Journal of Mobile Multimedia, Vol. 1, No.4 (2006) 307-326
`© Rinton Press
`
`
`
`
`
`
`SYSTEM ARCHITECTURE OF A WIRELESS BODY AREA SENSOR NETWORK
`FOR UBIQUITOUS HEALTH MONITORING
`
`CHRIS OTTO, ALEKSANDAR MILENKOVIĆ, COREY SANDERS, EMIL JOVANOV
`University of Alabama in Huntsville
`chrisaotto@yahoo.com, {milenka | jovanov}@ece.uah.edu, sanderscorey@yahoo.com
`
`Received October 1, 2005
`Revised January 10, 2006
`
`
`Recent technological advances in sensors, low-power microelectronics and miniaturization, and wireless
`networking enabled the design and proliferation of wireless sensor networks capable of autonomously
`monitoring and controlling environments. One of the most promising applications of sensor networks is for
`human health monitoring. A number of tiny wireless sensors, strategically placed on the human body,
`create a wireless body area network that can monitor various vital signs, providing real-time feedback to
`the user and medical personnel. The wireless body area networks promise to revolutionize health
`monitoring. However, designers of such systems face a number of challenging tasks, as they need to
`address often quite conflicting requirements for size, operating time, precision, and reliability.
`
`In this paper we present hardware and software architecture of a working wireless sensor network system
`for ambulatory health status monitoring. The system consists of multiple sensor nodes that monitor body
`motion and heart activity, a network coordinator, and a personal server running on a personal digital
`assistant or a personal computer.
`
`Key words: Wireless sensors, body area networks, health monitoring, wearable computing.
`Communicated by: A Ganz & R Istepanian
`
`
`Introduction
`1
`integration and
`in wireless networking, microelectronics
`Recent
`technological advances
`miniaturization, sensors, and the Internet allow us to fundamentally modernize and change the way
`health care services are deployed and delivered. Focus on prevention and early detection of disease or
`optimal maintenance of chronic conditions promise to augment existing health care systems that are
`mostly structured and optimized for reacting to crisis and managing illness rather than wellness [6].
`The anticipated change and emerging new services are well-timed to help cope with the imminent
`crisis in the health care systems caused by current economic, social, and demographic trends. The
`overall health care expenditures in the United States reached $1.8 trillion in 2004, though almost 45
`million Americans do not have health insurance [15]. On the other hand, many companies have
`already been plagued by high-rising costs of healthcare liabilities. With current trends in healthcare
`costs, it is projected that the total health care expenditures will reach almost 20% of the Gross
`Domestic Product (GDP) in less then 10 years from now, threatening the wellbeing of the entire
`economy. The demographic trends are indicating two significant phenomena: an aging population due
`to increased life expectancy and Baby Boomers demographic peak. Life expectancy has significantly
`increased from 49 years in 1901 to 77.6 years in 2003. According to the U.S. Bureau of the Census,
`
`Zepp Labs, Inc.
`ZEPP 1012
`Page 1
`
`

`
`
`308 System Architecture of WBAN for Ubiquitous Health Monitoring
`the number of elderly over age 65 is expected to double from 35 million to nearly 70 million by 2025
`when the youngest Baby Boomers retire [21]. This trend is global, so the worldwide population over
`age 65 is expected to more than double from 357 million in 1990 to 761 million in 2025. These
`statistics underscore the need for more scalable and more affordable health care solutions.
`Wearable systems for continuous health monitoring are a key technology in helping the transition
`to more proactive and affordable healthcare. They allow an individual to closely monitor changes in
`her or his vital signs and provide feedback to help maintain an optimal health status. If integrated into
`a telemedical system, these systems can even alert medical personnel when life-threatening changes
`occur. In addition, the wearable systems can be used for health monitoring of patients in ambulatory
`settings [7]. For example, they can be used as a part of a diagnostic procedure, optimal maintenance of
`a chronic condition, a supervised recovery from an acute event or surgical procedure, to monitor
`adherence to treatment guidelines (e.g., regular cardiovascular exercise), or to monitor effects of drug
`therapy.
`During the last few years there has been a significant increase in the number and variety of
`wearable health monitoring devices, ranging from simple pulse monitors, activity monitors, and
`portable Holter monitors, to sophisticated and expensive implantable sensors. However, wider
`acceptance of the existing systems is still limited by the following important restrictions.
`Traditionally, personal medical monitoring systems, such as Holter monitors, have been used only to
`collect data. Data processing and analysis are performed offline, making such devices impractical for
`continual monitoring and early detection of medical disorders. Systems with multiple sensors for
`physical rehabilitation often feature unwieldy wires between the sensors and the monitoring system.
`These wires may limit the patient's activity and level of comfort and thus negatively influence the
`measured results [12]. In addition, individual sensors often operate as stand-alone systems and usually
`do not offer flexibility and integration with third-party devices. Finally, the existing systems are rarely
`made affordable.
`One of the most promising approaches in building wearable health monitoring systems utilizes
`emerging wireless body area networks (WBANs) [8]. A WBAN consists of multiple sensor nodes,
`each capable of sampling, processing, and communicating one or more vital signs (heart rate, blood
`pressure, oxygen saturation, activity) or environmental parameters (location, temperature, humidity,
`light). Typically, these sensors are placed strategically on the human body as tiny patches or hidden in
`users’ clothes allowing ubiquitous health monitoring in their native environment for extended periods
`of time.
`A number of recent research efforts focus on wearable systems for health monitoring. Researchers
`at the MIT Media Lab have developed MIThril, a wearable computing platform compatible with both
`custom and off-the-shelf sensors. The MIThril includes ECG, skin temperature, and galvanic skin
`response (GSR) sensors. In addition, they demonstrated step and gait analysis using 3-axis
`accelerometers, rate gyros, and pressure sensors [18]. MIThril is being used to research human
`behaviour recognition and to create context-aware computing interfaces [5]. CodeBlue, a Harvard
`University research project, is also focused on developing wireless sensor networks for medical
`applications. They have developed wireless pulse oximeter sensors, wireless ECG sensors, and tri-
`axial accelerometer motion sensors. Using these sensors, they have demonstrated the formation of ad-
`hoc networks. The sensors, when outfitted on patients in hospitals or disaster environments, use the
`ad-hoc networks to transmit vital signs to healthcare givers, facilitating automatic vital sign collection
`and real-time triage [19,9].
`
`
`
`
`
`Zepp Labs, Inc.
`ZEPP 1012
`Page 2
`
`

`
`
`
`C. Otto, A. Milenkovic, C. Sanders, and E. Jovanov 309
`
`In this paper we describe a general WBAN architecture and how it can be integrated into a broader
`telemedical system. To explore feasibility of the proposed system and address open issues we have
`designed a prototype WBAN that consists of a personal server, implemented on a personal digital
`assistant (PDA) or personal computer (PC), and physiological sensors, implemented using off-the-
`shelf sensor platforms and custom-built sensor boards. The WBAN includes several motion sensors
`that monitor the user’s overall activity and an ECG sensor for monitoring heart activity. We describe
`the hardware and software organization of the WBAN prototype.
`The rest of the paper is organized as follows. Section 2 outlines the general WBAN architecture,
`defines the role of each component, and describes its integration into a broader telemedical system.
`Section 3 presents a case study, walking through a typical system deployment and its use. Section 4
`describes the hardware architecture. Section 5 details the software architecture of the personal server
`and sensor nodes, and introduces energy-efficient WBAN communication protocol. Section 6
`concludes the paper and discusses possible future research directions.
`
`2 System Architecture
`The proposed wireless body area sensor network for health monitoring integrated into a broader multi-
`tier telemedicine system is illustrated in Figure 1. The telemedical system spans a network comprised
`of individual health monitoring systems that connect through the Internet to a medical server tier that
`resides at the top of this hierarchy. The top tier, centered on a medical server, is optimized to service
`hundreds or thousands of individual users, and encompasses a complex network of interconnected
`services, medical personnel, and healthcare professionals. Each user wears a number of sensor nodes
`that are strategically placed on her body. The primary functions of these sensor nodes are to
`unobtrusively sample vital signs and transfer the relevant data to a personal server through wireless
`personal network implemented using ZigBee (802.15.4) or Bluetooth (802.15.1). The personal server,
`implemented on a personal digital assistant (PDA), cell phone, or home personal computer, sets up and
`controls the WBAN, provides graphical or audio interface to the user, and transfers the information
`about health status to the medical server through the Internet or mobile telephone networks (e.g.,
`GPRS, 3G).
`The medical server keeps electronic medical records of registered users and provides various
`services to the users, medical personnel, and informal caregivers. It is the responsibility of the medical
`server to authenticate users, accept health monitoring session uploads, format and insert this session
`data into corresponding medical records, analyze the data patterns, recognize serious health anomalies
`in order to contact emergency care givers, and forward new instructions to the users, such as physician
`prescribed exercises. The patient’s physician can access the data from his/her office via the Internet
`and examine it to ensure the patient is within expected health metrics (heart rate, blood pressure,
`activity), ensure that the patient is responding to a given treatment or that a patient has been
`performing the given exercises. A server agent may inspect the uploaded data and create an alert in
`the case of a potential medical condition. The large amount of data collected through these services
`can also be utilized for knowledge discovery through data mining. Integration of the collected data
`into research databases and quantitative analysis of conditions and patterns could prove invaluable to
`researchers trying to link symptoms and diagnoses with historical changes in health status,
`physiological data, or other parameters (e.g., gender, age, weight). In a similar way this infrastructure
`could significantly contribute to monitoring and studying of drug therapy effects.
`The second tier is the personal server that interfaces WBAN sensor nodes, provides the graphical
`user interface, and communicates with services at the top tier. The personal server is typically
`
`
`
`Zepp Labs, Inc.
`ZEPP 1012
`Page 3
`
`

`
`
`310 System Architecture of WBAN for Ubiquitous Health Monitoring
`implemented on a PDA or a cell phone, but alternatively can run on a home personal computer. This
`is particularly convenient for in-home monitoring of elderly patients. The personal server interfaces
`the WBAN nodes through a network coordinator (nc) that implements ZigBee or Bluetooth
`connectivity. To communicate to the medical server, the personal server employs mobile telephone
`networks (2G, GPRS, 3G) or WLANs to reach an Internet access point.
`
`
`
`WBANWBAN
`
`(Zigbee, Bluetooth)(Zigbee, Bluetooth)
`
`
`WWANWWAN
`
`WLANWLAN
`
`
`
`EE
`
`
`
`AA
`
`
`
`AA
`
`
`
`AA
`
`
`Tier 1:Tier 1:
`
`WBASNWBASN
`
`
`
`……
`
`
`User2User2User2
`
`User1User1User1
`
`WWANWWAN
`
`(GPRS)(GPRS)
`
`
`
`UserNUserNUserN
`
`
`Tier 3:Tier 3:
`
`MSMS
`
`
`Medical Medical
`
`ServerServer
`
`
`InternetInternet
`
`InternetInternet
`
`
`HealthcareHealthcare
`
`providerprovider
`
`
`Informal Informal
`
`caregivercaregiver
`
`
`
`ncnc
`
`
`Tier 2:Tier 2:
`
`PSPS
`
`
`WLANWLAN
`
`(Wi-Fi)(Wi-Fi)
`
`
`
`
`
`………
`
`
`
`
`
`UserNUserNUserNUserN
`
`
`
`
`
`User2User1User2User1User2User1
`
`
`
`
`
`
`
`EmergencyEmergency
`
`Figure 1 Health Monitoring System Network Architecture
`
`
`
`
`The interface to the WBAN includes the network configuration and management. The network
`configuration encompasses the following tasks: sensor node registration (type and number of sensors),
`initialization (e.g., specify sampling frequency and mode of operation), customization (e.g., run user-
`specific calibration or user-specific signal processing procedure upload), and setup of a secure
`communication (key exchange). Once the WBAN network is configured, the personal server manages
`the network, taking care of channel sharing, time synchronization, data retrieval and processing, and
`fusion of the data. Based on synergy of information from multiple medical sensors the PS application
`should determine the user’s state and his or her health status and provide feedback through a user-
`friendly and intuitive graphical or audio user interface.
`The personal server holds patient authentication information and is configured with the medical
`server IP address in order to interface the medical services. If the communication channel to the
`medical server is available, the PS establishes a secure communication to the medical server and sends
`reports that can be integrated into the user’s medical record. However, if a link between the PS and
`the medical server is not available, the PS should be able to store the data locally and initiate data
`
`
`
`
`
`Zepp Labs, Inc.
`ZEPP 1012
`Page 4
`
`

`
`C. Otto, A. Milenkovic, C. Sanders, and E. Jovanov 311
`
`
`uploads when a link becomes available. This organization allows full mobility of users with secure and
`near real time health information uploads.
`A pivotal part of the telemedical system is tier 1 – wireless body area sensor network. It
`comprises a number of intelligent nodes, each capable of sensing, sampling, processing, and
`communicating of physiological signals. For example, an ECG sensor can be used for monitoring
`heart activity, an EMG sensor for monitoring muscle activity, an EEG sensor for monitoring brain
`electrical activity, a blood pressure sensor for monitoring blood pressure, a tilt sensor for monitoring
`trunk position, and a breathing sensor for monitoring respiration, while the motion sensors can be used
`to discriminate the user’s status and estimate her or his level of activity.
`Each sensor node receives initialization commands and responds to queries from the personal
`server. WBAN nodes must satisfy requirements for minimal weight, miniature form-factor, low-
`power consumption to permit prolonged ubiquitous monitoring, seamless integration into a WBAN,
`standards based interface protocols, and patient-specific calibration, tuning, and customization. The
`wireless network nodes can be implemented as tiny patches or incorporated into clothes or shoes. The
`network nodes continuously collect and process raw information, store them locally, and send
`processed event notifications to the personal server. The type and nature of a healthcare application
`will determine the frequency of relevant events (sampling, processing, storing, and communicating).
`Ideally, sensors periodically transmit their status and events, therefore significantly reducing power
`consumption and extending battery life. When local analysis of data is inconclusive or indicates an
`emergency situation, the upper level in the hierarchy can issue a request to transfer raw signals to the
`next tier of the network.
`Patient privacy, an outstanding issue and a requirement by law, must be addressed at all tiers in the
`healthcare system. Data transfers between a user’s personal server and the medical server require
`encryption of all sensitive information related to the personal health [22]. Before possible integration
`of the data into research databases, all records must be stripped of all information that can tie it to a
`particular user. The limited range of wireless communications partially addresses security within
`WBAN; however, the messages can be encrypted using either software or hardware techniques. Some
`wireless sensor platforms have already provided a low power hardware encryption solution for ZigBee
`communications [14].
`
`3 Case Study
`In this section we present a hypothetical case study to illustrate the usefulness of our proposed system.
`The patient presented is fictitious, but representative of common issues a recovering heart attack
`patient would face. We discuss the issues and describe how our system can be used to both address the
`problem and provide advantages over typical present day solutions.
`Juan Lopez is recovering from a heart attack. After the release from the hospital he attended
`supervised physical rehabilitation for several weeks. His physicians prescribed an exercise regime at
`home. During the physical rehabilitation it was easy to monitor Juan and verify he completed his
`exercises. Sadly, when left to his own self-discipline, he does not rigorously follow the exercise as
`prescribed. He exercises, but is not honest to himself (or his physician) as to the intensity and duration
`of the exercise. As a result, Juan’s recovery is slower than expected which raises concerns about his
`health prognosis, and his physician has no quantitative way to verify Juan’s adherence to the program.
`Our health monitoring system offers a solution for Juan. Equipped with a WBAN, tiny sensors
`provide constant observation of vital statistics, estimate induced energy expenditure, and assist Juan’s
`
`
`
`Zepp Labs, Inc.
`ZEPP 1012
`Page 5
`
`

`
`
`312 System Architecture of WBAN for Ubiquitous Health Monitoring
`exercise. Tiny electronic inertial sensors measure movement while electrodes on the chest can
`measure Juan’s heart activity. The time, duration, and level of intensity of the exercise can be
`determined by calculating an estimate of energy expenditure from the motion sensors. Through the
`Internet, his physician can collect and review data, verify Juan is exercising regularly, issue new
`prescribed exercises, adjust data threshold values, and schedule office visits. Juan’s physician need
`not rely on Juan’s testament, but can quantify his level and duration of exercise. In addition, Juan’s
`parameters of heart rate variability provide a direct measure of his physiological response to the
`exercise serving as an in-home stress test. Substituting these remote stress tests and data collection for
`in-office tests, Juan’s physician reduces the number of office visits. This cuts healthcare costs and
`makes better use of the physician’s time. In urgent cases, however, the personal server can directly
`contact Emergency Medical Services (EMS) if the user subscribes to this service. Figure 2 illustrates
`one possible data flow.
`
`MS
`
`Internet
`Internet
`
`nc
`
`PS
`1. Events and
`Data are Collected
`
`2. Relayed to MS
`
`5. Relayed to PS
`
`3. Physician can retrieve
`and analyze data
`
`4. Based on analysis,
`physician recommends
`patient increase exercises
`
`
`
`AAA
`
`6. Juan can review
`new prescribed exercises.
`Figure 2 Data flow in the proposed healthcare monitoring system
`
`
`
`WBAN
`
`
`
`AAA
`
`
`
`EEE
`
`
`
`AAA
`
`4 Hardware Architecture
`In the spirit of the system architecture presented in Section 2, we have developed a prototype
`healthcare monitoring system. Figure 3 shows a photograph of the prototype components. The fully
`operational prototype system includes two activity sensors (ActiS), an integrated ECG and tilt sensor
`(eActiS), and a personal server. Each sensor node includes a custom application specific board and
`uses the Tmote sky platform [16] for processing and ZigBee wireless communication. The personal
`server runs either on a laptop computer or a WLAN/WWAN-enabled handheld PocketPC. The
`network coordinator with wireless ZigBee interface is implemented on another Tmote sky that
`connects to the personal server through a USB interface. For an alternative setting we have developed
`a custom network coordinator that features the ZigBee wireless interface, an ARM processor, and a
`compact flash interface to the personal server (Figure 3).
`The Tmote sky from Moteiv acts as the primary embedded platform for all sensors in our system.
`Each Tmote sky board utilizes Texas Instrument’s MSP430F1611 microcontroller and Chipcon’s
`
`
`
`
`
`Zepp Labs, Inc.
`ZEPP 1012
`Page 6
`
`

`
`C. Otto, A. Milenkovic, C. Sanders, and E. Jovanov 313
`
`
`CC2420 radio interface. The microcontroller is based around a 16-bit RISC core integrated with 10
`KB of RAM and 48 KB of flash memory, analog and digital peripherals, and a flexible clock
`subsystem. It supports several low-power operating modes and consumes as low as 1 µA in standby
`mode; it also has very fast wake up time of 6 µs. The CC2420 wireless transceiver is IEEE 802.15.4
`compliant and has programmable output power, maximum data rate of 250 Kbps, and hardware
`support for error correction and 128-bit encryption. The CC2420 is controlled by the MSP430
`microcontroller through the Serial Peripheral Interface (SPI) port and a series of digital I/O lines with
`interrupt capabilities. The Tmote sky platform features a 10-pin expansion connector with one
`Universal Asynchronous Receiver Transmitter (UART) and one I2C interface, two general-purpose I/O
`lines, and three analog input lines.
`The activity sensor, ActiS, consists of the Tmote sky platform and an Intelligent Activity Sensor
`(IAS), implemented as a daughter card. The IAS monitors motion using two dual-axis accelerometers
`arranged to provide three orthogonal motion axes (X, Y, Z). The IAS utilizes an on-board
`MSP430F1232 microcontroller for pre-processing and filtering of sampled data. The IAS connects to
`the Tmote sky platform via the extension header and sends digital sensor data using a simple serial
`communication protocol.
`
`Figure 3 Prototype WBAN. From left to right:
`the Personal Server with Network Coordinator, ECG sensor with electrodes, and a motion sensor.
`
`
`
`
`Similarly, the integrated ECG and tilt sensor (eActiS) consists of the Tmote sky platform and an
`intelligent signal processing module (ISPM). The ISPM is similar to the ActiS IAS board, but includes
`
`
`
`Zepp Labs, Inc.
`ZEPP 1012
`Page 7
`
`

`
`
`314 System Architecture of WBAN for Ubiquitous Health Monitoring
`a single-channel bio-amplifier for three-lead ECG/EMG. Electrodes are connected and placed on the
`chest for monitoring heart activity. When the sensor is worn on the chest, it also serves as an upper
`body tilt sensor.
`5
`Software Architecture
`In this section we describe the software architecture of the prototype WBAN system, illustrated in
`Figure 4. It encompasses software modules running on the IAS/ISMP, the Tmote sky platform, the
`network coordinator, and the personal server. Our focus has been on developing solutions for real-
`time on-sensor processing, WBAN communications, time synchronization [16], maximizing battery
`life [13], managing data and events, and an easy to use user interface. These issues relate to the lower
`tiers of the network, and as such we describe our prototype software for the WBAN.
`5.1 Sensor Node Software
`The sensor node software samples and collects physiological data, analyzes the signals in real-time,
`and transmits the results wirelessly to the personal server. In our prototype this software runs on the
`Tmote sky platform and custom application specific daughter cards. We have developed software for
`two types of sensors. An Activity Sensor (ActiS) samples three-axis accelerometers to determine
`orientation, type of activity (walking, sitting, etc.), estimates activity induced energy expenditure
`(AEE) based on an algorithm proposed by Bouten, et. al. [1], and performs step detection in real-time.
`An ECG and tilt sensor (eActiS) monitors heart activity and samples a two-axis accelerometer for
`orientation (upper body tilt). Sensor node and network coordinator software is implemented in the
`TinyOS environment.
`
`ActiS
`(IAS/ISPM)
`
`
`Tmote skyTmote sky
`
`InterfaceInterface
`
`Filtering/Filtering/
`
`Pre-processingPre-processing
`
`Data Data
`
`AcquisitionAcquisition
`
`
`
`PS
`(PDA)
`
`User InterfaceUser Interface
`
`Network CoordinatorNetwork Coordinator
`
`(Telos)(Telos)
`
`ActiSActiS
`
`WWAN/WLANWWAN/WLAN
`
`ProtocolProtocol
`
`CommunicationCommunication
`
`
`
`Messaging ControlMessaging Control
`
`
`InterfaceInterface
`
`(USB/CF)(USB/CF)
`
`
`
`StorageStorage
`
`Interface
`(USB/CF)
`Main Control
`(Messaging, Fusion, Buffering)
`
`TimeSync
`
`Flash Storage
`
`Wireless
`Transceiver
`
`NC
`(Tmote sky / custom CF card)
`
`ActiS
`(Tmote sky)
`
`ActiS Application Layer
`
`ActiS
`Protocol
`
`Flash
`Storage
`
`Signal
`Processing
`
`Sensor
`Interface
`
`TimeSync
`
`Messaging
`Buffering
`
`Wireless
`Transceiver
`
`Figure 4 Block Diagram of Software Components in a WBAN.
`
`
`
`
`
`
`
`
`
`
`
`Zepp Labs, Inc.
`ZEPP 1012
`Page 8
`
`

`
`C. Otto, A. Milenkovic, C. Sanders, and E. Jovanov 315
`
`
`TinyOS components
`TinyOS is a lightweight open source operating system for wireless embedded sensors. It is designed to
`use minimal resources and its configuration is defined at compile time by combining components from
`the TinyOS library and custom-developed components. Well-defined interfaces are used to connect
`and define the data flow between components. A TinyOS application is implemented as a set of
`component modules written in nesC [20]. The nesC language extends the C language with new
`support for task synchronization and task management. This approach results in a natural modular
`design, minimal use of resources, and short development cycles.
`TinyOS fully supports the Tmote sky platform and includes library components for the Chipcon
`CC2420 radio drivers and other on-chip peripherals.
` Radio configuration, MAC
`layer
`communications, and generic packet handling are also natively supported.
`Figure 5 depicts the ActiS software architecture using the TinyOS component model. The
`components in yellow are those reused from the TinyOS library. GenericComm provides generic
`packet handling and basic SendMsg, ReceiveMsg interfaces using TinyOS messages. A TinyOS
`message is a generic message structure with a reserved payload for application data [20].
`GenericComm would also interface to low-level platform specific TinyOS hardware drivers. The
`ActiS application layer consists of custom components shown in blue.
`
`
`
`
`ActiS Application LayerActiS Application Layer
`
`
`
`RawOutput ActisEvent ActisConfig ActisCalibrate ActisInventoryRawOutput ActisEvent ActisConfig ActisCalibrate ActisInventory
`
`
`
`StepProcessingStepProcessing
`
`
`
`ActisCommActisComm
`
`
`
`StepProcessorStepProcessor
`
`
`
`55
`
`
`
`55
`
`
`
`SendMsg ReceiveMsgSendMsg ReceiveMsg
`
`
`
`GenericCommGenericComm
`
`
`
`SensorAccSensorAcc
`
`
`
`AccAcc
`
`
`
`Figure 5 Simplified sensor node interface connection (ActiS).
`
`
`Communication Protocol and Time Synchronization
`Our communication protocol was designed to minimize resources and uphold the spirit of the ZigBee
`star network topology [24]. Figure 6 shows our communication super frame. All communications are
`between a sensor node and the network coordinator. Each communication super frame is divided into
`50ms timeslots used for message transmissions. Each sensor uses its corresponding timeslot to
`transmit sensor data, command acknowledgements, and event messages. The first timeslot, however,
`
`
`
`Zepp Labs, Inc.
`ZEPP 1012
`Page 9
`
`

`
`
`316 System Architecture of WBAN for Ubiquitous Health Monitoring
`belongs to the network coordinator and is used for transmitting configuration commands from the
`personal server. The network coordinator also transmits periodic beacon messages used to
`synchronize the start of super frames. This organization also serves as practical collision avoidance,
`making more efficient use of the available bandwidth when compared to using only CC2420 Collision
`Sense Multiple Access (CSMA) scheme.
`Time synchronization is crucial in providing a means for network communication protocol, as well
`as for event correlation. In the WBAN, sensor nodes are distributed about the user’s body and are only
`wirelessly connected; the sensors operate for extended periods, sampling and analyzing physiological
`data. It becomes necessary to correlate detected events between sensors within the sampling interval.
`A time stamp mechanism can be employed; however, without a global time reference the timestamp
`has no meaning outside of the scope of a sensor node. Synchronizing the session start times and
`utilizing a local time reference is not sufficient for the problem at hand. Even if two sensors could
`precisely agree on the start of a health monitoring session, a running local time would only work for
`short session durations. Each sensor in the WBAN has a local clock source with an associated skew.
`The skew is a measure of the difference in frequency between the local clock source and an ideal clock
`source. As a result of skew, any elapsed time based on a local clock, over time, will differ between
`any two sensors. This error is cumulative. Consider two 32 KHz crystals differing by a typical 50
`ppm (parts per million). Over the course of a few hours, the sensor clocks can differ by more than
`several hundred milliseconds. For an accurate healthcare monitoring system and correlating step and
`gait analysis between two sensors this is unacceptable, while the optimization of communication
`channel sharing would be impossible.
`
`
`Super Frame
`
`
`
`BeaconBeacon
`
`
`
`Slot#1Slot#1
`
`
`
`Slot#2Slot#2
`
`
`
`Slot#3Slot#3
`
`
`
`……
`
`
`
`BeaconBeacon
`
`
`
`……
`
`Listen
`
`Tr.
`
`Listen
`
`Time
`
`
`
`1000 ms
`150 ms
`100 ms
`Figure 6 Communication Super Frame and activity of Sensor #2
`We address the issue by employing a modified version of the Flooding Time Synchronization
`Protocol (FTSP) developed at Vanderbilt University [11]. FTSP generates time synchronization by
`dynamically electing a master node. The master node transmits periodic beacons containing global
`time stamps. FTSP features MAC layer time stamping for increased precision and skew compensation
`with linear regression to account for clock drift. We modified FTSP for use in our prototype WBAN
`[4]. Our modified version exploits the WBAN’s star network topology. Beacon messages, used to
`delineate the super frame, also serve to distribute the global timestamps. Sensor nodes in the WBAN
`use beacons as a timing reference. For testing of the time synchronization protocol, we developed a
`test bed where the network coordinator and WBAN sensor nodes are all connected to a common wired
`
`50 ms
`
`
`
`
`
`Zepp Labs, Inc.
`ZEPP 1012
`Page 10
`
`

`
`C. Otto, A. Milenkovic, C. Sanders, and E. Jovanov 317
`
`
`signal. Sensors measured the signal changes in jiffy ticks, where one jiffy is 30.5 µs, determined by
`the clock frequency of the on-board crystal (32.768 KHz). In most cases the node’s error was within
`±1 jiffy and the average error was approximately 0.1*Tjiffy or 3 µs.
`Power Management
`Long-life, persistent sensor nodes require efficient power management. Ease of use and the perceived
`unobtrusiveness is affected by sensor weight, interval between battery changes, and the level of user
`interaction required. Because battery life is proportional to battery size (weight) it is our challenge as
`designers to minimize sensor power consumption and thus maximize battery life for a selected battery
`size. In designing our prototype we have held low power consumption as a major design goal – both
`in processor and technology selection as well as software organization. We have selected the MSP430
`microprocessor family for their excellent MIPS/mW ratio and 802.15.4 in part because of its unique fit
`for low power, low data rate applications. Beyond this, it is possible to extend each node’s lifetime by
`clever network organization and making
`