Exquisite Textiles Sensors and Wireless Sensor Network Device for ...

30th Annual International IEEE EMBS Conference Vancouver, British Columbia, Canada, August 20-24, 2008

Exquisite Textiles Sensors and Wireless Sensor Network Device for Home Health Care Wen-Tzeng Huang, Chin-Hsing Chen, Yuan-Jen Chang, You-Yin Chen, Jung-Lin Huang, Chang Ming Yang, and Tzu Lin Yang Abstract In this study, we propose a wearing system with four sensors, ECG (electrocardiogram), three-axis accelerometer, temperature, and tight-switch, applied for remote monitoring system in home-care. The sensors ECG, measured with wearable electrodes made of the steel textile to generate the real-time heart-rate estimator, tight-switch, made of the steel textile to check whether wearing person dresses properly, accelerometer, and temperature parameters are received via the ZigBee receiver within an exquisite belt. Since the movable textile electrodes will cause of unfixed contacts when the wearing person is in motion, making the heart-rate estimation much a sophisticated work, the tight-switch sensor and FIR (Filter Impulse Response) filter technology are applied here to get the more satisfiable heart-rate. The other two bio-sensors can detect the whether fall-down or not and normal body-temperature of this wearing person. Moreover, the ZigBee device with small size, low-power consumption, and high-reliability characteristics is designed to transmit the detected bio-information from these four sensors. Therefore, the vital system embedded with the capability of real-time heart-rate estimation and transmission makes it highly suitable for applications of remote healthcare and wellness. I. Introduction A rapid trend of interest in new sensing and monitoring devices is for healthcare during these years. Since the world’s aged population is growing, the need for a comfortably wearable system with capability to measure and wireless transmit vital signals is becoming more important [1]. The wearable sensors especially made of textiles with suitable materials have been flourishing recently. Manuscript revised June 26, 2008. Wen-Tzeng Huang (Corresponding author), Associate Professor of Department of Computer Science and Information Engineering, Minghsin University of Science and Technology. (e-mail: [email protected]) Chin-Hsing Chen, Associate Professor of Department of Management Information Systems, Central Taiwan University of Science and Technology, Taiwan. (e-mail: [email protected]) Yuan-Jen Chang, Assistant Professor of Department of Management Information Systems/3Institute of Biomedical Engineering and Material Science, Central Taiwan University of Science and Technology, Taiwan. (e-mail: [email protected]). You-Yin Chen, Assistant Professor of Department of Electrical and Control Engineering, National Chiao Tung University, Taiwan. (e-mail: [email protected]) Jung-Lin Hunag, Graduate student of Department of Electronic Engineering, National Taipei University of Technology, Taiwan. (e-mail: [email protected]) Chang Ming Yang, and Tzu Lin Yang are with Ming Young Biomedical Corp. No. 27, Guangfu Rd., Jhunan Miaoli 350, Taiwan. (WWW: http://www.my-cares.com, e-mail: [email protected])

978-1-4244-1815-2/08/$25.00 ©2008 IEEE.

546

For instance, the usage of wearable medical bent for monitoring respiration activity ECG signal [2] and home-care has been proposed. Studies for developing wearable medical devices have been started such as the VTAMN project in France, the WEALTHY project in Europe, and the Life Shirt in USA [3]. At present, using conventional sensor techniques as monitoring component often cause skin irritation and hampering wires. Skin irritation and allergic contact dermatitis have been reported, caused by the adhesive part and the gel used in conventional electrocardiograph electrodes [4]. Moreover, for long-term physiological monitoring inside or outside a hospital setting, a reliable, wearable monitoring system would be a convenient platform if biomedical sensors are securely placed in appropriate positions. An article of clothing is an attractive platform to implement such a wearable system. It is highly desirable that the sensors be designed and integrated into the garment in an unobtrusive way. Although several such items of wearable electronic textile technology have been developed, it still confronts the great challenge to obtain new differentiated constructions which could find application as reliable sensors for monitoring heartbeat. Firstly, biomedical sensors serve as front-end elements of this wearable health monitoring system, and as such are directly, indirectly or invasively placed into contact with the human body to measure its physiological status. These sensors must be significantly different from the currently in market, conventional, bulky and rigid versions. The most desirable sensor forms for the wearable health monitoring system are fabrics/textiles themselves, i.e. textile-based sensors. We look forward to designing a wear is possible and easy to utility. Traditionally, modern electro-conductive textiles were used to construct textile sensors characterized by the changes of resistance under deformations but the resistance change by textile elongation is small [5]. Our target is to design the textile sensors overcoming the problem above easily integrated into one exquisite bent. Then, the received data can be further processed to extract more meaningful information. In this paper, we apply the well-known the FIR filters to accurately estimate the heart rate. Simulations demonstrate that this algorithm can successfully estimate the heart rate and its variability from the received ECG data when the wearing person begins with sitting still to walking through jogging. We had presented the ZigBee system to be applied in the home-care system [6].

when the system is in the normal operation case, this indicator is on; the other case is in the abnormal case. System timer: when the system starts, this system timer is displayed for recording the tag of bio-information. HR (BPM) (heart-rate by beat per minute) : the heart-rate is detected from the R wave signal of the sensor ECG, measured with wearable electrodes made of the steel textile to generate real-time heart-rate estimator [7]. Also, the number of this heart-rate is related to ECG information. Body-temperature: when wearing person wears this bio-bent, the system will send out the alarm news while one’s temperature is abnormal state. Three-axis accelerations: this three-axis accelerometer is used to measure the motions of wearing person whether one has already fallen or not. This fall information will send out the urgent signal promptly to attendants.

II. The System Structure The structure of our presented system is divided into two parts, the textile bent knitted with vital sensors and a ZigBee control box, namely Device [6], used for signal integration, receiving, and transmission as shown in Fig. 1(a). The belt contains two ECG sensor electrodes, a temperature sensor, and a tight-switch sensor all sewn in the belt such that the wearing person can feel comfortable shown in Fig 1(b). The device includes signal processing circuits, a three-axis accelerometer, and a ZigBee transceiver; the signal processing circuits contains some filters, the three-axis accelerometer is used to measure the motions of the wearing person, and ZigBee transceiver is to receive remote signals as well as to transmit the data obtained from various sensors to remote mobile phones or computers for further applications.

(a)

Fig. 2. Remote display interface for typically measured data. Then, discussing the sensor tight-switch, the arc shape sensor tight-switch, which pulling force is over 500 pounds to be on, of the belt as shown in Fig. 1(b) is designed here which its base part is no woven textile and the upper arc shape made of SBR (Styrene); each side of contact face is knitted a piece of electronic conductive fiber as shown in Fig. 3 (a); its structure is shown as Fig. 3 (b)(c).

(b) Fig. 1. (a) Structure of wireless nodes with vital belt and (b) Belt with sensors. In this study, items in the remote display interface for typically measured data are shown in Fig. 2. Although there are just 2 input signals and 1 floating reference with high impendence, our designed textile bent is not belong to one type of standard leads. Then, ECG indicator: when the system get the ECG signal and then analyses it, one ECG indicator indicate the input signal as the on state. ECG data: this ECG data, measured with wearable electrodes made of the steel textile to generate real-time heart-rate estimator, indicates the electric signal of one’s heart state. Tight-switch indicator: if this tight-switch indicator is on, it indicates that wearing person dresses properly. In this on case, the system can obtain better signal than the other case, which tight-switch indicator is off state. System emergency button: while abnormal operations happen in the signal or the system, in order to avoid sending out the wrong message, one pushes this emergency knob such that it can stop the system operation. System operation indicator:

Fig. 3. (a) Real sample of the arc sensor tight-switch, (b) Structure of the arc sensor tight-switch, (c) Conductive fiber marked by the red line. The ECG sensor of the belt as shown in Fig. 1(b) in this system has two electrodes realized by the steel textile. The signals from the electrodes are passed through filters with their specifications conformed to the American National Standard [8]. The filtered signals are sampled via an analog/digital converter at the rate of 200 samples per second and then transmitted by ZigBee to the remote computer. In other words, the sampling rate of ECG is 1/200 second. Each ECG pixel to be drawn one scale on the 547

degree of bent elasticity is in the moderate case, a better ECG signal can be measured. From our results as shown in Figs. 4 (a) and (b), we know that the ECG measured signal quality of textile ECG sensor with tight-switch sensor as shown in Fig. 4(a) is better than that of textile ECG sensor without tight-switch sensor as shown in Fig. 4(b). Moreover, we learn that the performance of textile ECG sensor with tight-switch sensor is near equal to the conventional ECG electrodes as shown in Fig. 4(c). Moreover, since FIR can remove the base drift on the ECG signal as shown in Fig. 4(a) and then get the more satisfiable heart-rate in this design as shown Fig. 4(d).

axis of time, every 50 pixels is to draw one thick mark. Then, the three-axis accelerometer integrated circuit is mounted on the control box which normally is carried along the waist. The body temperature sensor is obtained via a thermistor knitted in the bent and connected using the steel fiber to the control box. The characteristics of our designed ZigBee sensor node are small size, low-power consumption, high-reliability [6], 18 M transmission distances, multiple-sensor inputs as shown in Fig. 1(c). III. Post FIR Filter Signal Processing The purpose of post FIR signal processing is to use the measured data for extracting meaningful information or knowledge which can either help doctor diagnose or judge the physical status of the wearing person whether calling emergency help is needed. This system has been equipped with signal processing capabilities to obtain the heart rate from the ECG. The system function of a FIR (Filter-Infinite Impulse Response) filter is shown in Eq. (1). Let H(z) be the transferring function, bM-1 be the Filter coefficients, and z1-M be the Z transfer function with (1–M) time delay [9][10]. And, its impulse response h(n) is shown as Eq. (2). M −1

H ( z ) = b0 + b1 z −1 + ... + bM −1 z1− M = ∑ bn z −n

Fig. 4. In rest case, (a) textile ECG sensor with tight -switch sensor, (b) ECG sensor without tight -switch sensor, (c) conventional ECG electrodes, (d) Post FIR filter processing. When the wearing person walks at the speed of 0.8m per second, the 4 experiments, which are similar to the rest cases, of ECG signal are measured and shown in Figs. 5(a)-(d), respectively.

(1)

n =0

⎧b ,0 ≤ n ≤ M − 1 h( n) = ⎨ n ⎩0, else

( 2)

While, the difference equation y(n) is shown as Eq. (3).

y ( n) = b0 x( n) + b1 x ( n − 1) + ... + bM −1 x ( n − M + 1)

(3)

Then, from above equations result of filter design, the two filters, high-pass filter with over 5 Hz and band-pass filter with between 1 Hz to 40 Hz, are implemented here. Therefore, after the base drift of an ECG signal with passes through this high-pass filter and then passes again through a band-pass filter, this base drift of ECG signal can be removed. That is, this high-pass filter can get rid of the electric signal of muscle-electric signal; and then, the high-pass filter can keep the wanted ECG signal. The experiments are presented to show the comparisons between four results, textile ECG sensor with tight-switch sensor, ECG sensor without tight-switch sensor, conventional ECG electrodes, and post FIR processing. The rest experiments shown in Figs. 4(a)-(d), respectively. The bent elasticity degree, which causes of unfixed contacts, of the wearing persons will affect the measuring ECG signal. 500 pounds of tight-switch is designed here to control the ECG sensors to be under the fixed contacts and in the better tested condition. When the body of wearing person presses the tight-switch of the bent over 500 pounds, then, ECG signal will be starting to measure. In this case, since the

Fig. 5. In walking case, (a) Textile ECG sensor with tight-switch sensor, (b) ECG sensor without tight-switch sensor, (c) Conventional ECG electrodes, (d) Post FIR filter processing.

Fig. 6. In jogging case, (a)Textile ECG sensor with tightswitch sensor, (b)ECG sensor without tight- switch sensor, (c)Conventional ECG electrodes, (d) Post FIR filter processing. There is high vibration, which cause more shift and large amount of low-frequency muscle-electric signal, in the jogging case. This signal will interfere ECG tested signal and easy cause to get the mistake signal. Similarly, 4 experiments are shown in the Fig. 6 (a)-(d), respectively. 548

When the person begins jogging, however, the heart rate is increased tremendously and the FIR filter algorithm can successfully obtain estimation of heart rate. IV. Wireless Sensor Performance Evaluation The major function of this structure is finished within the real-time transmission. Let Coordinator allocate the transmission space. When Coordinator has the ability to receive the data, it will send a command CMD to Device, which can send the data now. Then, Gateway sends this data from Coordinator out. A completed transmission cycle takes 16.5ms shown in Fig. 1. Coordinator takes 4ms to send a CMD to Device; Device takes 4 ms and 4.5 ms to response and then transmits an ACK and data back to Coordinator, respectively; finally, Coordinator takes 4ms to send an ACK to Device. Hence, there are about 60 packages within one second, and the transmission rate is about 38 Kbps (bit per second). According to the beaconless mode [11], its ideal transmission rate is about 140 Kbps without any ACK. In our design, one data takes 4.5 ms to send, and our transmission rate of full speed is about 142 Kbps in conformity with the beaconless mode [11]. In the real design here, the transmission rate is shown in Eq. (4) to send one completed data, which costs each CMD, ACK, Data, and ACK time, respectively, namely Tcmd, Tack, Tdata, and Tack, respectively. Therefore, the transmission rate of our design is about 38 Kbps. Let SysSpeed be the real transmission speed, Tcmd be the CMD time, Tdata be the physical data, and FullSpeed be 142 Kbps.

Fig. 7. Schematic of correct rate transmitted rate. steadily to jogging, demonstrate that the presented method obtains heart-rate estimates which satisfactorily reflect the motion status of the person. Moreover, about the ZigBee system with small size, low-power consumption, and high-reliability characteristics, the package correct transmitted rate is more than 97.5% and simultaneous six devices with these four sensors under the full speed when the transmitting distance is smaller than 18 M. Hence, this system can be used for remote health monitoring and diagnosis. This system is especially useful for reducing the incidence of life-style related and chronic diseases; it also can be used for self-care and management of patient’s health. Reference [1]

S. Park and S. Jayaraman, “Enhancing Quality of Life through Wearable Technology”, IEEE Eng. In Medicine and Biology Magazine, pp. 41-48, May/June 2003. [2] R. Paradiso, A. Gemingnani, E. P. Scilingo, and E. De Rossi, “Knitted bioclothes for cardiopulmonary- monitoring,” Proc. of 25th Int. Conf. IEEE EMBS, pp. 3720-3723, 2003. [3] F. Axisa, P. M. Schmitt, C. Gehin, G. Delhomme,E. McAdams, and A. Dittmar, “Flexible technologiesand smart clothing for citizen mdeicine, home healthcare,and disease prevention,” IEEE Trans. InformationTechnology in Biomedicine, Vol. 9, No. 3, pp. 325-336, Sept. 2005. [4] M. Avenel-Audran, A. Goossens, E. Zimmerson, M.Bruze, Contact dermatitis from electrocardipgraph-monitoring electrodes: role of p-tert-butylphenol-formaldehyde resin, Contact Dermatitis 48, pp.108-111, 2003 [5] M. Catrysse, R.Puers, C. Hertleer, L. Van Langenhove, H.van Egmond and D. Matthys, “Towards the Integration of Integration of Textile Sensors in a Wireless Monitoring Suit” 2004 Elsevier B.V. pp.302-311. [6] C. H. Chen, Y. J. Chang, Y. Y. Chen, W. T. Huang, W. H. Chang, J. J. Chen, and C. Y. Kuo, “"Guardian"- The Study of Construction Wireless Sensor Network for Remote Monitoring System in Home Care”, Information Education and Technological Applications Conference, IETAC2007, Nov. 2007. [7] N. V. Thakor, J. G. Webster, and W. J. Tompkins, “Estimation of QRS complex power spectra for design of QRS filter,” IEEE Trans. Biomed. Eng., vol. BME-31, pp. 702-706, Nov. 1984. [8] American National Standard, Cardiac monitors, heart rate meters, and alarms, ANSI/AAMI EC 13:2002. [9] A. Van, J. A.; Schilder, T. S, “Removal of Base-Line Wander and Power-Line Interference from the ECG by an Efficient FIR Filter with a Reduced Number of Taps,” on Biomedical Engineering, IEEE Transactions on Volume BME-32, Issue 12, Dec. 1985, pp. 1052-1060. [10] W. Philips,“Adaptive noise removal from biomedical signals using warped polynomials”, on Biomedical Engineering, IEEE Trans. Vol. 43, Issue 5, May 1996, pp:480- 492. [11] T. Sun, L. J. Chen, C. C. Han, G. Yang, and M. Gerla, “Measuring effective capacity of IEEE 802.15.4 beaconless mode”, Wireless Communications and Networking Conference, WCNC 2006, Vol. 1, pp:493-498.

Tdata ∗ FullSpeed (4) Tcmd + Tack + Tdata + Tack Since the transmission data of the two sensors, ECG and three-axis accelerometer, is combined into one, there are 200 data sets. That is, its transmission rate is 200 set*4 (B/set) * 8 (bits/B)=6.4 Kbps. Since there is one data set rate per second to represent the body temperature and tight-switch, its transmission rate is 0.032 Kbps (1 set*4 (B/set)*8 (bit/B)=32 bits). Since the maximum transmission rate of our system is 38.79Kbps, there are six devices with these four sensors under the full speed. It shows that the package correct transmitted rate is more than 97.5% when the transmitting distance is smaller than 18 M. SysSpeed =

V. Conclusion A vital wearing system is presented in this paper. The exquisite bent is sewn with steel textile for realizing ECG sensors, tight-switch sensor, and temperature. In addition to the bent, a small control box with mounted a three-axis accelerometer and a ZigBee transceiver such that the posture of wearing person can be detected and wireless transmission of data is realized. Experiments for ECG data from the person with initially sitting still through walking 549