jong-Gwan Yook (S’89–M’97–SM’12) was born in Seoul, South Korea. He received the B.S. and M.S. degrees in electronics engineering from Yonsei University, Seoul, South Korea, and the Ph.D. degree from the University of Michigan, Ann Arbor, MI, USA, in 1987, 1989, and 1996, respectively. Currently, he is a Professor with the School of Electrical and Electronic Engineering, Yonsei University. His main research interests are in the areas of theoretical/numerical electromagnetic modeling and characterization of microwave/millimeter-wave circuits and components, design of radio frequency integrated circuits (RFICs) and monolithic microwave integrated-circuits (MMICs), and analysis and optimization of high-frequency high-speed interconnects, including signal/power integrity (EMI/EMC), based on frequency as well as time-domain full-wave methods. Recently, his research team developed various biosensors, such as carbon-nano-tube RF biosensors for nanometer size antigen-antibody detection as well as remote wireless vital signal monitoring sensors.
