398
www.DownloadPaper.ir
IEEE SYSTEMS JOURNAL, VOL. 3, NO. 4, DECEMBER 2009
Wireless Body Sensor Network With Adaptive Low-Power Design for Biometrics and Healthcare Applications Shih-Lun Chen, Ho-Yin Lee, Chiung-An Chen, Hong-Yi Huang, Member, IEEE, and Ching-Hsing Luo, Member, IEEE
Abstract—A four-levels hierarchical wireless body sensor network (WBSN) system is designed for biometrics and healthcare applications. It also separates pathways for communication and control. In order to improve performance, a communication cycle is constructed for synchronizing the WBSN system with the pipeline. A low-power adaptive process is a necessity for long-time healthcare monitoring. It includes a data encoder and an adaptive power conserving algorithm within each sensor node along with an accurate control switch system for adaptive power control. The thermal sensor node consists of a micro control unit (MCU), a thermal bipolar junction transistor sensor, an analog-to-digital converter (ADC), a calibrator, a data encoder, a 2.4-GHz radio frequency transceiver, and an antenna. When detecting ten body temperature or 240 electrocardiogram (ECG) signals per second, or 220.4 . By the power consumption is either 106.3 switching circuits, multi sharing wireless protocol, and reducing transmission data by data encoder, it achieves a reduction of 99.573% or 99.164% in power consumption compared to those without using adaptive and encoding modules. Compared with published research reports and industrial works, the proposed method is 69.6% or 98% lower than the power consumption in thermal sensor nodes which consist only of a sensor and ADC (without MCU, 2.4-GHz transceiver, modulator, demodulator, and data encoder) or wireless ECG sensor nodes which selected Bluetooth, 2.4-GHz transceiver, and Zigbee as wireless protocols.
W
W
Index Terms—Adaptive low power, biometrics, wireless body sensor network, wireless ECG sensor, wireless thermal sensor.
I. INTRODUCTION IRELESS sensor networks [1] are offering the next evolution in biometrics and healthcare monitoring applications. By wireless body sensor network (WBSN) system, a hospital can monitor physiology signals (such as body temperature, electrocardiogram (ECG), pulse oxygen saturation, blood pressure, blood glucose, etc.) for 24 h without affecting the life of a patient, and the health department of a national government can control epidemic disease like SARS [2] efficiently. The WBSN plays an important role for healthcare monitoring applications.
W
Manuscript received January 08, 2009; revised September 01, 2009. First published October 23, 2009; current version published January 27, 2010. S.-L. Chen, H.-Y. Lee, C.-A. Chen, and C.-H. Luo are with the Department of Electrical Engineering, National Cheng Kung University, Taiwan (e-mail:
[email protected]). H.-Y. Huang is with the Graduate Institute of Electrical Engineering, National Taipei University, Taiwan. 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/JSYST.2009.2032440
Similar to other wireless mobile applications, a long-term use WBSN is necessary. Researchers have proposed efficient lowpower protocols [3], [4] for wireless sensor networks, which focused on reducing power consumptions of communication protocols and attaining great contribution. In order to compact with an open communication system, it is unavoidable to implement hardware with the high complexity and high cost. There are many proposed wireless ECG sensor node developed by open communication protocols and systems. Ekstrom [5] implemented a small sized ECG sensor system which can be wirelessly connected to a personal digital assistant (PDA) by Bluetooth. Park et al. [6] present an ultra-wearable wireless lowpower ECG monitoring system which is integrated by a wearable and ultra low-power wireless sensor node. Its communication is through 2.4-GHz transceiver and GFSK modulator. Lee et al. [7] propose a biomedical sensor node designed into PDA and cellular phone for 24-h healthcare. The communicating is through Zigbee and CDMA modem. Sakaue et al. [8] develop a wireless biosignal monitoring device which consists of a micro processing unit, a three-axis acceleration sensor, an ECG amplifier, and a Bluetooth wireless communication module. The above-mentioned proposed systems consume huge power for the communication as a result of no power management and efficient protocol designed into the systems. In general wireless sensor network, each wireless sensor node has to receive packages from other sensor nodes and select adjacent sensor nodes to send data out according to the header of each package. Although there are proposed many complex communication protocols [9], [10] and routing algorithms for wireless sensor network, the complexity, high cost, and handshake challenges must be solved in hardware implementation. Nowadays, the PC is powerful enough for controlling, scheduling [11], and routing transmission path within the wireless biomedical sensor network (WBSN) system. The proposed WBSN system architecture is separated into communication and control paths for biomedical applications [12]. It makes the implementation of the WBSN system easier by hierarchy architecture and reduces hardware cost with a simple sensor node design because of reducing complex communication protocols and routing algorithms. Moreover, in order to reduce transmission power consumptions in the WBSN system, a time sharing protocol is used as communication protocol. A communication cycle is constructed to synchronize the whole WBSN system, which allows the central control system to pipeline the WBSN system like a digital circuit. It advances the performance of the WBSN system by
1932-8184/$26.00 © 2009 IEEE Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4623 UTC from IE Xplore. Restricon aply.
www.DownloadPaper.ir
CHEN et al.: WBSN WITH ADAPTIVE LOW-POWER DESIGN
399
Fig. 1. Wireless sensor network system hierarchical architecture.
adding a few control functions into the original control system. All the applications will be controlled and scheduled completely by the control path. In addition to low complexity, low cost, and high performance, a low-power design is necessary for long-term healthcare monitoring applications. There are two ways to incorporate low-power design, one is through components and the other is through system operation. Recently, the advancement of low-power design through components contributes greatly to reducing power consumption especially in transceivers [13], [14]. In this paper, we will focus on adaptive low-power system that is designed with its starting point at the system level [15]. Adaptive low-power management provides optimal power management through an adaptive power conserving algorithm for the wireless sensor nodes of the WBSN system [16]. Additionally, a data encoder provides a power saving effect. The low-power design is always an interesting topic for mobile devices and sensor network. In biometrics and biomedical applications, the wireless recorder and monitor are battery-operated devices. For long-term or 24 h working, the power-constrained is strict and limited. The effective adaptive low-power devices and system are demanded to ensure adequately longtime working. The rest of the paper is organized as follows. Section II introduces the four levels architecture, communication path, and control path of the WBSN system. Section III illustrates adaptive low-power design by the hierarchy architecture, communication time sharing protocol, pipeline control, and adaptive power conserving algorithm. Section IV uses a thermal sensor node for example to describe the details of a wireless body sensor node. Section V shows the analog-to-digital converter (ADC), radio frequency (RF), modulator, and demodulator chip results. Finally, the body temperature and ECG applications are used for examples to analyze the specifications and power consumptions of the WBSN system. II. WBSN SYSTEM ARCHITECTURE A. System Architecture As shown in Fig. 1, there are four levels in the WBSN system with hierarchy and the relationship between each layer. The definitions and distinctions of each layer are described below. Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4623 UTC from IE Xplore. Restricon aply.
1) System layer. This layer is a PC in the hospital, which consists of the control path and the data path for the WBSN system. The control path includes the function control and power control signals. The main actions of the system layer are sending commands to the application layer and receiving data from it. 2) Application layer. This layer includes several sensor groups, and controls all the sensor groups by sending commands and receiving data from them. 3) Sensor group layer. This layer includes several sensor nodes, and controls all the sensor nodes in the sensor group layer by sending commands and receiving data from them. 4) Sensor layer. This layer is just a single sensor node receiving commands from a sensor group and sending the detected data to the sensor group. The four layers hierarchy system architecture helps the WBSN system development and verification. First of all, a sensor node development environment is built for wireless sensor node integrating. Then a sensor group is designed as a geometric node of a Voronoi diagram [17], [18], which provides a system to test the connection between sensor nodes and sensor groups. Next, an application layer is partitioned into cells, which supplies an environment to evaluate the communication ability when sensor nodes move. In this WBSN system, each wireless sensor node is collocated with a unique identification code and the sensor groups are the geometric nodes of a Voronoi diagram. The Voronoi diagram of a set of sensor groups partitions the sensor network area into cells [17], [18]. Every sensor node in a given cell is closer to the sensor group in this cell than to any other sensor group. Thus, the sensor group only transmits commands and classifies received data in its Voronio cell, and it also means no other sensor group can connect it. The unique identification code for each person and Voronoi diagram for sensor network area make people moving freely in a large area to be possible by changing sensor network group. It also makes sure that the biomedical data can be smooth received in PC-end. Furthermore, the proposed WBSN system can be integrated into a metropolitan or geographic network for real-time online healthcare monitoring system. The adaptive low-power wireless sensor node can also be the foundations for improving mobility conditions of people to move in a large-area. A virtual sensor network group system can be also designed into a mobile phone, PDA, or notebook,
www.DownloadPaper.ir
400
IEEE SYSTEMS JOURNAL, VOL. 3, NO. 4, DECEMBER 2009
Fig. 2. Hierarchical architecture for power control path.
and connect with wireless sensor nodes on human body. The control commands or detected data can be received and sent out through Bluetooth, Wi-Fi, or global system for mobile communcations (GSM) of wireless consumer electric products. This low-power design is an efficient and easy implement component for integrating into a large-area system. B. Communication Path The most important purpose of the data path is to transmit detected data from each sensor node to the PC. In this design, 2.4 GHz should be used as the communication transceivers in the sensor group layer. As shown in Fig. 1, the processed biomedical signals are transmitted from sensor nodes (sensor layer) to the collecting point (sensor group layer) through the 2.4-GHz band wireless communication system. After collecting and merging the detected signals from the sensor groups, the application layer sends the merged data to the central control system (system layer) by the 2.4-GHz band. C. Control Path The control path is an hierarchical architecture from the system layer to the sensor layer. It consists of the adaptive low-power control signals [16] and the function control signals. This control path not only gains the benefits of flexible, easy development, and run-time reconfigurability [19], but also significantly reduces power consumption. The details of adaptive low-power control design will be described in the next section. III. ADAPTIVE LOW-POWER CONTROL DESIGN A. Hierarchy Architecture As shown in Fig. 2, the power control signals are hierarchical from the PC to individual functions of each sensor node. When a user measures body temperatures in the WBSN system, the power control signals switch each function within a sensor node with accurate pipeline control through each layer. Each function circuit will be only powered on when necessary and the redundant functions in each cycle will be powered off to reduce unnecessary power consumptions. B. Communication Protocol In order to improve the performance of the WBSN system, the pipeline methodology was design into this system. It promotes the performance well, but it also increases the complexity Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4623 UTC from IE Xplore. Restricon aply.
in accurately controlling timing [11]. A communication cycle is proposed to synchronize the WBSN system. It can be multiple or equal to the synchronous clock in the digital circuit of sensor nodes, which varies with applications. Transmission of body temperature data takes less time than that of ECG or blood pressure application, since the data amount of the former is much less. Consequently, the communication cycle is set by the application which demands most communication time. Fig. 3 shows a five-application example of the time sharing protocol for the WBSN system. First of all, the sensor nodes are reset by a reset command from the sensor group. In order to synchronize the timing of sensor group and each sensor node, a handshaking action is applied for synchronization after reset. By this means, the sensor group and each sensor node have the same communication cycle starting point. The sensor 1 sends first transmission data out after synchronization, and after ( is the number of sensor nodes in this group) communication cycle time counting by the micro control unit (MCU), the sensor 2 sends the first transmission data out. The MCU of each sensor node counts down one communication cycle time. The transmitter of each sensor node sleeps from the end of current data transmission to the start of next transmission. The transmitter in each sensor node is powered on only when it is necessary to transmit data. The transmitting time of sensor 4 and sensor 5 data are much longer than sensor 1, sensor 2, and sensor 3 because the signal amounts of blood pressure and ECG are huger than body temperature, pulse oxygen, and blood Glucose. The sensor group receives detected data sensor by sensor. As shown in Fig. 3, the sensor group receives the first transmission data (data0) which includes the first transmission data of the sensor 1, sensor 2, sensor 3, sensor 4, and sensor 5. The sensor group will receive the next transmission data (data1) during the next communication cycle. C. Accurate Pipeline Control The adaptive power control path also works well with accurate pipeline control. The adaptive power control path can be separated into five independent steps which include system, application, sensor group, sensor node and function. With storage at the end of each step, the adaptive power control signals can be sent out step by step. This allows the control signals to issue control path at the processing rate of the communication cycle which is the slowest step. The communication cycle is much faster than the time needed to perform all steps at once. As shown in Fig. 4, the WBSN system can execute all functions in
www.DownloadPaper.ir
CHEN et al.: WBSN WITH ADAPTIVE LOW-POWER DESIGN
401
Fig. 3. Five-application example of time sharing protocol for the WBSN system.
Fig. 4. Pipeline for adaptive power control path.
the same time after the fourth communication cycle. Healthcare monitoring applications will benefit through better performance in the WBSN system by the pipeline control.
Fig. 5. Adaptive power conserving algorithm for each thermal sensor node.
D. Adaptive Power Conserving Algorithm in the Sensor Node An adaptive power conserving algorithm is proposed for power management in each sensor node. Fig. 5 shows an example of the body temperature application. While the communication cycle comes, the MCU powers on the receiver of the transceiver to receive the commands from the sensor group. After receiving the correct commands, the MCU checks the calibration status. Once the thermal sensor node is calibrated, the MCU powers on the ADC at the same time, and then powers on the data encoder to compress temperature values. At Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4623 UTC from IE Xplore. Restricon aply.
the end of the communication cycle, the MCU powers on the transmitter of the transceiver to transmit merged compression data to the sensor group. After transmission, the MCU powers off all functions except thermal sensor in the wireless thermal sensor node and wait until the next communication cycle starts. IV. WIRELESS THERMAL SENSOR NODE DESIGN Sensor nodes lie in the lowest layer of the WBSN system. The block diagram of the wireless sensor nodes is shown in
402
www.DownloadPaper.ir
IEEE SYSTEMS JOURNAL, VOL. 3, NO. 4, DECEMBER 2009
2)
Fig. 6. Block diagram of a sensor node.
3)
4) Fig. 7. FSM in the MCU power controller.
5)
Fig. 8. Simple example of Huffman coding algorithm.
Fig. 6. The details of each function in the thermal sensor node are described as follows. 1) MCU: The MCU is the controller of a thermal sensor node. It receives commands from the RF transceiver, and then sends both the power and function control signals to each function in the sensor node. Fig. 7 shows the finite state machine (FSM) of a power controller in the MCU. The design of the FSM is in accordance with the adaptive power conserving algorithm. Each function of the thermal sensor node is powered on only when necessary. In time in sleep state, and all the functions of a sensor node are powered off except the MCU and thermal sensor. Although the MCU Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4623 UTC from IE Xplore. Restricon aply.
and thermal sensor are always powered on, the power consumptions of both are small. Thermal BJT Sensor: Traditionally, thermistors are resistor-type detecting devices in which the resistance varies with temperature [20]. However, it is well known that the resistance is very inaccurate in the CMOS process. Therefore, as 400 series thermistors from YSI, they are unsuitable for integration on CMOS silicon chips [21]. Conversely, the transistors are commonly employed to generate the basic sensing signals for temperature sensors [22], [23] because these devices can provide a higher accuracy than resistor-type sensor in the sensor chip applications. In observance of this, the transistors are generally chosen for integration on the CMOS silicon chips. In a thermal sensor [21], a bipolar transistor has a strong exponential re) and the lationship between the base-emitter voltage ( collector current ( ). A thermal sensor is utilized for temperature sensing since its own characteristic of a negative temperature coefficient of resistivity. This phenomenon is in the sensory signal of a bipolar expressed with the transistor. ADC: The ADC consists of a sigma delta modulator (SDM), a digital filter (DF), and a downsampling circuit [21]. Following the response of sensors to variations in the biosignals, the SDM continuously samples the changes in signal voltage readings and converts these readings into a digital signal. The digital signal passes through a DF for filtering and then goes through downsampling by the downsampling circuit. The biomedical signals can be further increased in the ADC, so that in-band noise is effectively suppressed and shaped. Calibrator [21]: Temperature sensors are calibrated by trimming the number of substrate BJT transistors. The 8-bit calibrator controller signal tunes the sensor to approach a calibration point by increasing or decreasing the number of transistors connected in parallel. Data Encoder: The data compression consists of a predictor and an entropy encoder. The detected thermal values range from 0 to 1023 [21], which need 10-bits for a single value during transmission. Body temperature varies quite slow (far less than 1 Hz), thus the variations of the adjacent temperature values are very small. As demonstrated by Table I, 10-bit temperature values are too large to store or transmit. After prediction, the difference values by the predictor from the original temperature data are very close to zero. Huffman coding [24] was chosen as the data compression algorithm. The probability of Huffman encoding will be decided by the characteristics of different biomedical signals. For example, the variation of body temperature is very small. The probabilities are concentrated on small difference values. On the other hand, if the biomedical signals with more variation like ECG, the probabilities are dispersed to each difference value. An example of a Huffman coding algorithm is shown in Fig. 8, more probable prediction values are encoded with shorter codes, and less probable values are encoded with longer codes. Table I clearly shows that 73% of the data is reduced by the predictor and entropy coding. This linear prediction and
www.DownloadPaper.ir
CHEN et al.: WBSN WITH ADAPTIVE LOW-POWER DESIGN
403
TABLE I INTERMEDIUM PREDICTION VALUES AND OUTPUT BIT LENGTHS FOR ORIGINAL AND COMPRESSED DATA
TABLE II TYPICAL APPLICATION REQUIREMENTS
The transmitter is implemented with single balanced mixer, cross couple pair voltage control oscillator (VCO), and cascade buffer amplifier as PA. The adopted circuits in the receiver are cascoded source degeneration LNA, single balanced mixer, cross couple pair VCO, and demodulator.
Fig. 9. Layout figure of MCU.
Huffman coding algorithm is also selected as the lossless compression of ECG signals. After experiment, the averagecompression ratio is 1.92 in all case of MIT-BIH Arrhythmia database [25], which means there are 48% data reduced by this lossless compression method. The data encoder shortens the stream of detected temperature data effectively, which also means large amount of power saving by transmitting a shorter stream of data. 6) RF Transceiver, Modulator, Demodulator, and Antennas: The RF transceiver consists of a transmitter and a receiver designed with adaptive power switches. As shown in Fig. 10, the transceiver is controlled by the MCU which sends the control signals through Tctrl and Rctrl to switch for the transmiiter power on/off from VDD to and modulator or for the receiver and demodulator. These control signals also switch the modulator and demodulator. The transceiver, modulator, or demodulator is powered on only if necessary. Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4623 UTC from IE Xplore. Restricon aply.
V. EXPERIMENTAL RESULTS Table II illustrates the requirements of six important applications for the WBSN system. The sampling rates of the body temperature, Pulse Oxygen, and blood Glucose signals are much less than the blood pressure, ECG, and electroencephalography (EEG) signals. The compression rates of body temperature, pulse oxygen, and blood glucose signals are 3.33 with lower variations, and the compression rates of blood pressure, ECG, and EEG signals are 1.92 with higher variations. The setup and control overhead time is the time of initializing the RF transceiver, modulator, and demodulator, and executing the control signals. The switch overhead time is the time of switching transmitter or receiver in the same transceiver. The total transfer time per second means that how much time is necessary for each application to transfer per second. The duty cycle [26] is shown as follow: (1) where is active time (RF on), is allocated sleep time (time between wakeups), and is the number of retransmissions. The specifications of fabricated process, area, Vdd, Current, and power consumption of each component in a wireless thermal sensor node and wireless ECG sensor node are listed in Table IV. Fig. 9 shows the layout of the MCU circuit which was synthesized by TSMC 0.18cell library. Except specifications of the MCU is post layout simulation and the ECG sensor and amplifier by the commercial product, AD8553 [27], the others are tapeout chips. As shown in Figs. 11 and 12, the
404
www.DownloadPaper.ir
IEEE SYSTEMS JOURNAL, VOL. 3, NO. 4, DECEMBER 2009
Fig. 10. RF transceiver circuit with switch controls.
The 2.4-GHz RF transceiver circuit can provide 40 channels for FSK modulation in 80 MHz bandwidth. The specifications for the 2.4-GHz RF transceiver circuit are shown in Table III. In the transmitter path (mixer, oscillator and PA), the transmitted power is shown as follows: (2) (3)
Fig. 11. Photograph of 2.4-GHz RF transceiver circuit.
TABLE III 2.4-GHZ NODE-TO-NODE LINK BUDGET
tapeout chip of 2.4-GHz transceiver was fabricated using a standard 0.18by TSMC, and the tapeout chip of modulator and demodulator was fabricated using a standard 0.25 by TSMC. Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4623 UTC from IE Xplore. Restricon aply.
is the transmitted power, is the gain of receive where antenna, is the gain of transmit antenna, is the received power that means sensitivity and path loss is the power lost in the air. In the receiver path (mixer, oscillator and LNA), the sensitivity should be concerned and shown as follows:
(4) is the signal energy per bit-to-noise spectral denwhere sity ratio. These specifications are needed as follows. . 1) Output power in transmitter is . 2) Sensitivity in receiver is 3) Noise figure in receiver is 25 dB. 4) IF frequency is 320 MHz ~ 400 MHZ. 5) Channel bandwidth is 1 MHz. Examples of 10 values/s power consumption in the thermal sensor node and 240 values/s power consumption in the wireless ECG sensor node are also shown in Table IV. The power consumption of ECG sensor and amplifier is adapted from the data sheet of the commercial product, AD8553 [27]. Obviously, the power consumptions of the transceiver, modulator, and demodulator are much greater than other components in this system. When all functions are switched on by the power controller, the ) in each wireless thermal sensor node is total energy ( (5)
www.DownloadPaper.ir
CHEN et al.: WBSN WITH ADAPTIVE LOW-POWER DESIGN
405
TABLE IV SPEC OF EACH COMPONENT, 10 VALUES/S AND 250 VALUES/S POWER CONSUMPTION EXAMPLES IN THE WIRELESS THERMAL AND ECG SENSOR NODES
TABLE V PERCENTAGE CHANGE BETWEEN AVERAGE AND REDUCED POWER CONSUMPTION BETWEEN WIRELESS THERMAL SENSOR NODES WITH ADAPTIVE AND DATA ENCODING DESIGNS
where is the total energy of the MCU, is the total energy of the thermal BJT sensor, is the total energy of the calibrator, is the total energy of the analog to digital converter, is the total energy of the data encoder, and is the total energy of the communication. The total communication energy making up of the most power consumption in the whole system can be divided into four parts [32]:
(6) where , , , and are the power consumptions of the transmitter, receiver, modulator, and demodulator independently. Further more, they can be divided into several operation time procedures.
Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4623 UTC from IE Xplore. Restricon aply.
where the
or , , , are the power consumptions when the transceiver is transmitting or receiving, initializing, turning on, and switching. , , and The are the power consumptions of the modulator modulating, initializing, and powering on. The , , and are the power consumptions of the demodulator demodulating, initializing, and turning on. Total energy without adaptive design is
406
www.DownloadPaper.ir
IEEE SYSTEMS JOURNAL, VOL. 3, NO. 4, DECEMBER 2009
TABLE VI PERCENTAGE CHANGE BETWEEN AVERAGE AND REDUCED POWER CONSUMPTION BETWEEN WIRELESS ECG SENSOR NODES WITH ADAPTIVE AND DATA ENCODING DESIGN
), calibrator ( ), analog to digital converter ( ( data encoder ( ), and communication ( The total energy with adaptive design is
), ).
Fig. 12. Photograph of modulator and demodulator.
where is run-time and the total energy ( ) is integrated with the sum of each component’s power multiplied by it’s run-time [33]. Total power consumption with adaptive design is :
Fig. 13. Hardware emulation of the wireless thermal sensor node.
(8)
The total power consumptions without adaptive design is
(7) where the total power consumption ( sum of power consumption in the MUC ( Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4623 UTC from IE Xplore. Restricon aply.
) is a ), thermal sensor
where the total power consumption with adaptive design ( ) is the sum of that each component needs when it is turned on during the adaptive process. Initializing, turning-on, and switching energy [32] are considered in addition to power simulation [34], [35]. Table V shows the average power consumption and the power consumption reduction percentages using an adaptive algorithm and data encoding design for a thermal sensor node. Comparing power consumption reduction percentages from adaptive design with or without the data encoder, the data encoder reduces average power consumption more effectively with a greater amount of recorded values. For example, 4000 samples in Table V, the power consumption of a wireless thermal sensor without adding ) is 37.8% larger than it added a a data encoder (739.58 data encoder (459.98 ). As shown in Table IV, when a
www.DownloadPaper.ir
CHEN et al.: WBSN WITH ADAPTIVE LOW-POWER DESIGN
407
TABLE VII COMPARISON OF WIRELESS THERMAL SENSOR POWER CONSUMPTION WITH PREVIOUS WORKS
TABLE VIII COMPARISON OF WIRELESS ECG SENSOR POWER CONSUMPTION WITH PREVIOUS WORKS
wireless thermal sensor detecting 1 temperature per minute and 1, 5, 10, 50, or 4000 temperatures per second, adaptive and data encoding design reduces power consumption by 99.577%, 99.575%, 99.575%, 99.573%, 99.559%, or 98.154% compared to that without adaptive design. Table VI shows the results of ECG application, when a wireless ECG sensor detecting 120, 200, 240, 400, or 1000 values per second, adaptive and data encoding design reduces power consumption by 99.383%, 99.237%, 99.164%, 98.872%, or 97.778% compared to that without adaptive design. The power consumption of a wireless ECG sensor without adding a data encoder (642.7 ) is 8.8% larger than it added a data encoder (585.8 ) when it detecting 1000 values per second. This adaptive power control path was incorporated into the WBSN system for healthcare monitoring applications. The MCU, 2.4-GHz wireless transceiver communication system, and data encoder are added to build a wireless thermal sensor network. Table VII presents a power consumption report comparing the performances with previous works [22], [23], [33]–[35]. The power consumptions of two thermal sensors in the industry have also been presented in Table VII [38], [39]. After comparisons with published works [22], [23], [36], [37] and previous works in the industry [38], [39] which consist only of a sensor and ADC (without MCU, 2.4-GHz transceiver, and data encoder), the power consumption of the wireless thermal Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4623 UTC from IE Xplore. Restricon aply.
sensor node with adaptive low-power and data encoding design is reduced by over 69.6% [23]. Table VIII presents a power consumption report to compare the achieved performance with previous works [5], [6] of wireless ECG sensor. The power consumptions of the small sized ECG sensor system [5] are 170 mW for 400 ECG values per second. The power consumptions of the wearable and ultra low-power wireless sensor node [6] are 30 mW for 1000 ECG values per second. The power consumptions of the biomedical sensor node [7] are 1 W. The power consumptions of the biosignal monitoring device [8] are 756 mW for 200 ECG values per second. The power consumptions of this work are 201.2 , 297.4 , or 585.8 for 200, 400, or 1000 ECG values/s. After comparisons with published works [5]–[8] which selects Bluetooth [5], [8], 2.4 G Transceiver [6] and Zigbee [7] as wireless protocols, the power consumption of the proposed work with adaptive low-power and data encoding design is reduced by over 98% [6]. We tested and verified functions of a wireless thermal sensor node by hardware emulation. As Fig. 13 shows, the function of MCU is emulated by FPGA. VI. CONCLUSION In this paper, we have proposed a system of lower power consumption for a wireless thermal sensor node in WBSN. A communication cycle is constructed to synchronize the WBSN
408
www.DownloadPaper.ir
IEEE SYSTEMS JOURNAL, VOL. 3, NO. 4, DECEMBER 2009
system with a pipeline to advance system performance. The transceiver consumes the most power in the sensing node, and the data encoder reduces average power consumption more effectively when there is a greater amount of detected values. When detecting ten body temperatures or 240 ECG signals per or 220.4 second, the power consumption is either 106.3 . It ends up with a reduction 99.573% or 99.164% in power consumption using the proposed adaptive power conserving algorithm and data encoding design. The effect of the data encoder is obvious in ECG or blood pressure application since high data rate results. The benefits of data compression are that it not only reduces power consumption by transmitting a shorter stream of data, but also allows adding more advanced error control coding to the transmitting data. REFERENCES [1] D. Culler, D. Estrin, and M. Srivastava, “Overview of sensor networks,” IEEE Computer (Special Issue on Sensor Networks), Aug. 2004. [2] W. T. Chiu et al., “Infrared thermography to mass-screen suspected Sars patients with fever,” Asia Pac. J. Public Health, vol. 17, pp. 26–28, Jan. 2005. [3] M. Avvenuti, P. Corsini, P. Masci, and A. Vecchio, “Energy-efficient reception of large preambles in MAC protocols for wireless sensor networks,” Electron. Lett., vol. 43, no. 5, Mar. 2007. [4] C. C. Enz, A. El-Hoiydi, J. D. Decotignie, and V. Peiris, “WiseNET: An ultralow-power wireless sensor network solution,” IEEE Computer, vol. 37, no. 8, pp. 62–70, Aug. 2004. [5] M. Ekstrom, “Small Wireless ECG Bluetooth Communication to a PDA,” M.S. thesis, Dept. Comput. Sci. Electron., Malardalen Univ., , Vasteras, Sweden, 2006. [6] C. Park, P. H. Chou, Y. Bai, R. Matthews, and A. Hibbs, “An ultra-wearable, wireless, low power ECG monitoring system,” in Proc. IEEE Biomedical Circuit System Conf., London, U.K., Nov. 2006, pp. 241–244. [7] T. S. Lee, J. H. Hong, and M. C. Cho, “Biomedical digital assistant for ubiquitous healthcare,” in Proc. IEEE Engineering in Medicine and Biology Int. Conf., Lyon, France, Aug. 23–26, 2007. [8] Y. Sakaue and M. Makikawa, “Development of wireless biosignal monitoring device,” in Proc. 6th Int. Special Topic Conf. on ITAB, Tokyo, Japan, 2007. [9] M. H. Jin, R. G. Lee, C. Y. Kao, Y. R. Wu, D. F. Hsu, T. P. Dong, and K. T. Huang, “Sensor network design and implementation for health telecare and diagnosis assistance applications,” in Proc. IEEE Int. Conf. Parallel and Distributed Systems, Fuduoka, Japan, Jul. 2005. [10] E. Lamprinos, A. Prentza, E. Sakka, and D. Koutsouris, “Energy-efficient MAC protocol for patient personal area networks,” in Proc. IEEE Int. Conf. Engineering in Medicine and Biological Soc., Shanghai, China, Sep. 2005. [11] T. ElBatt and A. Ephremides, “Joint scheduling and power control for wireless ad hoc networks,” IEEE Trans. Wireless Commun., vol. 3, no. 1, pp. 74–85, Jan. 2004. [12] S.-L. Chen, H.-Y. Lee, C.-A. Chen, H.-Y. Huang, and C.-H. Luo, “Wireless sensor network system by separating control and data path (SCDP) for bio-medical applications,” in Proc. Eur. Microwave Conf. 2007 (EuMC’07), Munich, Germany, Oct. 8–12, 2007. [13] H. M. Seo, Y. K. Moon, Y. K. Park, D. Kim, D. S. Kim, Y. S. Lee, K. H. Won, S. D. Kim, and P. Choi, “A low power fully CMOS integrated RF transceiver IC for wireless sensor networks,” IEEE Trans. VLSI Syst., vol. 15, no. 2, pp. 227–231, Feb. 2007. [14] D. C. Daly and A. P. Chandrakasan, “An energy-efficient OOK transceiver for wireless sensor networks,” IEEE J. Solid-State Circuits, vol. 42, no. 5, pp. 1003–1011, May 2007. [15] V. Raghunathan, C. L. Pereira, M. B. Srivastava, and R. K. Gupta, “Energy aware wireless systems with adaptive power-fidelity tradeoffs,” IEEE Trans. VLSI Syst., vol. 13, no. 2, pp. 211–225, Feb. 2005. [16] H.-Y. Lee, S.-L. Chen, C.-A. Chen, and C.-H. Luo, “Wireless thermal sensor network with adaptive low power design,” in Proc. IEEE Eng. Medicine and Biology Int. Conf. 2007 (SFGBM), Lyon, France, Aug. 23–26, 2007. [17] F. Aurenhammer, “Voronoi diagrams—A survey of a fundamental geometric data structure,” ACM Comput. Surv., vol. 23, pp. 345–405, 1991. [18] S. Fortune, D. Du, and F. Hwang, “Voronoi diagrams and Delaunay triangulations,” Euclidean Geom. and Comput., 1992. Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4623 UTC from IE Xplore. Restricon aply.
[19] T. Arslan, N. Haridas, E. Yang, A. T. Erdogan, and W. G. J. Howells, “ESPACENET: A framework of evolvable and reconfigurable sensor networks for aerospace – based monitoring and diagnostics,” in Conf. Rec. 2006 1st NASA/ESA Conf. Adaptive Hardware and Systems (AHS’06), pp. 323–329. [20] D. Barrettino, M. Graf, H. S. Wan, K. U. Kirstein, A. Hierlemann, and H. Baltes, “Hotplate-based monolithic CMOS microsystems for gas detection and material characterization for operating temperatures up to 500 C,” IEEE J. Solid-State Circuits, vol. SC-39, no. 7, pp. 1202–1207, Jul. 2004. [21] H.-Y. Lee, S.-L. Chen, and C.-H. Luo, “A CMOS smart thermal sensor for biomedical application,” Inst. Electron., Inform., and Commun. Eng. (IEICE), vol. E91-C, no. 1, pp. 96–104, Jan. 2008. [22] M. A. P. Pertijs, A. Niederkorn, X. Ma, B. McKillop, A. Bakker, and J. H. Huijsing, “A CMOS smart temperature sensor with a 3 inaccuracy of 0:5 C from 50 C to 120 C,” IEEE J. Solid-State Circuits, vol. 40, no. 2, pp. 454–461, Feb. 2005. [23] M. A. P. Pertijs, K. A. A. Makinwa, and J. H. Huijsing, “A CMOS smart temperature sensor with a 3 inaccuracy of 0:1 C from 55 C to 125 C,” IEEE J. Solid-State Circuits, vol. 40, no. 12, pp. 2805–2815, Dec. 2005. [24] D. A. Huffman, “A method for the construction of minimum redundancy codes,” in Proc. IRE., 1951, vol. 40, pp. 1098–1101. [25] G. B. Moody and R. G. Mark, “The impact of the MIT/BIH arrhythmia database,” IEEE Eng. Med. Biol. Mag., vol. 20, no. 3, pp. 45–50, May/ Jun. 2001. [26] O. Omeni, A. C. W. Wong, A. J. Burdett, and C. Toumazou, “Energy efficient medium access protocol for wireless medical body area sensor networks,” IEEE Trans. Biomed. Circuits Syst., vol. 2, no. 4, pp. 251–259, Dec. 2008. [27] AD8553 Data Sheet Analog Devices Products [Online]. Available: http://www.analog.com [28] U. Varshney, “Improving wireless health monitoring using incentivebased router cooperation,” Computer, vol. 41, no. 5, pp. 56–62, May 2008. [29] D. Huber, L. Falco-Jonasson, M. Talary, W. Stahel, N. Stadler, F. Dewarrat, and A. Caduff, “Multi data fusion for non-invasive continuous glucose monitoring,” in Proc. Information Fusion 10th Int. Conf., Jul. 9–12, 2007. [30] P. Li, S. H. Sur, R. E. Mistlberger, and M. Morris, “Circadian blood pressure and heart rate rhythms in mice,” Amer. J Physiol., vol. 276, pp. R500–R504, 1999. [31] C.-T. Lin, L.-W. Ko, J.-C. Chiou, J.-R. Duann, R.-S. Huang, S.-F. Liang, T.-W. Chiu, and T.-P. Jung, “Noninvasive neural prostheses using mobile and wireless EEG,” Proc. IEEE, vol. 96, no. 7, pp. 1167–1183, Jul. 2008. [32] J. Polastre, J. Hill, and D. Culler, “Versatile low power media access for wireless sensor networks,” in Proc. 2nd ACM Conf. Embedded Networked Sensor Systems (SenSys), Nov. 3–5, 2004. [33] G. Zhou, C. Huang, T. Yan, J. A. Stankovic, and T. F. Abdelzaher, “MMSN: Multi-frequency media access control for wireless sensor networks,” in Proc. IEEE INFOCOM, Barcelona, Spain, 2006. [34] R. Jurdak, P. Baldi, and C. V. Lopes, “Adaptive low power listening for wireless sensor networks,” IEEE Trans. Mobile Comput. Arch., vol. 6, no. 8, pp. 988–2004, Aug. 2007. [35] W. Ye, J. Heidemann, and D. Estrin, “Medium access control with coordinated, adaptive sleeping for wireless sensor networks,” IEEE Trans. Networking, vol. 12, no. 3, pp. 493–506, Jun. 2004. [36] A. Bakker and J. H. Huijsing, “Micropower CMOS temperature sensor with digital output,” IEEE J. Solid-State Circuits, vol. 31, no. 7, pp. 933–937, Jul. 1996. [37] M. Tuthill, “A switched-current, switched-capacitor temperature sensor in 0.6-m CMOS,” IEEE J. Solid-State Circuits, vol. SC-33, pp. 1117–1122, Jul. 1998. [38] LM92 Data Sheet National Semiconductor Corporation, Mar. 2008. [39] DS1626 Data Sheet Maxim Int. Products, May 2008 [Online]. Available: http://www.maxim-ic.com
6
0
6
0
Shih-Lun Chen received the B.S. and M.S. degrees in electrical engineering from National Cheng Kung University (NCKU), Tainan, Taiwan, in 2002 and 2004, respectively. He is currently pursuing the Ph.D. degree in the Department of Electrical Engineering, NCKU. His research fields include wireless sensor networks, adaptive power management, image processing, biomedical signal processing, and reconfigurable architecture.
www.DownloadPaper.ir
CHEN et al.: WBSN WITH ADAPTIVE LOW-POWER DESIGN
Ho-Yin Lee was born in Hong Kong in 1979. He received the B.S. and Ph.D. degrees in electrical engineering from National Cheng Kung University, Tainan, Taiwan, in 2002 and 2007, respectively. His research fields include medical analog circuit design, medical signal processing, and wireless sensor network.
Chiung-An Chen received the B.S. degree in electronic engineering from Chung Yuan Christian University, Chung Li, Taiwan, in 2005. She is currently pursuing the Ph.D. degree in the Wireless Mixed Biochip Lab, National Cheng Kung University, Tainan, Taiwan. Her research fields include radio frequency circuit in wireless systems, adaptive power management, and wireless sensor networks.
Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4623 UTC from IE Xplore. Restricon aply.
409
Hong-Yi Huang (S’89–M’94) was born in Taiwan, in 1965. He received the B.S. degree in nuclear engineering from National Tsing-Hua University, Hsinchu, Taiwan, in 1987 and the M.S. and Ph.D. degrees from the Institute of Electronics, National Chiao-Tung University, Hsinchu, in 1989 and 1994, respectively. He was with Industrial Technology Research Institute (ITRI) from 1994 to 1999, engaged in mixedsignal integrated circuits design. He joined the Department of Electronic Engineering, Fu-Jen Catholic University, Taiwan, from 1999 to 2006. Since 2006, he has been an Associate Professor with the Graduate Institute of Electrical Engineering, National Taipei University, Taiwan. His research interests are in biocircuits and systems, mixedsignal and RF circuits, embedded memory, and communication circuits. He holds over 30 patents on VLSI circuits.
Ching-Hsing Luo (M’05) received the B.S. degree in electrophysics from National Chiao-Tung University, Hsinchu, Taiwan, and the M.S. degrees in electrical engineering from National Taiwan University, Taipei, in 1982 and in biomedical engineering from the Johns Hopkins University, Baltimore, MD, in 1987. He received the Ph.D. degree in biomedical engineering from Case Western Reserve University, Cleveland, OH, in 1991. He is a Distinguished Professor in the Department of Electrical Engineering, National Cheng Kung University, Tainan, Taiwan, since 2005. His research interests include biomedical instrumentation-on-a-chip, assistive device and agent system, cell modeling, signal processing, RFIC, implantable wireless biosensing chip, genome simulation, and pulse diagnosis platform in traditional Chinese medicine.
