A Bluetooth Low Energy Approach for Monitoring Electrocardiography and Respiration

Transcription

1 A Bluetooth Low Energy Approach for Monitoring Electrocardiography and Respiration Bing Zhou 1,2, Xianxiang Chen 1, Xinyu Hu 1,2, Ren Ren 1,2, Xiao Tan 1,2, Zhen Fang 1, Shanhong Xia 1 1 State Key Laboratory of Transducer Technology Institute of Electronics, Chinese Academy of Sciences, Beijing, China 2 University of Chinese Academy of Sciences, Beijing, China Abstract This paper presents a design and a contrast test of an ultra-low power wireless health monitoring system capable of measuring a subject s ECG (Electrocardiography), respiration, and body temperature. The system is based on the BLE (Bluetooth Low Energy) technology which is most valued for its ultra-low power consumption. Compared to our former design using MSP430 MCU and Bluetooth 2.1, this new design is much more highly integrated and can reduce power consumption significantly, which is generally a vital problem should be considered in WSN (Wireless Sensor Network) issues. The new system can save as much as nearly 75% power consumption than the former design when working in the same mode with the sampling rate at 250 Hz, which means that the battery life can extend to 107 hours compared to 26 hours of the former one, both using a 3.7 V lithium polymer battery with the capacity of 1100 mah. Meanwhile, the new design can connect 3 nodes simultaneously with a single PC or smart phone wirelessly; this number may increase with the new BLE stack will be released in the future. Keywords Bluetooth Low Energy; wireless sensor; ECG; respiration I. INTRODUCTION The ability to monitor the health status of elderly patients or patients undergoing therapy at home enables significant advantages in terms of both cost and comfort of the subject. However, such non-clinical applications of biomedical signal monitoring require various improvements not only in terms of size and comfort of the acquisition systems, but also in terms of their power dissipation [1]. Wireless sensor networks have been developed for elderly care [2] and clinical monitoring [3, 4]. Most tele-cardiology systems use wearable devices (such as portable ECG recorder, Sphygmomanometer, Pulse Oximetry, and so on) to collect remote cardiac patients physiological data (including 2 or 3 lead ECG, Blood pressure, Pulse rate) [5]. ECG is perhaps the most important diagnostic signal for a health monitor, hence why almost all implementations include it. Besides, monitoring respiration is important in a large number of cases. It s critical to monitor respiration in patients with cardiopulmonary issues such as congestive heart failure and patients who are on medication which suppresses breathing [6]. So our team chose to work on a project to monitor the ECG, respiration and body This paper was supported by funds of National High Technology Research and Development Program of China (863 program- 2012AA040506, 2011AA040405, 2013AA041201), National Natural Science Foundation of China (No , No ). temperature simultaneously with a single highly integrated and small-sized device. Our latest design which will be discussed in this paper matches the two requirements for comfort and low power mentioned above very well with the size of 40 mm * 67 mm and an ultra-low current consumption of about ma. The main improvement compared to our former design is the implement of BLE Technology, based on TI s (Texas Instruments) Bluetooth low energy chip CC2540, which plays the role both as the MCU and wireless transmission module, replacing the MSP430 micro controller and Bluetooth 2.1 module in the former one, hence makes the ultra-low power consumption possible. Two ADS1292R (each with 2 ADC channels) Analog-To-Digital Converters are adopted to acquire body potential signals for monitoring ECG, heart rate, which is derived from ECG signals by determining the R-R intervals, body temperature, and the respiration. This paper mainly consists of 5 sections: in section 2, the hardware design of the device is introduced, especially on the microcontroller and data acquisition module; Bluetooth low energy protocol are introduced in section 3; in section 4, we tested the power consumption and analyzed it in detail, besides, the final results are presented in this section; the last section is about the conclusion. II. HARDWARE DESIGN OF THE DEVICE Small size for comfortable acquisition and low power consumption for longer battery life should be considered as two important factors in the design of the system. The block diagram of Fig. 1 illustrates the hardware framework of the monitoring system. Except for BLE112 and ADC, TPS73201 and TPS73033 are used to provide suitable voltage for different modules. In this section, we will mainly discuss the microcontroller and wireless communication module and signal acquisition module. A. Microcontroller and wireless communication module A microcontroller is the core of most health monitoring measurement applications. For the sake of easy implementation and higher integration, we chose the Bluegiga s (Bluegiga Technologies) Bluetooth low energy module BLE112 as the microcontroller as well as the wireless communication module in our new design. The BLE112 module is a Bluetooth low energy single mode /13/$ IEEE 116

2 module based on TI s Bluetooth 4.0 single mode chip CC2540 that is targeted for low power sensors and accessories. BLE112 offers all Bluetooth low energy features: radio, stack, profiles and application space for our own applications [7], so no external processor is needed. The module also provides flexible hardware interfaces to connect sensors which make it possible to act as the microcontroller as well in our design. Fig. 1. Hardware framework of the system BLE112 has many new features compared to our former design using classic Bluetooth technology. It supports master and slave modes, and can connect 3 slaves simultaneously in master mode. In our design, the USB dongle works in master mode and the sensor nodes work in slave mode, the USB dongle can send and receive data from three sensor nodes simultaneously, so it is possible for us to monitor 3 subjects at the same time with only one PC or a smart phone, which is a main advance compared to our former design which can monitor only one subject at one time. Another important feature is its ultra-low current consumption: 27 ma (Tx power at 0 dbm) in transmitting and only 0.4 A in sleep mode. Despite of its advantages such as low power consumption and multi-connections, there are also some disadvantages. The primary disadvantage of the BLE technology is the lack of compatibility compared to standard Bluetooth. We can easily make our device communicate with most smart phones, tablets or any PC with a Bluetooth module through some common tools or software developed by ourselves. But, currently, only Apple s ios devices with Bluetooth 4.0, such as iphone4s/5, ipad 4, ipad mini, are compatible with the BLE devices, while other smart phones, like android or windows phone, have no official BLE SDK yet. However, the standard is being adopted by more and more smart phone manufacturers; Samsung has released its first beta BLE SDK for its galaxy series smart phones lately. Another disadvantage compared to the standard Bluetooth is its lower bandwidth. However, the data rate in our design is relatively small, and the bandwidth is sufficient for our demands, so this limitation should be manageable. B. Signal acquisition module Signal acquisition is very critical to the performance of the system as the measurement needs very accurate and stable acquisition, and the design of the front end signal acquiring circuit and the choice of ADC is the key. Consider the intention of our system design, we want to make it highly integrated, low power and as small as possible for confortable wearing; TI s ADS1292R is chosen as the ADC of the circuit for its ultra-low power consumption (335 W/channel), high resolution, high integration and small package. ADS1292R is a multichannel, simultaneous sampling, 24-bit, delta-sigma (ΔΣ) analog-to-digital converter with a built-in programmable gain amplifier (PGA), internal reference, and an onboard oscillator [8]. With high levels of integration and exceptional performance, the ADS1292R successfully makes our design at significantly reduced size, power, and overall cost. The ADS1292R also has a built-in right leg driver amplifier, lead-off detection and test signals, which makes it very suitable for ECG signal acquisition. Moreover, the integrated respiration measurement can be used to monitor respiration directly. Wheatstone bridge circuit is adopted to measure the body surface temperature. The internal reference programmed to 2.42 V is used as reference voltage of the ADC and the driving voltage of the bridge, so the effect of driving voltage variation on resistance detection sensitivity can be cancelled. In our design, two ADCs share the same SPI interface. In order to synchronize the two chips, one of them is set as the master for the clock source with the internal oscillator enabled, and the internal clock of the master is brought out to be used as the external clock source for the other one. III. BLUETOOTH LOW ENERGY PROTOCOL Bluetooth Low Energy has many advantages over Classic Bluetooth and other wireless technologies, that s why we choose it as the wireless transmit solution as well as the microcontroller in our design. Simple star topology reduces implementation complexity significantly; meanwhile, it is very robust through frequency hopping compared to other wireless technologies with very low power, which is the highlight of this technology. The BLE protocol stack architecture is illustrated on Fig. 2. It consists of two sections: the controller and the host. The separation of controller and host goes back to standard Bluetooth BR/EDR devices, in which the two sections were often implemented separately. Any profiles and applications that are being used sit on top of the GAP and GATT layers of the stack [9]. Here is a brief introduction of some layers which needed to be aware of in our development: PHY layer is a 1 Mbps adaptive frequency-hopping GFSK (Gaussian Frequency-Shift Keying) radio operating in the unlicensed 2.4 GHz ISM (Industrial, Scientific, and Medical) band. It has 40 channels with 2MHz channel spacing, from 2402 MHz to 2480 MHz. The minimum transmit power is 0.01 mw, and the maximum is 10 mw, which is very important to realize low energy consumption. The TX power which can be modified within the source code flashed into CC2540 directly determines the range of the 117

3 wireless transmission: about 30 meters with 0 dbm TX power and -70 dbm RX sensitivity and over 100 meters with 10 dbm TX power and -90 dbm RX sensitivity. Typically, the devices have 0~4 dbm Tx power and -85~-90 dbm sensitivity. Fig. 2. BLE Protocol Stack The LL essentially controls the RF state of the device, with the device being in one of five possible states: standby, advertising, scanning, initiating, or connected. The Link Layer controls 40 channels mentioned in the PHY part: 3 advertising channels used for discoverability and connectability, broadcasting, and avoid known frequencies; 37 data channels used for reliably sending application data in a connection and Adaptive Frequency Hopping for co-existence and robustness. The HCI layer provides a means of communication between the host and controller via a standardized interface. This layer can be implemented either through a software API, or by a hardware interface such as UART, SPI, or USB. In our design, one CC2540 communicates with two ADS1292R through SPI interface. The GATT layer is a service framework that defines the sub-procedures for using ATT. All data communications that occur between two devices in a BLE connection are handled through GATT sub-procedures. Therefore, the application and/or profiles will directly use GATT. We don t need to care about other layers in our development. More details are in the Bluetooth specification Version 4.0. IV. POWER ANALYSIS AND FUNCTION TEST Power consumption is a main factor that should be considered in many designs, especially in wireless sensor network related projects. There are many factors that affect the power consumption of the design, we can analyze it from two aspects: hardware aspect and software aspect. For the hardware aspect, we have explained in section 2. How much we can reduce the power consumption largely depends on how we design the software flashed into the CC2540. Firstly, as RF consumes a lot of power, we should calculate the data transmit rate, which depends on the sampling rate. For ADS1292R, the number of data outputs is (24 status bits + 24 bits * 2 channels) = 72 bits. In our design, we don t need the status bits, so we just need to read out 48 bits in a sampling cycle from one chip. As we used two ADS1292R working simultaneously at the sampling rate of 250 Hz, one for EKG and respiration, the other for respiration and body temperature. So the data rate is as follow: DATA RATE = ( )*250 bits/s = 1.5 kb/s According to the Bluetooth specification, in a BLE connection between two devices, a frequency-hopping scheme is used, in which the two devices each send and receive data from one another on a specific channel, then meet at a new channel at a specific amount of time later. This meeting where the two devices send and receive data is known as a connecting event. The connection interval is the amount of time between two connection events, in units of 1.25 ms. The minimum value of the connection interval is 6 (7.5 ms), meanwhile in each connecting event, 4 notifications can be sent which each can carry 20 bytes at most. So the maximum transfer rate through notifications is as follow: TRANS RATE = 20*4 bytes/7.5 ms = kb/s From the above calculation, we can know that the transfer speed of BLE through notifications can fulfill our need in the data acquiring system. In fact, we don t need to gather all the output bits in every sampling cycle as the body temperature doesn t need such a high sampling rate. We chose to read the body temperature channel every 1000 sampling cycles and send it wirelessly. This leads to a decrease in power consumption nearly 6.3% compared to sending all the data acquired in every sampling cycle. TABLE I. CURRENT CONSUMPTION IN DIFFERENT WORKING MODES As mentioned above, the RF state of the Bluetooth module can be in one of five possible states: standby, advertising, scanning, initiating, or connected. In our design, the Bluetooth module works as a slave, so its state can be in standby, advertising or connected. The power consumption is closely associated with the RF state. And whether the ADS1292R is working also influences the overall power consumption. Table 1 shows a typical current consumption 118

4 value measured with a digital multimeter when the system is running in different modes with the Tx power set to 0 dbm and at the sampling rate of 250 Hz. The software flashed into BLE112 is developed using the BLE software development kit from TI, which consists of five major sections: the OSAL (Operation System Abstraction Layer), HAL (Hardware Abstraction Layer), the BLE Protocol, profiles, and the application. The OSAL is a control loop that allows software to setup the execution of events. The BLE protocol stack, the profiles, and all applications are built around the OSAL. When the OSAL is running, it will keep polling all the tasks that were set up in the initialization. If there are no events to execute for a loop, the system will be put into sleep. The Standby mode here means that the system is in sleep and the two ADC converters are powered down by holding the RESET pin to low. Ideally, the current should be only about ma of BLE devices working in Standby mode, while our device still with the ADCs powered down and the core processor put into sleep. We analyzed our circuit again, and found out that the power regulators we used to supply power for the BLE112 and ADCs maybe the problem. They are both belongs to LDO (Low Dropout regulator) regulators, which provide essential performance characteristics for many precision analog and mixed-signal applications including low noise, fast transient response, high power supply rejection ration and small size [10]. In our design, we need to acquire very weak signals and the system is very sensitive to noise, so the LDO regulators are chosen to supply the power. But the efficiency of LDO regulators are relatively low, the greater the D-value between the input and output voltage, the lower the efficiency. As it s our first BLE approach demo, this problem will be solved in our next design, whether by redesigning the circuit or reconsidering to choose more proper power management chips. In order to analyze the power consumption more deeply, we captured the current wave using an Oscilloscope and a 50 ohm series resistor between the battery and the device. By measuring the voltage on the series resistor, we can calculate the current consumption of the device. The resistance of the series resistor can t be too large or too little, it will low down the voltage for the device too much if the resistance is too large, hence produces bigger errors; however, the resistance value should not be too small, otherwise the noise will influences our results, making the current wave difficult to recognize. As the typical current consumption is around 10 ma when working normally, we chose a 50 ohm resistor. It may lower down the voltage about 500 mv, so the real voltage for the device is about 3.2 V, which is in the range for proper working. Here are two figures which illustrate the current consumptions in different working modes, in different time scales, one at 20 ms and the other at 10 ms, both with Tx power set to 0 dbm, respectively. Fig. 3. Current consumption in different working modes (channel 1 (yellow) is for device connected without sending data and channel 2 (green) is for device in advertising mode but not connected, both at a sampling rate of 250 Hz.) Fig. 4. Current consumption in different working modes (channel 1 (yellow) is for device running normally and channel 2 (green) is for device connected without sending data, both at a sampling rate of 250 Hz.) In our test, in order to ensure stable data transmission and make the power consumption as low as possible, the connection interval is set to 12 (15 ms), so a connection event happens every 15 ms, producing a current pulse with a period of 15 ms, which matches the wave illustrated in the two figures above. We can see in the Fig. 3 that even when the device is not connected, there are still pulses with a period about 110 ms; it s caused by advertising, by sending specific data letting any central device (the USB dongle in our design) know that it is a connectable device, which consumes some power. When the device is running normally and keeps sending data, the pulses are much more frequent, so the average current consumption increases. When the device is in connected state, even if there is no application data to be sent or received, the two devices will still exchange link layer data to maintain the connection. In Fig. 4, one device is connected but not transmitting data, so the pulses amplitudes are the same with a fixed time period of 15 ms, while the other one is transmitting data continuously, the pulses are higher than the former one. Meanwhile, when sending data continuously, the SPI interface will be working and there are extra data processing 119

5 works to do, so the base line of channel 1 (yellow) is a little higher than channel 2. Fig. 5 shows a power consumption comparison test between the former design and the new design in power consumptionboth with the Tx power set to 0 dbm. The results are measured using a digital multimeter with the same sampling rate each time. Fig. 7. ECG and Respiration Fig. 7 shows the results of ECG and respiration, respectively. A labview project is developed to process the data received from the device: filter the noise and eliminate baseline drift by using five-point sliding average method and other algorithms, calculate parameters we need such as heart rate, and display ECG, respiration and temperature. Fig. 5. Current Consumption comparison under different sampling frequency According to the test results, we can draw a conclusion that our new device can work for more than 107 hours continuously without a recharge when powered by a lithium polymer battery with 1100 mah, which is about 4 times of the previous one of 26 hours, and completely meet the application of family health monitor. It saves about 75% power consumption than the former design. The circuit PCB of the device is shown in Fig. 6. Fig. 6. PCB of the system (SD card slot on the back) Functional performance of the device was tested. As body temperature is relatively easy to monitor, the result is not presented here. There are five lead wires in our design, three of which are used for ECG monitoring, including a right leg drive. In this test, respiration shares the two electrodes (LL, RA) of ECG which are placed on the chest horizontally. The other two electrodes are also designed to measure the respiration as we plan to extract the ICG (Impedance Cardiogram) from the two channels of respiration, which is the next step of our work. V. CONCLUSIONS This paper presented a design of an ultra-low power and highly-integrated portable health monitoring system capable of measuring a subject s ECG, respiration, and body temperature. The highlight of this design is for its ultra-low power consumption, which can save as much as 75% power consumption compared to our former design when realize the same functions. Power consumption was analyzed in detail, and we also tested the performance of the device. The results turned out to be satisfactory which demonstrate that the device is stable to monitor the ECG, respiration and body temperature for portable family use. REFERENCE [1] E. Sardini, M. Serpelloni*. Instrumented Wearable Belt for Wireless Health Monitoring, Procedia Engineering 5 (2010) [2] A. Wood, G. Virone, T. Doan, Q. Cao, L. Selavo,Y. Wu, L. Fang, Z. He, S. Lin, and J. Stankovic. Alarm-net: Wireless sensor networks for assisted-living and residential monitoring, Technical Report CS ,Dec [3] D. W. Curtis, E. J. Pino, J. M. Bailey, E. I. Shih,J. Waterman, S. A. Vinterbo, T. O. Stair, J. V. Guttag,R. A. Greenes, and L. Ohno- Machado. Smart an in-tegrated wireless system for monitoring unattended pa-tients, Journal of the American Medical Informatics Association, [4] J. Ko, R. Musăloiu-Elefteri, J. H. Lim, Y. Chen,A. Terzis, T. Gao, W. Destler, and L. Selavo. MEDiSN: medical emergency detection in sensor networks. In SenSys, [5] Fei Hu, Meng Jiang, Laura Celentano, Yang Xiao. Robust medical ad hoc sensor nerworks (MASN) with wavelet-based ECG data mining, Ad Hoc Networks 6 (2008) [6] Reza Naima, John Canny. The Berkeley Tricorder: Ambulatory Health Monitoring, Body Sensor Networks,2009. [7] Bluegiga Technologies, BLE112 Datasheet [8] Texas Instruments, ADS1292R Datasheet. [9] Texas Instruments, TI_BLE_Software_Developer's_Guide. [10] Marasco, Ken. Low-Noise LDOs Offer Advantages for Noise Sensitive Analog and RF Applications, Electronic Component News;Nov2010, Vol. 54 Issue 13, p14 View publication stats 120