资源描述
An Ultra-Wearable, Wireless, Low Power ECG Monitoring SystemChulsung Park and Pai H. Chou University of California,Irvine, CA 92697-2625 USAEmail: chulsung, phchouuci.eduYing Bai, Robert Matthews, and Andrew Hibbs Quantum Applied Science & Research (QUASAR),Inc.5764 Pacific Center Blvd., Suite 107 San Diego, CA 92121, USAAbstractWearableel ectrocar diograph(ECG) monitoring systemstoday useelectrod esthatrequireskin preparationin advance,andrequire pastes or gels to make electrical contact to the skin. Moreover, they are not suitable for subjects at high levels of activity due to high noise spikes that can appear in the data. To address these problems, a new class of miniature, ultra low noise, capacitive sensor that does not require direct contact to the skin, and has comparable performance to gold standard ECG electrodes, has been developed. This paper presents a description and evaluation of a wireless version of a system based on these innovative ECG sensors. We use a wearable and ultra low power wireless sensor node called Eco. Experimental results show that the wireless interface will add minimal size and weight to the system while providing reliable, untethered operation.I. INTRODUCTIONElectrocardiograph (ECG) is one of the most widely used biomedical sensing procedures to date. The heartbeat is the definitive indicator for a wide range of physiological conditions. Although ECG instruments were quite bulky, miniaturization in recent years has opened up brand new applications by enabling wearable versions to collect data in scenarios that were not possible before.A. Current ECG sensorsMany wearable ECG systemshave been proposed to date. Virtually all of them use some form of electrodes that must make electrical contact with the subjects skin surface. This necessitates the use of sticky pads, pastes or gel. While this method works for stationary patients, it suffers from several problems. First, the material used to construct the electrode or the paste could cause skin irritation and discomfort, especially if the subject is performing rigorous physical exercise and may be sweating. Another problem is that, during motion, the electrodes may become become loose, breaking electrical contact and causing high noise spikes in the data. Paste/gel-free resistive contact ECG sensors have been developed. Many of them still suffer from similar noise levels to “wet” electrodes, and the contact can still cause irritation problems as well as being more sensitive to motion.B. Insulated BioelectrodesRecent breakthroughs have been made in the form of insulated bioelectrodes (IBEs). They can measure the electric potential on the skin without resistive electrical contact and with very low capacitive coupling. This has been made possible by a combination of circuit design and the use of a new, low dielectric material. These IBEs enable through-clothing measurements, and results over 40 subjects have shown them to be capable of over 99% correlation with gold standard conventional electrodes.Fig. 1. QUASARs ECG Sensor and Eco shown with a dime coin for scaleC. Wireless Enabled IBEsTo realize the full potentials of these IBEs, they need to be built in a miniature, wearable form factor and be wirelessly enabled. To accomplish this, we are integrating an existing wireless sensor node platform with the IBE. These IBEs can be very sensitive to outside noise sources and adding a digital wireless interface could introduce noise in different forms. Therefore, an important consideration in this case is to test the data quality in such a setup.II. BACKGROUNDSeveralwireless ECG monitoring systems have been proposed.Allofthemuseconventional“wet”ECGsensors. For data sampling and wireless transmission, they use either existing standard wireless interfaces or general-purpose wireless sensor nodes. This combination results in many system-level drawbacks such as big form factor, low transmission speed, short battery lifetime, and lack of wearability. In this section, we first review the previous works and discuss their shortcomings. Next, we demonstrate the design goals of our system to resolve the discussed problems.A. Related WorkThe most recent work 1 is a system which uses the Tmote Sky platform 7 and a 3-lead system based on conventional ECG sensors. Tmote has an 802.15.4 radio interface (for Zigbee) at 250 Kbps and is controlled by the MSP430F1611 microcontroller. The authorsdesignedaninterfaceboardbetweenTmoteandECGsensors. Also, they use a PDA for data collection and Wi-Fi or GPRS for the host interface. This system samples one ECG sensor at 1 KHz (16-bit resolution). However, this system is too obtrusive to be worn,considering its sensor node is larger than 66(L) 32(W) 15(H) 3mm . Also, its conventional ECG sensors must be attached directly onto the skin, which is another main obstacle towards wearability.Other work done at Harvard University 2 used the MICA series and Telos platforms for their first prototypes. In order to make more integrated systems, they developed their own mote platform called Pluto. Pluto is functionally identical to Tmote Sky . It 3 measures 57(L) 36(W) 16(H) mm , which is still as large as Tmote Sky. It also uses standard ECG sensors.Researchers at Imperial College developed their own wireless sensor node, called the BSN node. It measures 28(L) 37(W) 3 12(H) mm (w/ a sensor board and w/o a battery). They used this platform to design a wireless ECG monitoring system . This system also used the 802.15.4 radio and conventional ECG sensors. Most systems use the 802.15.4 radio, even though it was originally developed for event-detection applications rather than real-time monitoring ones. There are two exceptions . The first one uses a CC1050 transceiver (at 76.8 Kbps max), which is similar to MICA2s transceiver. The other one uses a Bluetooth interface (721Kbps max). Neither radio interface was originally designed for realtime monitoring applications.B. Design GoalsTo address the problems descried above, we are designing a new ECG monitoring system. There are four main design goals: Ultra-Wearability, High Throughput, Low Power, Universal Connectivity. Ultra-Wearability Wearability is the most crucial issue in designing a wireless ECG monitoring system. However, to the best of our knowledge, none of the existing miniature sensing systems can be considered truly wearable in the strict sense, not just because they are still bulky but also because conventional ECG sensors can cause skin irritation. Therefore, we are using QUASARs innovative ECG sensor and an ultra-compact wireless sensor node specially designed or wearable applications. High Throughput The other design goal is to achieve high network throughput, which is necessary for a low latency/realtime monitoring system. We decided to use a 1 Mbps proprietary radio instead of 802.15.4. Although in-sensor processing can reduce bandwidth demand, our chosen faster radio with a simpler Media Access Control (MAC) is actually more energy efficient, as discussed next. Low Power Low power consumption is another highly important design goal. Low power consumption contributes not only to prolonged lifetime, but also to system miniaturization, because the size of a battery occupies more than 50% of system volume. It is well known that the most power hungry component in a wireless monitoring system is the wireless transceiver. Therefore, we carefully chose a very low power transceiver that consumes less than 10 mA in transmission mode (1 Mbps, 0 dBm) and 22 mA in receiving mode.Universal Connectivity Universal connectivity means that the nodes should able to connect to virtually any computer on one of its communication interfaces. This versatility is necessary for applications that not only collect and record or replay data, but also integrate them with actuators and other infrastructures. Thus, wedesigned our system to be able to transmit data via most common communication interfaces including USB, Ethernet, and Wi-Fi.III. SYSTEM DESIGNIn order to achieve the design goals described in the previous section, we are developing a new ECG monitoring system that takes advantage of QUASARs ECG sensors 11 and Eco wireless sensor nodes 12. The QUASAR sensor is a wearable, tiny, low-power ECGsensingdevice,andEcoisanultra-compact,low-powerwireless sensor node. Fig. 1 (a) and (b) show the QUASAR ECG sensor and Eco with a US dime coin for reference of scale. They are similar in size, power, and are well matched in terms of data rate. This section first shows various system architectures for an ECG monitoring system that integratesthesetwotechnologies.Next,weexaminethespecifications of the QUASAR ECG sensor, Eco node,andthe base station in detail. Fig. 2. System Architecture of ECG Monitoring SystemA. System ArchitecturesOur ECG monitoring system can be functionally divided into four subsystems: ECG Sensors, Data Sampling, Wireless Transmission, and Host Interface. ECG signals are first digitized by ADCs andtransmitted wirelessly to a base station that interfaces with a host computer via USB, Fast Ethernet, or 802.11b. We propose three different system architectures as shown in Fig. 2. The first architecture Fig. 2(a) consists of multiple ECG sensors, a data-sampling module, an Eco node, and a base station. In this architecture, the system has a separate data-sampling module, which contains a microcontroller unit (MCU) and ADCs as shown in Fig. 1(c). All signals from ECG sensors are first sampled and buffered in this module. Then, data are fed to Eco via SPI and transmitted wirelessly to the base station. This architecture uses two MCUs to distribute workload. The MCU in the data-sampling module is dedicated to sampling signals. The other MCU in the Eco node handles wireless data transmission. By adopting this two-MCU architecture,we can achieve very accurate data sampling (low jitter) as well as high communication throughput and low latency 13. In addition, the system can have separate high resolution ADCs and increase monitoring granularity without changing the main MCU. However, this architecture imposes extra cost and volume for the data-sampling module, and it also needs a data transmission protocol between the two MCUs. Fig. 3. (a) QUASAR ECG sensor worn by subject, (b) Data Comparison:Green trace is from QUASAR sensor, Blue trace is from conventional electrodeThe second architecture Fig. 2(b) uses the Eco for both datasampling and wireless transmission. All the ECG signals are directly fed to the ADC channels on the Eco node. The ADC built into the main MCU on the Eco node digitizes and transmits data according to its timing requirements. In this architecture, one Eco has to run all sensing and communication tasks by itself. Therefore, systemperformance is degraded as the number of ECG sensors increases. Another system issue is that the number of ECG sensors that can be monitored simultaneously is limited by the number of Ecos ADC channels (8 channels in this case). Multiplexing also means that the samples on multiple channels cannot be taken precisely at the same time. However, this is a very simple and low cost design, which is suitable when the system is equipped with only two or three ECG sensors. The third architecture Fig. 2(c) uses one Eco node for each ECG sensor. This architecture is similar to the second one in the sense that the Eco node performs both sampling and communication tasks, except that an Eco node serves only one ECG sensor in this case. Therefore, we can achieve higher throughput and lower latency as well as lower jitter. This is also a more reliable and distributed architecture than the other two. However, we need a sophisticated MAC protocol to coordinate the wireless traffic among several Ecos. The cost will increase in proportion to the number of monitoring points.B. ECG SensorQUASARs sensor (Fig. 3) is a compact ECG sensor that does not require skin preparation, gels, or adhesives. It includes not only a sensing device, but also signal conditioning circuitry such as lownoise amplifiers and voltage reference chips. Its output signal range is adjustable from differential (-4.5 V to 4.5 V) to single-ended (0 V to 4.5 V). It measures only 15 mm (in diameter) 3.8 mm (in height) and weighs 5 g. Also, it consumes only 1 mW active power on average. These features enable our monitoring system to betruly wearable. This sensor measures ECG signals using capacitively coupled electrodes that do not require ohmic contact. As shown in Fig. 4, QUASARs sensors have at least equivalent and often superior signal quality and artifact rejection compared to the standard wet/resistive ECG sensors.Fig. 4. Eco Dimensions: 13(L) 11(W) 7(H) mmFig. 5 Eco PCB: (a) Top View, (b) Bottom View C. Eco Wireless Sensor Node Eco is an ultra-compact and low power wireless sensor node developed by the coauthors at UC Irvine. It measures only 13 mm(L) 11 mm(W) 7 mm(H) and weighs 2 grams (Fig. 5). Also, it consumes less than 10 mA in transmission mode (0 dBm) and 22 mA in receiving mode. Its maximum data rate and RF range are 1 Mbps and 10 m, respectively. Considering its small form factor and low power consumption, Eco is very suitable for real-time biomedical signal monitoring applications, which require relatively high throughput, low latency, and high wearability.Fig. 6 shows photos of the Eco hardware. Eco uses Nordic VLSIs nRF24E1, a 2.4 GHz RF transceiver with an embedded 8051-compatible MCU (DW8051). The MCU has a 512-byte ROM for a bootstrap loader, a 4 KB RAM for the user program, SPI (3-wire), RS-232, and a 9-channel ADC. The ADC is software-configurable for 612 bits of resolution. A 32 KB serial (SPI) EEPROM stores the application program. The nRF24E1s 2.4 GHz transceiver uses a GFSK modulation scheme with 125 frequency channels that are1 MHz apart. The transmission output power is also software-configurable for four different levels: -20 dBm, -10 dBm, -5dBm, and 0 dBm. The RainSun chip antenna (AN9520) measures 9.5 mm(H) 1.5 mm(W) 1 mm(H) and has a maximum gain of 1.5 dBi.Fig. 6. Base Stations for ECG monitoring SystemIn addition, Eco has a 3-axial acceleration sensor, Hitachi-Metals H34C, which measures acceleration from -3gto +3 g and temperature from 0 75?C. Eco also has a light sensor (S1087). Ecos power subsystem includes a regulator (LTC3410), battery protection circuitry, and a custom 40 mAh rechargeable Li-Polymer battery. LTC3410 is an adjustable boost regulator whose output voltage is set to 2.7 V. Its maximum output current is 300 mA, and its average efficiency over a Li-Polymer batterys output voltage range (3.0 V 4.2 V) is higher than 90% at Ecos maximum operating current of 30 mA.Eco has a flexible-PCB type expansion port that has 16 pins. This expansion port includes two digital I/O pins, two analog input lines, SPI, RS232, and voltage inputs for a regulator and battery charging. This port enables Eco to interface with other sensing devices such an image sensor, gyroscope, pressure sensor, or compass. We can charge the battery and program the EEPROM via this expansion port.D. Base StationAs shown in Fig. 7, we have developed three different types of base stations: USB, Fast Ethernet,and 802.11b/g.The USB one uses nRF24E1 as a wireless transceiver and MCU. Also, it uses Silicon Labs CP2102 UART-to-USB bridge (Max. 12 Mbps). The Fast Ethernet one uses Freescales MC9S12NE64, which has an HCS12 16-bit core and the Fast Ethernet interface (100 Mbps). It uses the nRF2401 as its wireless transceiver. The 802.11b/g base station use a PIC18F8720 MCU, nRF2401 transceiver and Linksys WCF12 CF 802.11b card (Max. 11 Mbps).IV. EVALUATIONIn this section, we evaluate our ECG monitoring system in terms of size, weight, power consumption, sampling rate, and latency. We considered the second system architecture (in Fig. 2), which has the lowest hardware cost among the three architectures. It also represents the lower bound in terms of performance and data quality.A Size, Weight, and Power Consumption.The total size and weight of the monitoring system are simply the sumofthoseoftheECGsensorandEco,becauseoursystemdoesnot require interface circuitry. Note that the signal conditioning circuitry is already included inside the ECG sensor. The total size of our 3 system is about 26(L) 15(W) 7(H) mm including one sensor, a wireless transceiver, and a battery. To the best of our knowledge, our system is the smallest wireless ECG monitoring system to date. The entire system weighs less than 17 grams and consists of one Eco plus three ECG sensors. Our ECG system consumes less than 30 mW max while in operation. When it is powered by the 40 mAh Li-Polymer battery, it can last for more than 12 hours at 20% duty cycle, 1 Kbps sampling rate.B. Sampling Rate and LatencyInthesecondsystemarchitecture,threeECGsensorsareconnected to one Eco. The size of one data sample is 12 bytes (4 bytes per sensor, 12-bit resolution, differential input). Our system can sustain sampling at 1 KHz. The measured latency is under 300 s.V. CONCLUSIONSWe are developing a truly wearable, wireless ECG monitoring system. Our system integrates novel capacitive ECG sensors, which have demonstrated 99% correlation with conventional electrodes, with Eco, an ultra-compact, low power wireless sensor node. Future work includes tighter integration of QUASARs sensor and Eco and improving wireless performance and power efficiency.一种超耐磨,低功耗的无线心电监护摘要如今使用的穿戴式电极心电图(ECG)监测系统,使用前需要让皮肤适应, 并要求使电极接触并粘到皮肤上。此外,他们都不适合在剧烈活动时使用,因为这会导致高噪声干扰。为了解决这些问题,一种新型的低噪声,可全方位显示的电极式心电图系统已经研制成功,它应用了电容式传感器,并且不需要直接与皮肤接触。本文介绍了描述和评价一种基于创新式心电图传感器的无线系统。我们使用了耐磨,超低功率无线传感器节点,称为Eco。实验结果表明,此系统系统可以提供可靠, 无线服务,还可以减小尺寸和重本文节选自美国加州大学发表的An Ultra-Wearable, Wireless, Low Power ECG Monitoring System(一种超耐磨,低功耗的无线心电监护系统)I.导言心电图测试仪( ECG )的是世界上使用广泛的传感仪器。心跳是一系列的生理状况检查的通用指标。原来的心电图仪器比较笨重, 近几年小型心电图仪器发展迅速,可以完成如采集数据等以前不可能的事情。A.当前的心电图传感器许多穿戴式心电图检测系统的研制已提上日程。几乎的心电图检测系统都需要将电极触头与受测者的皮肤表面接触。因此需要使用凝胶等物品。这虽然把电极触头固定住了,但也引发了其它问题。 第一,用来制造电极的材料或凝胶可能引起皮肤红肿等不良状况,尤其使用者是进行体育活动而出汗时。另一个问题是,在使用过程中,电极可能变得松散,导致电气隔离的失效,产生噪声误差。 浆糊/凝胶自由接触式电阻心电图传感器已经开发出来,但仍存在噪声,接触电极引起刺激性的问题. B. 绝缘电极最近绝缘式电极(IBEs)的研究取得了突破性的进展。他们可以依靠电容耦合作用测量皮肤的电位而不会产生电气接触问题。使用新的介电材料,结合电路的设计,这已变为可能。这种绝缘式电极安在衣服上,在40种使用方式中的统计结果表明其合格率能够达到99%以上。图1 钱币大小的星型心电图传感器C.无线绝缘式电极(IBEs) 为使IBEs发挥出全部潜力,需要建立一个穿戴形式,无线通信的模板。为做到这一点,我们整合了现有的无线传感器节点平台与IBEs。这些IBEs对外界噪声很敏感,添加的数字无线接口会引入不同形式的噪声。因此,在这种情况下,测试数据质量是一个重要的考虑因素。II.研究背景已经出现了几种无线心电图监测系统模型。它们使用了传统的湿心电图传感器。为实现数据采集和无线传输,它们使用了现有的标准无线接口和通用无线传感器节点。 这一组合,使系统产生了许多缺点,传输速率低,电池寿命短,缺乏耐磨性等。在这一节中,我们首先回顾了以往的工作,并分析了了它们的缺点。其次,我们论述了设计的目标-解决上述问题。 A. 有关成果最近的成果是利用tmote Sky platform与3-lead系统设计的常规心电图传感器系统。Tmote是由msp430f1611单片机控制,使用802.15.4无线接口(无线),传输速率为250 Kbps。作者设计了接口板与传感器之间的连接。它们应用PDA的数据采集卡和Wi-Fi或GPRS网络的主机接口。这一系统的心电传感器在1千赫(16位元解析度)。但是,为减少磨损,其节点大于66 (L) 32(W)15(H)mm3,使它过于笨重。同时,常规心电图传感器另一个主要问题是必须直接与皮肤接触。 哈佛大学在其他工作中,使用MICA系列与Telos平台原型机。 为了建立更多的综合服务体系,建立了自己的平台,并命名为Pluto。Pluto与tmote Sky(7)功能完全相同。其体积为57(L)36(W)16(H)mm3。它使用的是标准心电传感器。帝国学院的研究员建立了称为BSN的无线传感器节点。其尺寸为28 (L)37 (W)12(H)mm3(宽/应变片和W/电池)。他们利用这一平台,设计了无线心电图监测系统(3)。这个系统使用了802.15.4无线系统与常规心电图传感器。 大多数系统使用802.15.4无线系统,虽然它最初的研制目的是事件检测,而非实时监控。第一种使用了cc1050收发(10)(76.8 Kbps),这类似于MICA2s收发器,另一个使用了蓝牙接口(最高为721Kbps),而无线接口原本用于实时监测。 B. 设计目标为解决上述情况,我们正在设计一种新的心电监测系统。四个主要设计目标为:超耐磨,高吞吐量,低功率,通用性。超耐磨性-耐磨性是设计了无线心电图监测系统时的关键的问题。不过,据我们所知,在严格的意义上讲,现有的微型传感系统不算真正耐磨。它们仍然很笨重,而且常规心电图传感器会导致皮肤红肿。因此,我们正利用星型的心电图传感器和紧凑型无线传感器进行设计。 高吞吐量-另一设计目标是要实现网络高吞吐量,这对要求瞬时/实时监控的系统是必须的。我们决定1Mbps的专有无线代替802.15.4。虽然用传感器进行处理可以减少对带宽的需求,但我们选择了更有效,简单的媒体存取控制( MAC )技术。 低功耗-低功耗是另一个非常重要的设计目标。低功耗有利于延长寿命和实现系统的小型化,因为电池占系统体积的50%以上。人所共知无线收发器大都由无线监控系统组成。因此,我们选择了低功耗,收发消耗低于10马传输模式(每秒1兆字节,0DBM)和22mA的。 通用性-通用性意
展开阅读全文