ORIGINAL PAPER

Energy-aware Gateway Selection for Increasing the Lifetimeof Wireless Body Area Sensor Networks

Cuneyt Bayilmis & Mohamed Younis

Received: 21 September 2010 /Accepted: 25 October 2010 /Published online: 5 November 2010# Springer Science+Business Media, LLC 2010

Abstract A Wireless Body Area Sensor Network(WBASN) is composed of a set of sensor nodes, placedon, near or within a human body. WBASNs opt tocontinuously monitor the health conditions of individualsunder medical risk, e.g., elders and chronically ill people,without keeping them in a hospital or restraining theirmotion. A WBASN needs to stay connected to local orwide area networks using wireless technologies in orderto send sensor readings to a medical center. The WBASNnodes are implanted within the human body and wouldthus have limited energy supply. Since the mission of theWBASN is very critical, increasing the lifetime of nodesis essential in order to maintain both practicality andeffectiveness. This paper presents a new GatewaySelection Algorithm (GSA) that factors in the use ofenergy harvesting technologies and dynamically picks themost suitable WBASN node that serves as a gateway toother wireless networks. The goal of GSA is to balancethe load among the nodes by adaptively changing thegateway node in WBASN depending on the energyreserve of nodes. Computer modeling and simulations ofthe proposed GSA are carried out using OPNET. Thesimulation results demonstrate the effectiveness of theproposed GSA approach.

Keywords Wireless Body Area Sensor Networks(WBASN) . Energy harvesting . Network lifetime . Selectiveengagement of nodes . Gateway selection algorithm

Introduction

Nowadays, Wireless Body Area Sensor Networks(WBASN) is receiving an increased attention from theresearch community, motivated by numerous biomedicalapplications. Basically, WBASN can be a major asset forthe healthcare industry. For example WBASN can enablecontinual and unattended monitoring of the health con-ditions of elders, people with chronicle diseases, andathletes during their gymnastic training. In these applica-tions WBASN not only can mitigate health risk andexpedite incident response, but also bring about hugeeconomic advantages by cutting the labor cost and makinghealthcare more affordable.

Generally, a WBASN consists of several wirelesssensor nodes placed on, or implanted in a human body.The sensors probe the human body and measure heartrate, blood pressure, body temperature, respiration rate,etc. The collected data are sent to a medical center viafor physicians and caregivers to assess the conditions ofthe monitored individual. Since the main benefit of aWBASN is to avoid restraining the lifestyle of a patientor an elder, it should be possible to interconnect theWBASN with other networks in order to ensure that datacan reach the medical center while the monitoredindividual in a different place other than home, e.g.shopping mall, office, etc., as well as on the move, e.g.driving a car or riding a bus. Therefore, the capability ofconnecting to local and wide area networks is arequirement for WBASN [1–5].

C. Bayilmis (*)Department of Electronics and Computer Education,Faculty of Technical Education, University of Sakarya,Sakarya, Turkeye-mail: cbayilmis@sakarya.edu.tr

M. YounisDepartment of Computer Science and Electrical Engineering,University of Maryland, Baltimore County,Baltimore, MD, USAe-mail: younis@cs.umbc.edu

J Med Syst (2012) 36:1593–1601DOI 10.1007/s10916-010-9620-y

The nodes in WBASN are usually miniaturized andoperate on small batteries. In most setups, the well-being ofan individual should be continuously monitored using theWBASN. The natural question that would then arise is howthe nodes will get energy to sustain operation for longdurations, especially with the small form factor of theemployed nodes. Although it is theoretically feasible toreplace the batteries, there are numerous issues that makethis option impractical or at least very inconvenient.Basically, tampering with very small devices that areattached or implanted in a human body would be a majorhealth risk and would need professional attention, specifictools and clean environment in order to avoid the potentialof contamination. Therefore, increasing the lifetime ofWBASNs would have to be achieved through efficientusage of node batteries and by harvesting energy toaugment the onboard energy supply.

This paper tackles the energy management challenge inWBASN. First, it is important to note that a node consumesthe most of its energy in transmitting its data over wirelesslinks. By collecting all data at a single “sink” node andmaking a single transmission to the medical center, it wouldbe feasible to cut on the energy consumption of most nodessince they will be communicating with another node in thenetwork rather than an external hop. Nonetheless, the sinknode will be depleting its onboard energy supply at a highrate and would eventually seize its operation rather quickly.In this paper, a novel algorithm is proposed for selecting thegateway of the WBASN to the medical center. Theproposed Gateway Selection Algorithm (GSA) tracks theremaining energy of the individual nodes and factors in therate of energy harvesting from the human body in order topick the sink node that acts as a gateway and interfaces theWBASN to the outside world. In other words, GSAdynamically changes the sink node in a WBASN dependingon the energy level of the deployed sensor nodes. In thepresented research work, we assume that the sensor nodesare equipped with suitable devices to harvest energy fromthe human body, e.g. heat and motion, during the usualdaily activities such as walking, working, joggling, etc.. Animportant note is that the energy obtained from the humanbody using these devices is irregular, random, anddiscontinuous. While a sensor node can generate energy,another one may not harvest energy in a WBASN during atime period. As a result, the presented GSA is an attractivesolution for energy efficiency in WBASN.

The increased network lifetime achieved by GSA willboost the effectiveness of WBASN as a healthcare tool. Asmentioned replacing the batteries of implanted or attachedsensing devices will necessitate the involvement of pro-fessionals and may even requires the use of the operationroom. GSA enables sustaining the network operation usingexisting and scavenged energy resources and would thus

WBASN a non-intrusive tool for patients and elder. Theremainder of the paper is organized as follows. “Prior work onincreasing the lifetime of WBASN” presents an overview ofprevious work on reducing energy consumption and increas-ing lifetimes of WBASN. The used system model in theproposed solution is introduced in “Wireless body areasensor network system architecture”. GSA is explained in“Energy-aware gateway selection”. The simulation resultsare provided in “Performance validation”. “Conclusion”concludes the paper.

Prior work on increasing the lifetime of WBASN

Energy aware design and management of WBASN hasreceived significant attention from the research community.For example, Reusens et al. [6] performed measurements inorder to characterize the signal propagation model in asample WBASN. The goal is to provide design guidelinesfor forming a robust energy efficient network topology.They used the developed model to compare the single-hopand multi-hop topology and show that multi-hop dissemi-nation of data is the way to go. Braem et al. [7] and Ehyaieet al. [8] proposed the deployment of relay nodes inWBASN in order to reduce energy consumption in sendingthe sensors data. However, augmenting the WBASN withnodes that are attached to the human body is not alwayssuitable because it increases number of nodes and theincreased inconvenience. Another proposed solution in [6,7] is named as cooperation where the nodes cooperate inforwarding data from one node towards the sink node.However, these approaches assume a predetermined sinknode and do not factor in the remaining energy of nodes insetting the most suitable topology. Unlike these approaches,GSA dynamically adjusts the network topology anddesignates the sink node so that the network staysoperational for the longest time. In addition, prior workdid not consider the potential of energy scavenging on thenetwork topology.

To increase the lifetime of WBASN, GSA assumes thatthe sensor nodes are equipped with energy scavengingtechnology. The required node power may be generatedfrom human body by benefiting from the body’s heat,motion from leg or arm, chest expansion while breathing,vibration from heart. Starner and Paradiso [9, 10] investi-gated human generated power for mobile electronicsdevices and characterized the amount of energy that canbe harvested through the various technologies. Amor et al.[11] presented the basis for generating electrical energyfrom human body using vibration. They provided typicalexamples from daily activities such as eating (handmovement) and walking (foot movement) and argued thatthe generated energy from body motion is enough to make

1594 J Med Syst (2012) 36:1593–1601

hearing aid devices work. They further extended theirinvestigation to energy harvesting from breathing [12].Recognizing the potential of body-generated thermalenergy, Stark introduced the Thermo Life® device thatconverts heat into electrical energy [13]. Thermo Life is avery small device and would thus suits WBASN setups. Itis also worth noting that energy harvesting from humanbody has already been implemented in several productssuch as calculators, radios, Bluetooth headset, watches, etc.[9, 10, 14]. For example, Seiko Kinetic and ETAAutoquartz produced self-powered watches mainly benefitfrom vibration. Also, some Seiko produced watches runwith thermoelectric.

Some work also tried to reduce energy consumption atthe link layer, specifically by devising new MAC protocolsthat exploit the unique features of WBASN. For example,Omeni et al. [15, 16] developed a WBASN protocol thatfactors in the physical layer parameters and the nature ofthe application traffic in scheduling active and sleepdurations. On the other hand, Otal et al. [17] uses theapplication level quality of service requirements to derivethe node activation schedule that cut on medium accesscollisions and allows nodes to save energy wastage in idlelistening mode. Meanwhile S. Ullah and K.S. Kwak [18]propose a traffic-adaptive MAC protocol that dynamically

changes the duty cycle of the WBASN nodes based ontraffic pattern. A survey of other physical and link layertechniques for optimizing the design and operation ofWBASN can be found in [19]. The proposed GSAapproach considers the energy consumption at the level ofnetwork interface to the outside world and can becomplemented by additional means for energy savings atother layers in the communication protocols stack.

D.C. Hoang et al. [20] present an approach for selectingthe gateway based on the residual energy of the WBASNnodes. Unlike GSA, they do not use any threshold energylevel for leader node selection. In addition, they do not useenergy harvesting techniques from human body in theirsolution and assume a fix energy supply.

Wireless body area sensor network system architecture

Figure 1 shows the general system architecture for a WBASN.The system consists of three tiers: (i) the sensor network, (ii) abase station or network coordinator, and (iii) medical centerwhere the current status patients are monitored.

The sensor network It consists of multiple interconnectedsensor/actuator nodes attached to or implanted in a human

Wireless body sensor node

BS : Base Station WBASN: Wireless Body Area

Sensor Network

internet

WBASN

ECG

EEGBlood

Pressure

Glucose Monitoring

Motion

Pulse Oximetry

Outside

Home

BS

Hospital/Medical Center

Remote Monitoring

WWANGPRS

Doctor’s PDA

Patient Record

Database

Office

Home Monitoring and Base Station

Emergency Service

Fig. 1 General system architecture of a wireless body area sensor networks

J Med Syst (2012) 36:1593–1601 1595

body. The sensor nodes track vital signs such as electro-cardiography (ECG), electroencephalography (EEG), bloodpressure, heat, and respiratory from body and send thesensed/processed data through a base station to the medicalcenter for monitoring by doctors/caregivers.

The base station It can be a cell phone, a Personal DigitalAssistant (PDA) or a computer. It bridges the WBASN tothe medical center and carries out several different dataprocessing tasks. For example, it collects, analyzes anddisplays received signals from the sensor nodes. The basestation uses an internet connection through WLAN at homeor GPRS/WWAN in the outside to communicate with themedical center.

The medical center In a medical center, the received dataabout a patient are recorded and analyzed by doctors orcaregivers. If abnormal conditions are observed, themedical center intervenes to provide emergency servicesor activate on-body actuators such as insulin injection.

The WBASN is expected to operate autonomously andwithout a need to recharge the sensors batteries. It is thusassumed that energy scavenging capabilities are employedin the WBASN and the network is required to manage itsenergy resources wisely to ensure operation for long time.In our developed solution, we assume that the deployednodes are equipped with suitable energy harvestingdevices. Thus, the sensor nodes generate and storevarious amount of energy from the human body in dif-ferent times during daily human activities. The energyobtained from the human body using the energy harvestingdevices is irregular, random, and discontinuous. There-fore, the wireless sensor nodes have to store the obtainedenergy and carefully manage the power consumptioncompared to contemporary computing and communicationdevices.

Energy-aware gateway selection

The fundamental design principle of the proposed GatewaySelection Algorithm (GSA) is to pick the WBASN interfacebased on the energy reserve of the individual nodes and theenergy rate at which the node harvests energy from theenvironment. This entails making adjustment in the routingtopology dynamically in order to forward the data to thenode that can serve as a sink for the WBASN for thelongest time. GSA considers the fact that nodes consumemore energy in wireless transmission than in signalprocessing and that the gateway node has to transmit athigh power relative to other nodes in order to reach theaccess point of the command center. Therefore, the GSAalgorithm picks a node that can handle this task for the

longest time by assigning the gateway duty to the node withthe most energy.

As mentioned earlier, energy scavenging is essential forthe nodes to operate for an extended period and withoutcausing inconvenience to the individual served by theWBASN. Practically, energy harvesting techniques dependon the technology and the environment. While the effect oftechnology is obvious, the role of the environment deservesfurther elaboration. Basically, energy scavenging involvestransforming energy to a form suitable to be stored on abattery. If energy is harvested from natural resources, sunlight or winds, the effect of the environment in terms thetime of the day and the weather is fundamental. On the

Start

Y

All nodes send own energy level information each other

Begin working the selected node as sink node

Is there received packet from other nodes?

Send the received packet to base station or

network coordinator

Check out energy level? (Energy > Threshold)

N

Y N

Select the Leader node as Sink node

(it has the highest energy level node)

All nodes arrange own transceiver energy consumption in terms of

new selected the sink node

Lea

der

Nod

e W

orki

ng P

roce

ss

Lea

der

Nod

e Se

lect

Pro

cess

Energy Harvesting Process (The nodes generate energy according to used model)

Energy Harvesting Process (The nodes generate energy according to used model)

Fig. 2 Simplified flowchart of the Gateway Selection Algorithm

1596 J Med Syst (2012) 36:1593–1601

other hand, energy can be harvested from the motion madeby the monitored individual and would thus depends theenvironment he/she is in and what human activities areinvolved. For example, one might be in home, office, gym,shopping mall, etc., and the body activities would differsignificantly, e.g., sleeping, walking, juggling, etc., andconsequently the amount of harvested energy.

By continuously picking the gateway of the most energy,GSA factors in both the energy reserve and the rechargingrate. For example, a WBASN may have an ECG sensornode, a vibration/acceleration sensor on foot, a bloodpressure sensor node. While walking the accelerationsensor harvests the most energy in WBASN. In this case,the acceleration sensor node is selected as the new sinknode by using GSA. When the person is asleep, the ECGsensor node has the most energy harvest in WBASN and itcan thus work as a sink for all traffic and forward the datareports to the medical center. Consequently, the GSA willprovide an attractive solution in order to increase networklifetimes in WBASN. GSA requires each node to share itsenergy and rate of energy scavenging with the other nodesin the network. Simplified flow chart of the GSA is shownin Fig. 2. The GSA contains a Leader Node Select Process(LNSP) and a Leader Node Working Process (LNWP). Inthe former, all wireless sensor nodes send own energy levelto each other and the node with the highest energy levelassumes the responsibility of the gateway node andbecoming the sink for all sensor traffic. LNSP works inconjunction with a Look-up-Table (LT). The LT containsdistance information of between all nodes in WBASN. TheLT is used to optimize the energy consumption non-sinknodes by setting the output power of their transceiversaccording to the distance from the new selected gateway.The gateway node collects data from the other nodes inWBASN and sends them to the network coordinator. The

selected node stays as a gateway until its energy levelreaches a predefined threshold value. In addition, all nodescontinue to harvest energy according to their onboardcapabilities.

Performance validation

Computer modeling and simulations of the GSA areimplemented using OPNET. The GSA dynamically selectsa sensor node as the sink node in WBASN depending onthe energy level of nodes. In order to model and simulatethe GSA, we first identified and determined the energyconsumption model, energy harvesting and generatingdevice model, sink node energy threshold value andmaximum battery level of the sensor nodes.

Energy model

Energy consumption model We only consider the energyconsumption of the wireless communication since it ismuch larger than the energy used for data sensing andprocessing [5–7, 21, 22]. We have employed the radioenergy model used in [5–7] as summarized in Eqs. 1 and 2.The log-normal shadowing path loss model is assumed.

Etx k; d; nð Þ ¼ ETXelec:k þ EampðnÞ:k:dn ð1Þ

ErxðkÞ ¼ ERXelec:k ð2Þwhere,

Etx The transmission energy [J]Erx The receiver energy [J]ETXelec The energy dissipated by radio to run the

circuitry for the transmitter [nJ/bit]ERXelec The energy dissipated by radio to run the

circuitry for the receiver [nJ/bit]Eamp(n) The energy for transmit amplifier [J/(bit.mn)]k The number of transmitted bits [bit]

The radios are assumed to have power control andconsume the minimal energy needed to reach the receiver[6–8]. Table 1 shows the specific parameters of the NordicnRF2401 and Chipcon CC2420 transceivers which are usedfrequently in sensor networks [6–8]. We based the

Table 1 Energy consumption parameters for the Nordic nRF2401 andChipcon CC2420 transceiver [6–8]

Parameter nRF2401 CC2420

ETXelec (nJ/bit) 16.7 96.9

ERXelec (nJ/bit) 36.1 172.8

Eamp(n=3.11) (J/bit) 1.97e-9 2.71e-7

Eamp(n=5.9) (J/bit) 7.99e-6 9.18e-4

The supply of energy harvesting Power (W) Duration Energy (J)

Movements of hands [11] 2,592.10−6 1 h 9,3.10−3

Movements of foots [11] 9,452.10−6 1 h 34.10−3

Body’s heat [13] 30.10−6 1 h 108.10−3

Using shoe [22] 60.10−3 1 h 216

Table 2 Approximate amountof generated energy accordingto different energy harvestingtechniques

J Med Syst (2012) 36:1593–1601 1597

simulation on the Nordic transceiver due to its lower powerconsumption, which suits WBASN

Types of energy harvesting Micro energy is generated fromhuman body utilizing body’s heat, motion from leg or arm,chest expansion while breathing, vibration from heart, shoeetc. There are several studies on energy harvesting fromhuman body in the literature [10–12, 22]. Table 2 shows theapproximate amount of energy generated from human bodyaccording to different energy harvesting techniques. In thesimulation, we assumed the nodes generate energy accord-ing to Table 2.

Energy level and energy threshold for nodes We assumethat the wireless sensor nodes have two AA batteries (2 AAbatteries is 8640 Joule). Threshold energy value for sink nodeis half of the maximum battery energy (Energy Level * 0.5).Thus, we assume that all nodes are working independentunder energy threshold.

Simulation results

GSA was simulated under two different traffic modes thatreflect some daily activities of a typical healthy individualand an elder, respectively. Each traffic mode is based ondistinct sensor’s sample rate, packet size, distance to sinknode, different energy harvesting techniques, etc. Theseparameters are used in the energy harvesting and energyconsumption models. In addition, we used homogenous andheterogeneous scenarios in each traffic mode. The former

scenario contains only a single type of body sensor. Thelatter scenario has different body sensor types as seen inTables 3 and 4. On the other hand, we have used networklifetime as the performance metric to evaluate the effec-tiveness of the GSA algorithm. The network lifetime isdefined as the time of first node to die in the network. Wehave obtained the network lifetime results under threedifferent situations: (i) All wireless sensor nodes have onlya standard amount of energy (2 AA batteries with 8640 J),(ii) In addition to standard amount of energy all nodes havea suitable energy harvesting device as seen in Tables 3 and4, (iii) All wireless sensor nodes use the proposed GSA.

First model (daily life of a normal person)

This model tries to mimic the daily routine of a healthyperson. Figure 3 shows activities during 24 h of a normalperson. Energy harvesting and gateway selection processeswere performed according to this model. We have assessedthe performance under homogeneous and heterogeneousnode scenarios. Table 3 shows the body sensors and thetraffic parameters for this model.

Figure 4 shows the obtained network lifetime results forthe homogeneous scenario for the standard energy level,using energy harvesting device, and while employing theGSA. In this scenario, we used five pulse oxygen bodysensors. We have assumed that the network lifetime in thestandard energy level situation is 1 for normalized compar-isons. As seen from the Fig. 4, employing the proposed GSAincreases the network lifetime of the WBASN by approxi-mately 1.4 times compared to others situations.

Type of bodysensor

Samplingrate

Sensor data bitsper sample

Wirelesspacketsending rate

Type of used energyharvesting device

Temperature 1 reading/min 10 1 packet/min Body’s Heat

Blood Glucose 1 reading/min 10 1 packet/min Movements of Hands

Pulse Oxygena 1 reading/sec 16 1 packet/sec Movements of Hands

Blood Pressure 100 reading/sec

12 10 packet/sec Movements of Hands

Electrocardiogram(ECG)

240 reading/sec

12 24 packet/sec Body’s Heat

Table 3 Network trafficparameters used in first model

The samples are takenfrom [23]. TinyOS packetformat is used. Packetsize is fixed at 30 bytesa Pulse Oxygen sensor is only usedin homogeneous scenario.

Type of bodysensor

Samplingrate

Sensor data bitsper sample

Wirelesspacketsending rate

Type of used energyharvesting device

Temperaturea 1 reading/min 10 1 packet/min Body’s Heat

Pulse Oxygena 1 reading/sec 16 1 packet/sec Movements of Hands

Motion 10 reading/sec 10 5 packet/sec Shoes Pressure

Blood Pressurea 100 reading/sec 12 10 packet/sec Movements of Hands

Electrocardiogram (ECG)a 240 reading/sec 12 24 packet/sec Body’s Heat

Table 4 Network traffic param-eters used in the second model

a The samples are taken from [23].TinyOS packet format is used.Packet size is fixed at 30 bytes

Pulse Oxygen sensor is onlyused in homogeneous scenario

1598 J Med Syst (2012) 36:1593–1601

The obtained network lifetime results for the heteroge-neous scenario as a function three different situations(standard energy level, using energy harvesting device,employing the GSA) are shown in Fig. 5. The heteroge-neous scenario consists of five different body sensor typesincluding temperature, blood glucose, pulse oxygen, bloodpressure and ECG. The attributes of the body sensors aregiven in Table 3. Similar to the previous homogeneousscenario, the network lifetime in the standard energy levelsituation is assumed to be 1 for normalized comparisons.The standard energy level and the energy harvesting caseshave approximately the same network lifetime. EmployingGSA boosts the WBASN lifetime of a by about three timescompared to the other situations

Second model (daily life of an elderly person)

WBASN is most effective in monitoring the health of theelders. For this reason, we have studied the performanceusing a model that captures the daily routine of an elderperson. Figure 6 shows an example of such a routine.Energy harvesting and the gateway selection process wereperformed according to this model. Table 4 shows theengaged body sensors and the traffic parameters belong forthis model. Again, we have assessed the performance underhomogeneous and heterogeneous scenarios.

Figure 7 shows the network lifetime results for thehomogeneous scenario with the standard energy level,using energy harvesting device, and while employing the

Time(Hours)

Duration(Hours)

Type of Energy Harvesting

Distance to NC(m)

08 Body's Heat 2

1.5Movements of Hands

Body's Heat0.5 - 5

10 0.5Movements of Hands

Body's Heat1 - 2

0.2Movements of Hands

Body's Heat0.2 - 1

11 0.8 Body's Heat 0.2 - 1

0.5Movements of Hands

Body's Heat0.2 - 1

0.5Movements of Hands

Body's Heat1 - 2

19 7Movements of Hands

Body's Heat0.5 - 5

0.2Movements of Hands

Body's Heat0.2 - 1

20 0.8 Body's Heat 0.2 - 1

2Movements of Hands

Body's Heat0.5 - 5

0.5Movements of Hands

Body's Heat1 - 2

24 1.5 Body's Heat 1 - 2

Insi

de (

Hom

e)O

utsi

deIn

side

(Offi

ce)

Out

side

Insi

de(H

ome)

Action

Sleeping

Casual Activity

Eating

Walking

Driving / Bus

Walking

Eating

Office Activity

Walking

Driving / Bus

Casual Activity

Eating

Hobby Activity (Reading/Watching)

Fig. 3 The used model for anormal person’s daily routine

0

1

2

3

4

No Energy HarvestingDevice

Independent EnergyHarvesting Device

Gateway SelectionAlgorithm

Network Lifetime

Fig. 5 Network lifetime for a heterogeneous scenario based on dailyroutine of a health individual

0

0,5

1

1,5

No Energy HarvestingDevice

Independent EnergyHarvesting Device

Gateway SelectionAlgorithm

Network Lifetime

Fig. 4 Network lifetime under homogeneous scenario for the normalperson’s daily routine

J Med Syst (2012) 36:1593–1601 1599

GSA. In this scenario, we used five body pulse oxygensensors. Similar to the first model, we have assumed thatnetwork lifetime in the standard energy level situation isnormalized to 1 to use it as a base for comparison. Thestandard energy level and the only using energy harvestingconfigurations have almost same the network lifetime. TheGSA boosts the network lifetime by about 1.2 times.

Unlike the heterogeneous scenario of the first model, weused a motion sensor and a shoes pressure energy harvest-ing device instead of blood pressure sensor and movementsof hands energy harvesting device. As seen from theTable 2, the shoes pressure energy harvesting techniquegenerates more energy than the movements of hands. Onthe other hand, there are many reasons effecting the energy

consumption such as packet sending rate, distance tonetwork coordinator etc. Figure 8 shows the obtainednetwork lifetime results for the heterogeneous scenario forthe three different configurations. The GSA increases thenetwork lifetime of the WBASN by approximately 2.2times.

Conclusion

A WBASN enables monitoring patients’ conditions 24 hwithout disturbing their lifestyle. Continual network oper-ation requires careful management of the resources of theindividual nodes so that the WBASN sustains it function-

0

1

2

3

No Energy HarvestingDevice

Independent EnergyHarvesting Device

Gateway SelectionAlgorithm

Network Lifetime

Fig. 8 Network lifetime under heterogeneous scenario for the elderlyperson’s daily routine

0

0,5

1

1,5

No Energy HarvestingDevice

Independent EnergyHarvesting Device

Gateway SelectionAlgorithm

Network Lifetime

Fig. 7 Network lifetime under homogeneous scenario for the elderlyperson’s daily routine

Time(Hours) Action Duration

(Hours)Type of

Energy HarvestingDistance to NC

(m)0

Sleeping 8 Body's Heat 2

Casual Activity 2.5Movements of Hands

Shoes PressureBody's Heat

0.5 - 5

11 Eating 0.5Movements of Hands

Body's Heat 1 - 2

Walking 0.5Movements of Hands

Shoes PressureBody's Heat

0.2 - 1

12 Driving / Bus 0.5 Body's Heat 0.2 - 1

Hobby Activity(Reading/Watching)

1.5 Body's Heat 1 - 3

Eating 0.5 Movements of HandsBody's Heat

1 - 2

15 Shooping Activity 1Movements of Hands

Body's HeatShoes Pressure

0.5 - 5

Walking 0.5Movements of Hands

Shoes PressureBody's Heat

0.2 - 1

16 Driving / Bus 0.5 Body's Heat 0.2 - 1

Sleeping 2 Body's Heat 2

Casual Activity 2.5Movements of Hands

Shoes PressureBody's Heat

0.5 - 5

Eating 0.5 Movements of HandsBody's Heat

1 - 2

24Hobby Activity

(Reading/Watching) 3 Body's Heat 1 - 2

Insi

de (

Hom

e)O

utsi

deIn

side

(Caf

eter

ia/

Libr

ary/

Sho

pp)

Out

side

Insi

de(H

ome)

Fig. 6 General model for thedaily activities of an elder

1600 J Med Syst (2012) 36:1593–1601

ality for the longest time. In this paper, we have proposed anovel Gateway Selection Algorithm (GSA) for extendingthe network lifetime by optimized determination of thenode that interfaces the network to the outside world. Theidea is to dynamically adjust the data routing topology sothat the node with the highest energy reserve acts as thesink for all traffic. That sink node becomes the gateway thatinterfaces the network to doctors and caregivers. GSAfactors in both the energy reserve onboard a node and therate of scavenging from energy harvesting devices attachedto the body. GSA has been validated using OPNET. Thesimulation results showed that GSA could achieve up tothree times increase in the lifetime of a WBASN comparedto fixed data routing topology based schemes. Theincreased network lifetime makes GSA invaluable forincreasing the effectiveness of WBASN as a tool for elderand patient care.

Acknowledgment The work of C. Bayilmis was supported by TheScientific and Technological Research Council of Turkey (TÜBİTAK)during his visit as a Postdoctoral Researcher at University of MarylandBaltimore County between May 2009 and January 2010.

References

1. Hanson, M. A., et al., Body area sensor networks: Challenges andopportunities. IEEE Computer 42(1):58–65, 2009.

2. Cao, H., Leung, V., Chow, C., and Chan, H., Enablingtechnologies for wireless body area networks: Survey andoutlook. IEEE Commun. Mag. 84–92, December 2009.

3. Milenkovic, A., Otto, C., and Jovanov, E., Wireless Sensor Networksfor Personal Health Monitoring: Issues and an Implementation.Comput. Commun., Special issue on Wireless Sensor Networks:Performance, Reliability, Security, and Beyond 29:2521–2533, 2006.

4. Zhang, L., and Wu, X., Recent progress in challenges of wirelessbiomedical sensor network. In the Proceedings of the IEEE 3rdInternational Conference on Bioinformatics and BiomedicalEngineering (ICBBE 2009). Beijing, China, June 2009.

5. Yuce, M. R., Ng, P. C., and Khan, J. Y., Monitoring ofphysiological parameters from multiple patients using wirelesssensor network. J. Med. Syst. 32(5):433–441, 2008.

6. Reusens, E., et al., Characterization of on-body communicationchannel and energy efficient topology design for wireless body areanetworks. IEEE Trans. Inf. Technol. Biomed. 13(6):933–945, 2009.

7. Braem, B., et al., The need for cooperation and relaying in short-range high path loss sensor networks. In the Proceedings of the IEEEInternational Conference on Sensor Technologies and Applications(SENSORCOMM 2007). Valencia, Spain, October 2007.

8. Ehyaie, A., Hashemi, M., and Khadivi, P., Using relay network toincrease life time in wireless body area sensor networks. In theProceedings of the IEEE International Symposium on a World ofWireless, Mobile and Multimedia Networks & Workshops (WoWMoM2009). Kos, Greece, June 2009.

9. Starner, T., and Paradiso, J. A., Human generated power formobile electronics. In: Piguet, C. (Ed.), Low-Power ElectronicsDesign. CRC Press, Chapter 45, pp. 1–35, 2004.

10. Paradiso, J. A., and Starner, T., Energy scavenging for mobile andwireless electronics. IEEE Pervasive Comput. 4(1):18–27, 2005.

11. Amor, N. B., and Kanoun, O., Investigation to the use of vibrationenergy for supply of hearing aids. Instrumentation and Measure-ment Technology Conference-IMTC, Poland, 2007.

12. Amor, N. B., et al., Energy harvesting from human body forbiomedical autonomous systems. In the Proceedings of the IEEEon Sensors (SENSORS 2008), Leece, Italy, October 2008.

13. Stark, I., Thermal energy harvesting with thermo life. Proceedingsof the International Workshop on Wearable and Implantable BodySensor Networks (BSN’06), 2006.

14. Raju, M., ULP meets energy harvesting: a game-changingcombination for design engineers. Texas Instruments, WhitePaper, November 2008.

15. Omeni, O., Eljamaly, O., and Burdett, J. Energy efficient mediumaccess protocol for wireless body area sensor networks. In theProceedings of the 4th IEEE-EMBS International Summer Schooland Symposium on Medical Devices and Biosensors. Cambridge,UK, 29–32, 2007.

16. Omeni, O., Wong, A. C. W., Burdett, A. J., and Toumazou, C.,Energy efficient medium access protocol for wireless medicalbody area sensor networks. IEEE Trans. Biomed. Circuits Syst. 2(2):251–259, 2008.

17. Otal, B., Alonso, L., and Verikoukis, C., Highly reliable energy-saving MAC for wireless body sensor networks in healthcaresystems. IEEE J. Sel. Areas Commun. 27(4):553–565, 2009.

18. Ullah, S., and Kwak, K. S., An ultra low-power and traffic-adaptive medium access control protocol for wireless body areanetwork. J. Med. Syst. Online First™, 2 August 2010.

19. Ullah, S., et al., A comprehensive survey of wireless body areanetworks on PHY, MAC, and network layers solutions. J. Med.Syst. Online First™, 18 August 2010.

20. Hoang, D. C., Tan, Y. K., Chng, H. B., Panda, S. K., Thermalenergy harvesting from human warmth for wireless body areanetwork in medical healthcare system. In the Proceedings of theEight International Conference on Power Electronics and DriveSystems (PEDS2009). Taipei, Taiwan, pp. 1277–1282, November2009.

21. Welsh, M., Exposing resource tradeoffs in region-based commu-nication abstractions for sensor networks. SIGCOMM Comput.Commun. Rev. 34(1):119–124, 2004.

22. Shenck, N. S., and Paradiso, J. A., Energy scavenging with shoe-mounted piezoelectrics. IEEE Micro 21(3):30–42, 2001.

23. Chen, S., Lee, H., Chen, C., Huang, H., and Luo, C., Wirelessbody sensor network with adaptive low-power design forbiometrics and healthcare applications. IEEE Syst. J. 3(4):398–409, 2009.

J Med Syst (2012) 36:1593–1601 1601