milenkovic_compcomm06

8/9/2019 milenkovic_compcomm06

1/13

Wireless sensor networks for personal health monitoring: Issues andan implementation

Aleksandar Milenkovic *, Chris Otto, Emil Jovanov

Electrical and Computer Engineering Department, The University of Alabama in Huntsville, 301 Sparkman Drive, Huntsville, AL 35899, USA

Abstract

Recent technological advances in sensors, low-power integrated circuits, and wireless communications have enabled the design of low-cost, miniature, lightweight, and intelligent physiological sensor nodes. These nodes, capable of sensing, processing, and communicatingone or more vital signs, can be seamlessly integrated into wireless personal or body networks (WPANs or WBANs) for health monitor-ing. These networks promise to revolutionize health care by allowing inexpensive, non-invasive, continuous, ambulatory health moni-toring with almost real-time updates of medical records via the Internet. Though a number of ongoing research efforts are focusingon various technical, economic, and social issues, many technical hurdles still need to be resolved in order to have flexible, reliable,secure, and power-efficient WBANs suitable for medical applications. This paper discusses implementation issues and describes theauthors prototype sensor network for health monitoring that utilizes off-the-shelf 802.15.4 compliant network nodes and custom-builtmotion and heart activity sensors. The paper presents system architecture and hardware and software organization, as well as theauthors solutions for time synchronization, power management, and on-chip signal processing. 2006 Elsevier B.V. All rights reserved.

Keywords: Wireless sensor networks; Health monitoring; Hardware; Software; Signal processing; Time synchronization; Energy efficiency

1. Introduction

Current health care systems structured and optimizedfor reacting to crisis and managing illness are facing newchallenges: a rapidly growing population of elderly and ris-ing health care spending. According to the U.S. Bureau ofthe Census, the number of adults age 6584 is expected todouble from 35 million to nearly 70 million by 2025 whenthe youngest Baby Boomers retire. This trend is global, so

the worldwide population over age 65 is expected to morethan double from 357 million in 1990 to 761 million in2025. Also, overall health care expenditures in the UnitedStates reached $1.8 trillion in 2004 with almost 45 millionAmericans uninsured. In addition, a recent study foundthat almost one third of U.S. adults, most of whom heldfull-time jobs, were serving as informal caregivers mostly

to an elderly parent. It is projected that health care expen-ditures will reach almost 20% of the Gross Domestic Prod-uct (GDP) in less then 10 years, threatening the wellbeingof the entire economy [1]. All these statistics suggest thathealth care needs a major shift toward more scalable andmore affordable solutions. Restructuring health care sys-tems toward proactive managing of wellness rather than ill-ness, and focusing on prevention and early detection ofdisease emerge as the answers to these problems. Wearable

systems for continuous health monitoring are a key tech-nology in helping the transition to more proactive andaffordable healthcare.

Wearable health monitoring systems allow an individualto closely monitor changes in her or his vital signs and pro-vide feedback to help maintain an optimal health status. Ifintegrated into a telemedical system, these systems can evenalert medical personnel when life-threatening changesoccur. In addition, patients can benefit from continuouslong-term monitoring as a part of a diagnostic procedure,can achieve optimal maintenance of a chronic condition,

0140-3664/$ - see front matter 2006 Elsevier B.V. All rights reserved.

doi:10.1016/j.comcom.2006.02.011

* Corresponding author.E-mail addresses:[email protected](A. Milenkovic),chrisaotto@

yahoo.com(C. Otto),[email protected](E. Jovanov).

www.elsevier.com/locate/comcom

Computer Communications xxx (2006) xxxxxx

ARTICLE IN PRESS
mailto:[email protected]:chrisaotto@%20yahoo.commailto:chrisaotto@%20yahoo.commailto:[email protected]:[email protected]:chrisaotto@%20yahoo.commailto:chrisaotto@%20yahoo.commailto:[email protected]


2/13


3/13

and mode of operation), customization (e.g., run user-spe-cific calibration or user-specific signal processing procedure

upload), and setup of a secure communication (keyexchange). Once the WWBAN network is configured, thePS application manages the network, taking care of chan-nel sharing, time synchronization, data retrieval and pro-cessing, and fusion of the data. Based on synergy ofinformation from multiple medical sensors the PS applica-tion should determine the users state and his or her healthstatus and provide feedback through a user-friendly andintuitive graphical or audio user interface. Finally, if acommunication channel to the medical server is available,the PS establishes a secure link to the medical server andsends reports that can be integrated into the users medical

record. However, if a link between the PS and the medicalserver is not available, the PS should be able to store thedata locally and initiate data uploads when a link becomesavailable.

Tier 3 includes a medical server(s) accessed via the Inter-net. In addition to the medical server, the last tier mayencompass other servers, such as informal caregivers, com-mercial health care providers, and even emergency servers.The medical server typically runs a service that sets up acommunication channel to the users PS, collects thereports from the user, and integrates the data into theusers medical record. The service can issue recommenda-tions, and even issue alerts if reports seem to indicate anabnormal condition. More details about this architectureand services can be found in [14].

2.1. Deployment scenarios

Fig. 2illustrates three typical scenarios using WWBAN.The configuration on the left can be deployed at home, inthe workplace, or in hospitals. Wireless medical sensorsattached to the user send data to a PDA, forming ashort-range wireless network (e.g., IEEE 802.15.1 or802.15.3/4). The PDA equipped with a WLAN interface(e.g., IEEE 802.11a/b/g) transmits the data to the home

(central) server. The home server, already connected to

the Internet, can establish a secure channel to the medicalserver and send periodic updates for the users medical

record.The modified configuration in the middle is optimized

for home health care. The sensor network coordinator(nc inFig. 2) is attached to the home personal server thatruns the PS application. The medical sensor nodes andthe network coordinator form a wireless personal area net-work. By excluding the PDA, we can reduce system cost.However, this setting is likely to require more energy spentfor communication due to an increased RF output powerand lower Quality of Service (QoS), requiring frequentretransmissions.

The configuration on the right illustrates ambulatory

monitoring applicable any time and everywhere the PSapplication runs on a Wireless Wide Area Network(WWAN) enabled PDA/cell phone (e.g., 2G, 2.5G, and3G) that connects directly to the medical server. To illus-trate, we will follow a user (lets call him Sam) who has apredisposition for heart disorders. Sam continues his nor-mal daily activities, but now equipped with several non-in-vasive medical sensors applied in his clothes or as tinypatches on his skin. The PS application can discriminateSams state (walking, running, lying down, sitting, riding,etc.) using the data from the motion sensors, and can rec-ognize an arrhythmic event, analyzing the data from theECG sensor. If Sams activity, heart rate, and personalizedthresholds indicate an abnormal condition, he will receive awarning. In addition, a precise incident report can be sentto the medical server at Sams hospital or doctors office.

2.2. Requirements for wireless medical sensors

Wireless medical sensors should satisfy the mainrequirements such as wearability, reliability, security, andinteroperability.

2.2.1. Wearability

To achieve non-invasive and unobtrusive continuous

health monitoring, wireless medical sensors should be

nc

nc

InternetInternet

Tier 1:

WWBAN

WLAN

or

WWAN

A A

AE

Tier 2:

PS

Tier 3:

MS

Fig. 1. WWBAN integrated into a telemedical system for health monitoring.

A. Milenkovicet al. / Computer Communications xxx (2006) xxxxxx 3

ARTICLE IN PRESS


4/13


5/13

3.1. Hardware platform

The ActiS sensor node features a hierarchical organi-zation employed to offer a rich set of functions, benefitfrom the open software system support, and performcomputation and communications tasks with minimalpower consumption. Each ActiS node utilizes a commer-cially available wireless sensor platform Telos fromMoteiv [18] and a custom intelligent signal processingdaughter card attached to the Telos platform (Fig. 4).The daughter boards interface directly with physical sen-

sors and perform data sampling and in some cases preli-minary signal processing. The pre-processed data is thentransferred to the Telos board. The Telos platform cansupport more sophisticated real-time analysis and canperform additional filtering, characterization, featureextraction, or pattern recognition. The Telos platformis also responsible for time synchronization, communica-tion with the network coordinator, and secure datatransmission.

Fig. 5shows a block diagram of an ActiS with a Telosplatform and a customIntelligent Signal Processing Module(ISPM). Telos is powered by two AA batteries and featuresan ultra-low power Texas Instruments MSP430 microcon-troller [20]; a Chipcon CC2420 radio interface in the2.4 GHz band [21]; an integrated onboard antenna with50 m range indoors/125 m range outdoors; a USB portfor programming and communication; an external flashmemory; and integrated humidity, temperature, and lightsensors. The MSP430 microcontroller is based around a16-bit RISC core integrated with RAM and flash memo-

ries, analog and digital peripherals, and a flexible clocksubsystem. It supports several low-power operating modesand consumes as low as 1 lA in a standby mode; it also hasvery fast wake up time of no more than 6 ls. Telos Revi-sion A features a MS430F149 microcontroller with 2 KBRAM and 60 KB flash memory; Telos Revision B (nowTmoteSky) features a MSP430F1611 with 10 KB ofRAM and 48 KB of flash memory. The CC2240 wirelesstransceiver is IEEE 802.15.4 compliant and has program-mable output power, maximum data rate of 250 Kbs, andhardware support for error correction and encryption.The CC2240 is controlled by the MSP430 microcontrollerthrough the Serial Peripheral Interface (SPI) port and a ser-ies of digital I/O lines with interrupt capabilities. The Telosplatform features a 10-pin expansion connector with oneUART (Universal Asynchronous Receiver Transmitter)and one I2C interface, two general-purpose I/O lines, andthree analog input lines.

We developed two custom boards specifically for healthmonitoring applications, an ISPM and an IAS (IntelligentActivity Sensor). The ISPM board extends the capabilitiesof Telos by adding two perpendicular dual axis accelerom-eters (Analog Devices ADXL202), a bioamplifier with sig-nal conditioning circuit, and a microcontrollerMSP430F1232 (Fig. 5). The ISPMs two ADXL202 accel-

erometers cover all three axes of motion. One ADXL202 is

105 105.5 106 106.5 107 107.5 108 108.5 1090

500

1000

1500

2000

Step

105 105.5 106 106.5 107 107.5 108 108.5 1090.5

1

1.5x 10

acc

105 105.5 106 106.5 107 107.5 108 108.5 1091000

2000

3000

4000

Time [sec]

ECG

accX

accYaccZ

Fig. 3. ECG, acceleration, and foot switch data collected by the WWBAN prototype during normal walking.

Fig. 4. Photo of two ActiS sensors with customized daughter boards.


ARTICLE IN PRESS


6/13

mounted directly on the ISPM board and collects data forthe Xand Yaxes. The second ADXL202 is mounted on acard that extends vertically from the ISPM and collectsacceleration data on the Z axis. The users physiological

state can be monitored using an on-board bioamplifierimplemented with an instrumentation amplifier and signalconditioning circuit. The bioamplifier could be used forelectromyogram or electrocardiogram monitoring. Theoutput of the signal conditioning circuit is connected tothe local microcontroller as well as to the microcontrolleron the Telos board via the expansion connector. The ISPMhas its own MSP430F1232 processor for sampling and low-level data processing, selected primarily for its compact sizeand excellent MIPS/mW ratio. Other features that weredesirable for this design were the 10-bit ADC and the timercapture/compare registers that are used for acquisition of

data from accelerometers. The MSP430F1232 also has ahardware UART that is used for communications withthe Telos board. The IAS board is a stripped-down versionof the ISPM with only accelerometer sensors and signalconditioning for a force-sensing resistor that can be usedas a foot switch.

3.2. Software organization

The system software is implemented in a TinyOS envi-ronment [22]. TinyOS is a lightweight open source oper-ating system for wireless embedded sensors. It is designedto use minimal resources, and its configuration is definedat compile time by combining components from theTinyOS library and custom-developed components. ATinyOS application is implemented as a set of compo-nent modules written in nesC [23]. The nesC languageextends the C language with new constructs to facilitatethe component architecture and multitasking. By addingdirect language support for task synchronization andtask management, it allows rapid development and min-imizes resource usage. Fig. 6 shows a generalizedWWBAN software architecture, and from top to bottom,it shows the network coordinator software, WWBANnodes Telos software, and WWBAN nodes daughter

card software.

3.2.1. Network coordinator

The network coordinator is also implemented on a Telosplatform. It feeds the PS application through its USB con-nector and manages the WWBAN transmits the messages

from the PS that establish a session, assigns the individualsensor ID, distributes keys if secure data are encrypted, andassigns communication slots. The network coordinatorautonomously emits beacon messages for time synchroni-zation. After the initial setup, it receives data from individ-ual sensors, aggregates the data, and forwards it to the PSapplication.

3.2.2. Telos software

The Telos application software is implemented as multi-ple TinyOS components encompassing the following high-level functions: wireless communication, extended flash

storage, messaging software, board-to-board communica-tions, and signal feature extraction. Telos serves as a mas-ter controller, and it requests data from the daughtersensor card every 40 ms (25 Hz) by raising an interruptrequest line. The daughter sensor card sends preprocesseddata via an asynchronous serial interface. The receiveddata can also be processed and analyzed. For example,motion sensors can analyze acceleration signals to identifythe moment when a step has been made. A step detectionevent and the corresponding time stamp are sent to the per-sonal server. As an alternative, we can upload raw datafrom accelerometers at the price of increased power con-sumption. The processed data set can be stored in an exter-nal serial flash memory in the case of autonomousoperation or if the wireless channel is not available.

It should be noted that the flash memory, CC2420 radiointerface, and the daughter sensor card all share a singleserial interface of the MSP30 on the Telos platform. Thispresented its own set of challenges since the Telos platformis tasked with reliable communications to multiple devicesusing this single serial interface. For example, to communi-cate with the daughter card, the software must configurethe serial interface as a UART running at 115.2 kbps. Oncesensor data is received, the serial interface is dynamicallyreconfigured for SPI at 500 kbps, allowing communications

to both the on-board radio and flash. Because events are

ADXL202Accelerometer

ADXL202Accelerometer

ECG SignalConditioning

TIMSP430F1232

CC2420(ZigBee)

Flash

USB Interface

TI

MSP430F149

TelosISPM

ECG electrodes

Fig. 5. Block diagram of an ActiS sensor node.

6 A. Milenkovicet al. / Computer Communications xxx (2006) xxxxxx

ARTICLE IN PRESS


7/13

recognized asynchronously, accurate event time stamps canbe made, but often the messages must be buffered andqueued for transmission when the serial interface isavailable.

3.2.3. Sensor software

The sensor boards handle acquisition of physiologicalsignals and pre-processing. For example, the ISPM samplesthree independent accelerometer axes each at a rate of200 Hz. The raw accelerometer data is filtered and prepro-cessed. The filtering includes moving an average filter to

eliminate high frequency movement artifacts, and separa-tion of low and high frequency components of the acceler-ation signal. Sensor orientation can be calculated as theangle between low frequency accelerometer components.User activity is estimated with a function based on thesum of the integrals of the AC components in each channel[17].

4. Time synchronization

Time synchronization is a common requirement forwireless sensor networks since it allows collective signal

processing, sensor and source localization, data aggrega-

tion, and distributed sampling. In wireless body area net-works, synchronized time stamps are critical for propercorrelation of data coming from different sensors and foran efficient sharing of the communication channel. Forexample, our prototype needs to synchronize and time-stamp data from motion sensors and the heart sensor,and establish a protocol for sharing the communicationchannel. Precise time stamping is also important in the caseof intermittent communication that can significantly post-pone transmission of event messages.

A synchronization mechanism for a given application

is determined by the following: (i) the degree of precisionneeded, (ii) the longevity of synchronization, that is,whether we need to stay synchronized all the time or justwhen needed, (iii) the resources available (clocks), and (iv)the power and time budget available for achieving timesynchronization. A number of protocols and algorithmshave been proposed and implemented to provide timesynchronization in computer networks. However, theyare often ill-suited for wireless sensor networks since theyrequire significant computing resources and do not offerfault-tolerant solutions. Several protocols have been spe-cifically developed for wireless sensor networks [24

26,28,29].

Network Coordinator(Telos)

Telos

Sensor board

DataAcquisition

Filtering/Pre-processing

Transmitter

(UART)

Receiver(UART)

FeatureExtraction

WirelessReceiver/

Transmitter

Main Control(Messaging, Timer, Buffering)

Flash Storage

WirelessReceiver/

Transmitter

TimeSyncReceiver/

Transmitter (USB)

Main Control(Messaging, Fusion, Buffering)

Flash Storage

PersonalServer

Fig. 6. WBAN software architecture.


ARTICLE IN PRESS


8/13

One of the key protocols for time synchronization inWSNs is the Flooding Time Synchronization Protocol(FTSP) developed at Vanderbilt University[29]. It featuresMAC layer time stamping for increased precision and skewcompensation with linear regression to account for clockdrift. The FTSP generates time synchronization with peri-

odic time sync messages. The network can dynamicallyelect a root node. Whenever a node receives a time syncmessage, it rebroadcasts the message, thus flooding the net-work with time sync messages. The message itself containsa very precise timestamp of when the message was sent.The receiving node takes an additional local timestampwhen it receives the message. Because the timestamps aretaken deep in the radio stack, they eliminate non-determin-istic error sources and only include highly deterministicevents such as air propagation time, radio transmission,and radio reception time. Comparing the timestamps fromthe last several messages received, the node computes asimple linear regression to allow it to account for the offset

difference in its clock from global time as well as the rela-tive difference in frequency of local oscillators. This enableseach node to maintain an accurate estimate of global timedespite the fact that its local clock may be running slightlyfaster or slower than the global clock source.

We modified the original FTSP for WWBAN settings[30]. Our modification exploits the ZigBee[31]star networktopology to further minimize resources needed for timesynchronization. The prototype features a masterslavehierarchy where the network coordinator periodicallytransmits a beacon message to its slave nodes to maintain

the synchronized communication link; a slave node receivesthe beacon without re-transmission. This highly accuratetime synchronization allows a slave node to disable itsradio and enter a low power sleep mode, waking up justbefore the next message is due. In addition, we allowedan original implementation where a root can be dynamical-

ly chosen and the network flooded by the sync messages inthe case that the network coordinator fails or is turned off.For time synchronization to work, there must be a fixed

point in time from which both sender and receiver can ref-erence the timestamp in a given message. For a ZigBeemessage, this point is at the end of the Start of FrameDelimiter (SFD). Fig. 7 shows the interface between theChipcons CC2420 radio transceiver and the microcontrol-ler at the top, and the IEEE 802.15.4 physical frame formatand corresponding pin activity at the bottom. On the Telosplatform, the wireless transceiver provides a signal to themicrocontroller to indicate when the SFD byte has beenreceived or transmitted. As soon as the SFD byte is trans-

mitted, the radio transceiver raises the SFD pin that initi-ates a timer capture and an interrupt request on themicrocontroller. This allows the microcontroller to get atimestamp immediately after the SFD byte is transmitted.This timestamp is then inserted into the current time syn-chronization message (Fig. 8). It should be noted that themessage has already begun transmission when the time-stamp is made and added to the message. In a similarway, the receiver makes a local timestamp when it receivesthe SFD and stores it with the time synchronization mes-sage (Fig. 8). Later, the microcontroller will compare the

FIFOFIFOP

CCA

SFD

CSn

SISO

SCLK

GIO1

Interrupt

GIO0

SPI

GIO2MOSI

MOSO

SCLK

Data transmitted

over RFMAC Protocol DataLengthSFDPreamble

SFD Pin

Automaticallygenerated

preamble and SFD

Data fetched from TxFIFO CRC

Data received


SFD Pin

FIFO

Timer Capture

SPI

Fig. 7. Interfacing CC2420 RF transceiver (top). IEEE 802.15.4 frame format and pin activity (bottom).


ARTICLE IN PRESS


9/13

two timestamps to determine the offset between the localtime and the global time, and adjust the degree to whichthe local clock is running faster or slower than the globalclock (the network coordinator global time).

This is a platform-specific implementation of the origi-nal FTSP mechanism, and it results in very accurate time-stamps deep in the radio stack [30]. Unlike the originalimplementation of the FTSP on Mica2 boards where theprocessor directly controls the wireless bit-stream, the Tel-

oss wireless controller offloads channel access from themain processor. In addition, the CC2420 provides flexiblereceive and transmit FIFO buffers and communicates withthe main processor using a synchronous peripheral inter-face (SPI). Generally, when a message is being transmitted,the processor loads up the transmit FIFO with the entiremessage and then enables transmission. However, theFTSP messages contain a timestamp that is generated afterthe message has begun transmission. To implement this onthe Telos platform, most of the message is placed in theFIFO and transmission is enabled. When the SFD inter-rupt occurs, the captured timer value is retrieved and con-verted to a global timestamp. The timestamp is insertedinto the message and the rest of the message is placed inthe FIFO. Assuming this can all be done quickly enough,the entire message is transmitted properly. If however theprocess is too slow, the FIFO will under-run and the mes-sage transmission will fail. Therefore, the speed must bechecked for each individual application since ZigBee spec-ifies a fairly speedy effective bit rate of 250 kbps.

The SPI interface between the microcontroller and thewireless controller on the Telos platform runs at 500 kbps,twice as fast as the message is transmitted over the radio.Experimental time measurements were taken to ensure ade-quate margin and instill confidence that our protocol can

run reliably without FIFO under-run. It takes about

700 ls to transmit a time sync message and about 150 lsto calculate and insert the timestamp. It takes another300 ls to send the second part of the message over theSPI and finish filling the FIFO. All of this means that theentire message makes it into the FIFO with about 250 lsto spare, which is adequate for our application.

Software support for time synchronization is imple-mented as a nesC interface that provides applicationsaccess to the time synchronization information. It can give

the current global time, convert a local timestamp to globaltime, or calculate wait period until a future global time.This last facility is useful for an application that wants todo something at a specific global time. For example, weuse it to put the radio transceiver in the sleep mode untilthe next communication window. If the next window willoccur at a particular global time, the node can find outhow long it will have to wait and then set a timer accord-ingly so it can wake up and enable the radio.

All timestamps and clocks were based on the 32768 Hzcrystal on the Telos board. The crystal exhibits goodshort-term stability, an essential quality for our WWBANprototype to work properly. Though the low frequencycrystal does not allow a very high-resolution synchroniza-tion, the period of 30.5 ls is quite satisfactory for typicalWWBAN applications. It should be noted that theMSP340 microcontroller has a sophisticated clock modulewith a digitally controlled oscillator (DCO) with clock fre-quencies up to 5 MHz. Though a DCOs clock short-termstability is rather poor, an online modulation and clock sta-bilization can be employed if higher resolution is needed.

For testing of time synchronization protocol, we devel-oped a test bed where the network coordinator andWWBAN slave sensors are all connected to a commonwired signal connected to an MSP430s digital I/O port

with timer capture capability. All nodes are configured to

Propagation

Data transmitted


SFD Capture Timer

Timestamp

MAC Protocol DataLengthSFDPreamble TimestampData received

over RF

SFD Capture Timer

Synchronize local time(TinyOS)

Network

Coordinator

Process Send

NC

BAN

EE

AAAA

AA

Fig. 8. Time synchronization in WBAN prototype.


ARTICLE IN PRESS


10/13

report their respective global timestamps every time thecommon signal changes its state. For each event, the mas-ters timestamp was compared to the slaves timestamp todetermine the absolute error of the slaves timekeeping.

Testing the prototype indicated very good time synchro-nization. A number of experiments have been run varying

the frequency of synchronization messages and test dura-tion. In most cases the slave nodes error was within 1tick (30.5ls), and an average error was approximately0.1* Tcycle or 3 ls. It is expected that the more frequentlytime sync messages are sent, the better the network nodeswill be able to maintain synchronization. Somewhat sur-prisingly, we discovered that increasing the time betweenmessages improves time synchronization, because nodesare better able to estimate clock skew. However, the costof larger inter-beacon periods is an increase in algorithmconversion time. Consequently, we implemented a hybridscheme where messages are sent more often. During con-vergence, nodes will process every beacon. After a coarse

convergence is achieved, nodes will begin to process everyN-th beacon, allowing for a more precise skew estimation.

5. Energy efficiency

Energy consumption is a first class design constraint inwireless sensor networks since they are battery operated.To extend each nodes lifetime, it is necessary to reducepower dissipation as much as possible; dissipation below100 lW will enable operation on energy scavenged fromthe environment. Various design trade-offs between com-munication and on-sensor computation, collaborative pro-

tocols, and hierarchical network organization can yieldsignificant energy savings. Once the sensor network isdeployed, dynamic power management techniques can beemployed in order to maximize battery life.

In WWBAN systems, reducing total power consump-tion is crucial for several reasons. Size and weight of sen-sors are predominantly determined by the size and weightof the batteries. On the other hand, a batterys capacityis directly proportional to its size. Consequently, WWBANsensor nodes need to be extremely energy efficient, sincereducing energy requirements will allow designers to use

smaller batteries. Smaller batteries will result in furtherminiaturization of physiological sensors and, in turn, anincreased level of users comfort. Second, an extended peri-od of operation without battery changes is desirable,because frequent battery changes on multiple sensors arelikely to hamper users acceptance. In addition, longer bat-

tery life will decrease WWBAN operational costs.We have designed a custom, application-specific proto-col according to 802.15.4 recommendations [27]. In orderto satisfy medical application requirements, the networkprotocol specifies a 1-s super frame cycle (TSFC= 1 s) andeach slave node has its reserved time slot of 50 ms to trans-mit the data (Fig. 9). A super frame cycle starts with a bea-con message sent by the network coordinator; the beaconmessage carries time synchronization information. Eachsensor node wakes its radio interface up in a receive modeimmediately before the next expected beacon.

Fig. 10shows the power profiles recorded for a motionsensor using an environment for real-time power monitor-

ing[32]. We can clearly identify three distinct states: Listen,Transmit, and Inactive modes. As described in Section 3,the ultra-low power microcontroller on the daughter boardsamples 3-axes of acceleration with frequency of 200 Hz.The data is filtered and buffered. The processor on theTelos board wakes up every 40 ms (25 Hz) and raises aninterrupt requiring the data from the daughter card. Inthe Inactive mode the wireless transceiver on the Telosplatform is turned off, and the average current drawn forthe whole sensor is 1.53 mA. The motion sensor can beconfigured to send raw accelerometer data or detectedsteps. If raw accelerometer data is required, the amount

information to be sent per one super frame (1 s) is 3axes*40 Hz*2 bytes = 240 bytes, which is equivalent to12 TinyOS packets. If the step detection is performed onthe sensor only, information about that event and corre-sponding timestamp are sent (1 or 2 packets). The motionsensor draws on average 20.1 mA in the Transmit modeand 20.8 mA in the Listen mode. The main contributorto this figure is the wireless radio that draws 17.4 mA whenit is transmitting and 19.7 mA when it is receiving. Basedon these parameters, we can calculate the average currentas follows:

Fig. 9. Super frame cycle.


ARTICLE IN PRESS


11/13

Iavg TListen

TSFC

IListen TTransmit

TSFC

ITransmit

TSFC TListen TTransmit

TSFC

IInactive.

The average current can be used to estimate battery life.If only two messages are sent per super frame cycle (TLis-ten= 50 ms, TTransmit= 15 ms), the average current is2.77 mA. Two AA batteries on the Telos platform have2900 mAh capacity, so the expected operating time of themotion sensor is 1046 h or over 6 weeks. However, witha tiny 120 mA rechargeable battery, the operating time willbe slightly less than 2 days. Further optimizations are alsopossible: depending on the WWBAN deployment scenario,

we could decrease the output power during transmission, asuper cycle can be extended (the node would spend lesstime in the listen mode), or data can be stored locally ina compressed format and then later transmitted.

6. Conclusions

This paper demonstrates the use of WWBANs as a keyinfrastructure enabling unobtrusive, continual, ambulatoryhealth monitoring. This new technology has potential tooffer a wide range of benefits to patients, medical person-nel, and society through continuous monitoring in the

ambulatory setting, early detection of abnormal condi-

tions, supervised rehabilitation, and potential knowledgediscovery through data mining of all gathered information.

We have described a general WWBAN architecture,important implementation issues, and our prototypeWWBAN based on off-the-shelf wireless sensor platformsand custom-designed ECG and motion sensors. We haveaddressed several key technical issues such as sensor nodehardware architecture, software architecture, network timesynchronization, and energy conservation. Further effortsare necessary to improve QoS of wireless communication,reliability of sensor nodes, security, and standardizationof interfaces and interoperability. In addition, further stud-ies of different medical conditions in clinical and ambulato-ry settings are necessary to determine specific limitations

and possible new applications of this technology.

Acknowledgments

We thank John Gober for his help in the ISPM and IASprinted circuit board development and Dennis Cox for hishelp with time synchronization protocol. We also thankPiet de Groen from Mayo Clinic for inspiring exchangeof ideas about WWBAN.

References

[1] National Coalition on Health Care, in: , Accessed: June 2005.

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 50

5

10

15

20

25

Time [sec]

]Am[I

0.65 0.7 0.75 0.8 0.85 0.9

0

5

10

15

20

Time [sec]

]Am[I

Listen Transmit

Idle

Fig. 10. Actis current consumption during 5 s of operation. Red arrow marks one super frame cycle. (For interpretation of the references to color in thisfigure legend, the reader is referred to the web version of this paper.)


ARTICLE IN PRESS
http://www.nchc.org/facts/cost.shtmlhttp://www.nchc.org/facts/cost.shtmlhttp://www.nchc.org/facts/cost.shtmlhttp://www.nchc.org/facts/cost.shtml


12/13

[2] P. Binkley, Predicting the potential of wearable technology, IEEEEngineering in Medicine and Biology Magazine 22 (3) (2003) 2324.

[3] Seiko Pulse Graph Wrist Watch, Available at: (Japanese), Accessed: July 2005.

[4] Polar, Available at: , Accessed: July2005.

[5] Digi-Walker step counter, Available at: , Accessed: July 2005.

[6] Actigraph, Available at: ,Accessed: July 2005.

[7] Holter Systems, Med-electronics Inc., Available at: , Accessed: July 2005.

[8] Cardio Micro Sensor, Available at: ,Accessed: July 2005.

[9] T. Martin, E. Jovanov, D. Raskovic, Issues in wearable computingfor medical monitoring applications: a case study of a wearable ECGmonitoring device, in: International Symposium on Wearable Com-puters ISWC 2000, Atlanta, October 2000.

[10] R. Szewczyk, E. Osterweil, J. Polastre, M. Hamilton, A. Mainwaring,D. Estrin, Habitat monitoring with sensor networks, Commun. ACM47 (6) (2004) 3440.

[11] J. Burrell, T. Brooke, R. Beckwith, Vineyard computing: sensornetworks in agricultural production, IEEE Pervasive Computing 03(1) (2004) 3845.

[12] E. Jovanov, J. Price, D. Raskovic, K. Kavi, T. Martin, R. Adhami,Wireless personal area networks in telemedical environment, in:Proceedings of the Third International Conference on Informationtechnology in Biomedicine(ITAB-ITIS2000), Arlington, VA, 2000, pp.2227.

[13] E. Jovanov, A. Lords, D. Raskovic, P. Cox, R. Adhami, F. Andrasik,Stress monitoring using a distributed wireless intelligent sensorsystem, IEEE Engineering in Medicine and Biology Magazine 22(3) (2003) 4955.

[14] E. Jovanov, A. Milenkovic, C. Otto, P.C. de Groen, A wireless bodyarea network of intelligent motion sensors for computer assistedphysical rehabilitation, Journal of NeuroEngineering and Rehabili-tation, 2:6, March 1, 2005. Available at: .

[15] J. Welch, F. Guilak, S.D. Baker, A wireless ECG smart sensor forbroad application in life threatening event detection, in: Proceedingsof the 26th Annual International Conference of the IEEE Engineeringin Medicine and Biology Society, San Francisco, CA, September2004, pp. 34473449.

[16] D. Malan, T.R.F. Fulford-Jones, M.Welsh, S. Moulton, CodeBlue:an ad hoc sensor network infrastructure for emergency medical care,in: Proceedings of the MobiSys 2004 Workshop on Applications ofMobile Embedded Systems (WAMES 2004), Boston, MA, June,2004, pp. 1214.

[17] M.J. Mathie, B.G. Celler, A system for monitoring posture andphysical activity using accelerometers, in: Proceedings of the 23rdAnnual International Conference of the IEEE Engineering in

Medicine and Biology Society, 2001, pp. 36543657.[18] Moteiv: Wireless Sensor Networks, Available at: , Accessed: June 2005.

[19] C. Otto, J.P. Gober, R.W. McMurtrey, A. Milenkovic, Emil Jovanov,An implementation of hierarchical signal processing on a wirelesssensor in TinyOS environment, in: Proceedings of the 43rd ACMSoutheastern Conference, Kennesaw, GA, vol. 2, March 2005, pp.4953.

[20] MSP430 MCUs, Available at: ,Accessed: January 2005.

[21] Chipcon RF Tranceivers, Available at: ,Accessed: June 2005.

[22] TinyOS: an open-source operating system for sensor networks,Available at: , Accessed: June 2005.

[23] D. Gay, P. Levis, R. von Behren, M. Welsh, E. Brewer, D. Culler, ThenesC language: a holistic approach to networked embedded systems,in: Proceedings of Programming Language Design and Implementa-tion (PLDI) 2003, June 2003.

[24] J. Elson, D. Estrin, Time synchronization for wireless sensornetworks, in: Proceedings of the 2001 International Parallel andDistributed Processing Symposium (IPDPS 01), pp. 19651970.

[25] J. Elson, L. Girod, D. Estrin, Fine-grained network time synchroni-zation using reference broadcasts, in: Proceedings of the fifthsymposium OSDI 02, pp. 147163.

[26] S. Ganeriwal, R. Kumar, M.B. Srivastava, Timing-Sync Protocol forSensor Networks, in: Proceedings of the 1st International Conferenceon Embedded Network Sensor Systems (SenSys 03), pp. 138149.

[27] 802.15.4-2003 IEEE Standard for Information Technology-Part 15.4:Wireless Medium Access Control (MAC) and Physical Layer (PHY)specifications for Low Rate Wireless Personal Area Networks (LR-WPANS), 2003.

[28] M. Maroti, B. Kusy, G. Simon, A. Ledeczi, Robust multi-hop timesynchronization in sensor networks, in: Proceedings of the Interna-tional Conference on Wireless Networks (ICWN 04), vol. 1, pp. 454460.

[29] M. Maroti, B. Kusy, G. Simon, A. Ledeczi, The flooding timesynchronization protocol, in: Proceedings of the 2nd InternationalConference on Embedded Network Sensor Systems (SenSys 04), pp.3949.

[30] D. Cox, E. Jovanov, A. Milenkovic, Time synchronization for ZigBeenetworks, in: Proceedings of the 37th IEEE Southeastern Symposiumon System Theory (SSST05), Tuskegee, AL, March 2005, pp. 135138.

[31] ZigBee Alliance, Available at: , AccessedJune 2005.

[32] A. Milenkovic, M. Milenkovic, E. Jovanov, D. Hite, An environmentfor runtime power monitoring of wireless sensor network platforms,in: Proceedings of the 37th IEEE Southeastern Symposium on SystemTheory (SSST05), Tuskegee, AL, March 2005, pp. 406410.

Aleksandar Milenkovicis an assistant professor inthe Department of Electrical and ComputerEngineering at the University of Alabama inHuntsville. He currently directs the LaCASALaboratory (http://www.ece.uah.edu/~lacasa/).His research interests include advanced architec-tures for the next generation computing devices,low-power VLSI, reconfigurable computing,embedded systems, and wireless sensor networks.Dr. Milenkovic received his Dipl. Ing., MSc, andPhD degrees in Computer Engineering and Sci-

ence from the University of Belgrade, Serbia, in 1994, 1997, and 1999,respectively.

Chris Otto is an embedded software engineer atLewis Innovative Technologies, Inc. in Hunts-ville, Alabama. His interests include wirelesssensor networks, real-time systems, embeddedsystems and packet voice technologies. Mr. Ottoreceived his BS in Computer Engineering at theUniversity of Alabama in Huntsville in 1999 andis expected to receive his MS in 2006.


ARTICLE IN PRESS
http://www.seiko-pgt.or.jp/http://www.seiko-pgt.or.jp/http://www.polarusa.com/http://www.digiwalker.com/http://www.digiwalker.com/http://www.mtiactigraph.com/http://med-electronics.com/http://med-electronics.com/http://www.cardiomems.com/http://www.jneuroengrehab.com/content/2/1/6http://www.jneuroengrehab.com/content/2/1/6http://www.moteiv.com/http://www.moteiv.com/http://www.ti.com/msp430http://www.chipcon.com/http://www.tinyos.net/http://www.zigbee.org/http://www.ece.uah.edu/~lacasa/http://www.ece.uah.edu/~lacasa/http://www.zigbee.org/http://www.tinyos.net/http://www.chipcon.com/http://www.ti.com/msp430http://www.moteiv.com/http://www.moteiv.com/http://www.jneuroengrehab.com/content/2/1/6http://www.jneuroengrehab.com/content/2/1/6http://www.cardiomems.com/http://med-electronics.com/http://med-electronics.com/http://www.mtiactigraph.com/http://www.digiwalker.com/http://www.digiwalker.com/http://www.polarusa.com/http://www.seiko-pgt.or.jp/http://www.seiko-pgt.or.jp/


13/13

Dr. Emil Jovanovis an Associate Professor in theElectrical and Computer Engineering Depart-ment at the University of Alabama in Huntsville.He received the Dipl. Ing., MSc, and PhD degreesin electrical and computer engineering from theUniversity of Belgrade. Dr. Jovanov has beendeveloping wireless intelligent sensors for per-sonal health monitoring and mobile computing

for more than five years. He was principalinvestigator or co-investigator on several grantsfrom NSF and industry in the field of wireless

sensor networks, wireless intelligent sensors, and wearable health moni-tors. His research interests include ubiquitous and mobile computing,biomedical signal processing, and telemedical health monitoring. Dr.Jovanov serves as an Associate Editor of IEEE Transaction on Infor-mation Technology in Biomedicine and Applied Psychophysiology andBiofeedback.


ARTICLE IN PRESS

Documents

milenkovic_compcomm06