Chapter 6 Medium Access Control Protocols and Local Area Networks 802.11 Wireless LAN.
A Study of Medium Access Control Protocols for Wireless ...A Study of Medium Access Control...
Transcript of A Study of Medium Access Control Protocols for Wireless ...A Study of Medium Access Control...
1
A Study of Medium Access Control Protocols forWireless Body Area Networks
Sana UllahΨ, Bin Shen, S.M.Riazul Islam, Pervez Khan, Shahnaz Saleem,and Kyung Sup Kwak
Abstract—The seamless integration of low-power, miniaturised, invasive/non-invasive lightweight sensor nodes havecontributed to the development of a proactive and unobtrusive Wireless Body Area Network (WBAN). A WBAN provideslong-term health monitoring of a patient without any constraint on his/her normal dailylife activities. This monitoringrequires low-power operation of invasive/non-invasive sensor nodes. In other words, a power-efficient Medium AccessControl (MAC) protocol is required to satisfy the stringent WBAN requirements including low-power consumption. In thispaper, we first outline the WBAN requirements that are important for the design of a low-power MAC protocol. Then westudy low-power MAC protocols proposed/investigated for WBAN with emphasis on their strengths and weaknesses.We also review different power-efficient mechanisms for WBAN. In addition, useful suggestions are given to help theMAC designers to develop a low-power MAC protocol that will satisfy the stringent WBAN requirements.
Index Terms—WBAN, MAC, Low Power, Survey, IEEE 802.15.6.
F
1 INTRODUCTIONA Wireless Body Area Network (WBAN) al-lows the integration of intelligent, miniatur-ized, low- power, invasive/non-invasive sensornodes in/around a human body that are usedto monitor body functions and the surroundingenvironment. Each intelligent node has enoughcapability to process and forward informationto a base station for diagnosis and prescription.A WBAN provides long term health moni-toring of patients under natural physiologicalstates without constraining their normal activ-ities. It is used to develop a smart and afford-able health care system and can be a part ofdiagnostic procedure, maintenance of a chroniccondition, supervised recovery from a surgicalprocedure, and can handle emergency events.
• S.Ullah, B.Shen, S.M.R. Islam, P. Khan, and K.S. Kwak are withthe Graduate School of Telecommunication Engineering, InhaUniversity, Incheon (402-751) South Korea.
• Ψ. Author to whom correspondence should be addressed E-mail:[email protected]
• S. Saleem is with the Graduate School of Computer Science andEngineering, Inha University, Incheon (402-751) South Korea.
Manuscript received November 26, 2009; revised December 7, 2009;published (Sensors, vol. 10, no. 1: 128-145.) December 20, 2009.
Some of the common objectives in WBAN areto achieve maximum throughput, minimumdelay, and to maximize the network lifetime bycontrolling the main sources of energy waste,i.e., collision, idle listening, overhearing, andcontrol packet overhead. A collision occurswhen more than one packet is transmitted atthe same time. The collided packets have tobe retransmitted, which consumes extra energy.The second source of energy waste is idlelistening, meaning that a node listens to anidle channel to receive data. The third sourceis overhearing, i.e., to receive packets that aredestined to other nodes. The last source iscontrol packet overhead, meaning that con-trol information are added to the payload. Aminimal number of control packets should beused for data transmission. Medium AccessControl (MAC) protocols play an importantrole in solving the aforementioned problems.Generally they are grouped into contention-based and schedule-based MAC protocols. Incontention-based MAC protocols such as Car-rier Sense Multiple Access/Collision Avoid-ance (CSMA/CA) protocols, nodes contend forthe channel to transmit data. If the channelis busy, the node defers its transmission until
arX
iv:1
004.
3890
v2 [
cs.N
I] 3
1 Ju
l 201
0
2
TABLE 1CSMA vs. TDMA Protocols
Performance Metric CSMA/CA TDMA
Power consumption High Low
Traffic level Low High
Bandwidth utilisation Low Maximum
Scalability Good Poor
Effect of packet failure Low Latency
Synchronisation Not Applicable Required
it becomes idle. These protocols are scalablewith no strict time synchronization constraint.However, they incur significant protocol over-head. In schedule-based protocols such as TimeDivision Multiple Access (TDMA) protocols,the channel is divided into time slots of fixedor variable duration. These slots are assigned tonodes and each node transmits during its slotperiod. These protocols are energy conservingprotocols. The duty cycle of radio is reducedand there is no contention, idle listening andoverhearing problems. These protocols, how-ever, require frequent synchronization. Table 1compares CSMA/CA and TDMA protocols.
The development of a low-power MAC pro-tocol for WBAN has been a hot researchtopic for the last few years. Considerable re-search efforts have been dedicated to pro-pose/investigate new MAC protocols thatcould satisfy the crucial WBAN requirements.A number of researchers have considered IEEE802.15.4 [1] for WBAN since it supports lowdata rate applications, but it is not enough tosupport high data rate applications (data rate> 250 Kbps). Other protocols such as HeartbeatDriven MAC (H-MAC) [2], Reservation-basedDynamic TDMA (DTDMA) [3], Preamble-based TDMA (PB-TDMA) [4], and BodyMAC[5] protocols have been proposed/investigatedin the existing literature. In this paper, anoverview of the aforesaid protocols with focuson their strengths and weaknesses is presented.Useful suggestions are given to overcome theweaknesses of these protocols. Then, a compar-ative analysis of many power-efficient mech-anisms such as Low-power Listening (LPL),scheduled-contention, and TDMA mechanismsis presented in the context of WBAN. Examples
are given to validate the discussion.The rest of the paper is categorized into four
sections. Section 2 presents the WBAN require-ments. A study of different low-power MACprotocols proposed/investigated for WBAN isgiven in Section 3. Section 4 reviews variouspower-efficient mechanisms for WBAN withuseful guidelines. The final section concludesour work.
2 WBAN MAC REQUIREMENTS
The most important attribute of a good MACprotocol for WBAN is energy efficiency. Insome applications, the device should supporta battery life of months or years without inter-vention, while others may require a battery lifeof only tens of hours due to the nature of theapplications. For example, cardiac defibrillatorsand pacemakers should have a lifetime of morethan 5 years, while swallowable camera pillshave a lifetime of 12 hours [6]. Power-efficientand flexible duty cycling techniques are re-quired to minimize the idle listening, overhear-ing, packet collisions and control packet over-head problems. Furthermore, low duty cyclenodes should not receive frequent synchroniza-tion and control information (beacon frames) ifthey have no data to send or receive.
The WBAN MAC should support simulta-neous operation on in-body (Medical ImplantCommunications Service, also called MICS)and on-body frequency bands/channels [In-dustrial, Scientific and Medical (ISM) or UltraWide Band (UWB)] at the same time. In otherwords, it should support Multiple Physical lay-ers (Multi-PHYs) communication. Other impor-tant factors are scalability and adaptability tochanges in the network, delay, throughput, andbandwidth utilization. Changes in the networktopology, the position of the human body, andthe node density should be handled rapidlyand successfully. The MAC protocol for WBANshould consider the electrical properties of thehuman body and the diverse traffic nature ofin-body and on-body nodes. For example, thedata rate of in-body nodes varies, ranging fromfew kbps in pacemaker to several Mbps incapsular endoscope. Fig. 1 shows some of thepotential issues of a MAC protocol for WBANs.
3
Fig. 1. Potential issues of a MAC protocol for WBAN
The Quality of Service (QoS) is also an im-portant factor of a good MAC protocol forWBAN. This includes point-to-point delay anddelay variation. In some cases, real-time com-munication is required for many applicationssuch as fitness and medical surgery monitoringapplications. For multimedia applications, thelatency should be less than 250 ms and thejitter should be less than 50 ms [6]. How-ever, the reliability, latency, and jitter require-ments depend on the nature of the applications.For emergency applications, the MAC protocolshould allow in-body or on-body nodes toget quick access to the channel (in less thanone second) and to send the emergency datato the coordinator. One such example is thedetection of irregular heartbeat, high or lowblood pressure or temperature, and excessivelylow or high blood glucose level in a diabeticpatient. Another example is when the nodeis dying. Reporting medical emergency eventsshould have higher priority than non-medicalemergency (battery dying) events.
In WBAN, most of the traffic is correlated,i.e., a patient suffering from fever triggers tem-perature, blood pressure, and respiration sen-sors at the same time [7]. These changes mayalso affect the oxygen saturation level (SpO2) inthe blood. These kinds of physiological param-eters increase the traffic correlation. A singlephysiological fluctuation triggers many sensorsat the same time. In this case, a CSMA/CAprotocol encounters heavy collisions and extraenergy consumption. Also, in CSMA/CA pro-
tocol, the nodes are required to perform ClearChannel Assessment (CCA) before transmis-sion. However, the CCA is not always guar-anteed in the MICS band since the path lossinside the human body due to tissue heatingis much higher than in free space. Bin et.alstudied the unreliability of a CCA in WBANand concluded that for a given -85 dBm CCAthreshold, the on-body nodes cannot see theactivity of the in-body nodes when they are at3 meters distance away from the surface of thebody [8]. Sana et al. have studied the behaviorof the CSMA/CA protocol for WBAN and con-cluded that the CSMA/CA protocol encountersheavy collision problems for high traffic nodes[9]. TDMA-based protocols provide good solu-tions to the traffic correlation, heavy collision,and CCA problems. These protocols are energyconserving protocols because the duty cycle isreduced and there are no contention, idle listen-ing, and overhearing problems. However, com-mon TDMA needs extra energy for periodictime synchronization. All the sensors (withand without data) are required to receive theperiodic packets in order to synchronize theirclocks. Therefore, the design and implementa-tion of a new TDMA protocol is required thatcan accommodate the heterogeneous WBANtraffic in a power-efficient manner.
4
TABLE 2IEEE 802.15.4 frequency bands, data rates, and modulation methods
Frequency Bands/Channels Coverage Sub-channels Data Rate Data Modulation Chip Modulation
2.4 GHz Worldwide 16 250kbit/s 16-ary orthogonal OQPSK,2Mchips/s
868 MHz Europe 1 20kbit/s BPSK BPSK,300kchips/s
915 MHz Americas 10 40kbit/s BPSK BPSK,600kchips/s
3 EXISTING/PROPOSED MAC PROTO-COLS FOR WBAN
3.1 IEEE 802.15.4
IEEE 802.15.4 is a low-power standard de-signed for low data rate applications. It offersthree operational frequency bands: 868 MHz,915 MHz, and 2.4 GHz bands. There are 27sub-channels allocated in IEEE 802.15.4, i.e., 16sub-channels in 2.4 GHz band, 10 sub-channelsin 915 MHz band and one sub-channel in the868 MHz band, as given in Table 2. The ta-ble also shows the data rate and the modula-tion technique for each frequency band. IEEE802.15.4 has two operational modes: a beacon-enabled mode and a non-beacon enabled mode.In a beacon-enabled mode, the network is con-trolled by a coordinator, which regularly trans-mits beacons for device synchronization andassociation control. The channel is bounded bya superframe(s) structure as illustrated in Fig.2.
Fig. 2. IEEE 802.15.4 superframe structure
The superframe consists of both active andinactive periods. The active period containsthree components: a beacon, a Contention Ac-cess Period (CAP), and a Contention Free Pe-riod (CFP). The length of the entire super-frame (Beacon Interval, BI) and the length ofthe active part of the superframe (SuperframeDuration, SD) are defined as follows
BI = aBaseSuperframeDuration2BO (1)
SD = aBaseSuperframeDuration2SO (2)
where aBaseSuperframeDuration=960 sym-bols, BO=Beacon Order, and SO=SuperframeOrder. The coordinator interacts with nodesduring the active period and sleeps duringthe inactive period. There are maximum sevenGuaranteed Time Slots (GTSs) in the CFP pe-riod to support time critical traffic. In a beacon-enabled mode, a slotted CSMA/CA protocolis used in the CAP period while in a non-beacon enabled mode, an unslotted CSMA/CAprotocol is used.
IEEE 802.15.4 has remained the main focusof many researchers during the past few years.Some of the main reasons of selecting IEEE802.15.4 for WBAN are low-power communi-cation and support of low data rate WBANapplications. Nicolas et al. investigated the per-formance of a non-beacon IEEE 802.15.4 in [10],where low upload/download rates (mostly perhour) are considered. They concluded thatthe non-beacon IEEE 802.15.4 results in 10to 15 years sensor lifetime for low data rateand asymmetric WBAN traffic. However, theirwork considered data transmission on the basisof periodic intervals, which is not a perfectscenario in a real WBAN. Furthermore, thedata rate of in-body and on-body nodes is notalways low, i.e., it ranges from 10 Kbps to 10Mbps, and hence reduces the lifetime of thesensor nodes. Li et al. studied the behaviourof slotted and unslotted CSMA/CA protocolsand concluded that the unslotted mechanismperforms better than the slotted one in terms ofthroughput and latency, but with a high powerconsumption cost [11].
Intel Corporation conducted a series of ex-periments to analyze the performance of IEEE
5
TABLE 3Throughput at a 0 dBm Transmit Power [12]
Person is Standing (Destination Nodes) Person is Sitting (Destination Nodes)Chest Waist Ankle Chest Waist Ankle
Chest - 99% 84% - 99% 81%Waist 100% - 50% 99% - 47%Ankle 72% 76% - 77% 27% -
TABLE 4Throughput and Power (in mW) of IEEE 802.15.4 and IEEE 802.11e under AC BE and AC VO
[15]
IEEE 802.15.4) IEEE 802.11e (AC BE) ) IEEE 802.11e (AC VO))Throughput Wave-form 100% 100% 100%
Parameter 99.77% 100% 100%Power (mW) Wave-form 1.82 4.01 3.57
Parameter 0.26 2.88 2.77
802.15.4 for WBAN [12]. They deployed anumber of Intel Mote 2 [13] nodes on chest,waist, and the right ankle. Table 3 shows thethroughput at a 0 dBm transmit power when aperson is standing and sitting on a chair. Theconnection between ankle and waist cannotbe established, even for a short distance of1.5 m. All other connections show favourableperformance.
Dave et al. studied the energy efficiencyand QoS performance of IEEE 802.15.4 andIEEE 802.11e [14] MAC protocols under twogeneric applications: a wave-form real timestream and a real-time parameter measurementstream [15]. Table 4 shows the throughput andthe Power (in mW) for both applications. TheAC BE and AC VO represent the access cate-gories best-effort and voice in the IEEE 802.11e,respectively.
Since the IEEE 802.15.4 operates in the 2.4GHz unlicensed band, the possibilities of inter-ference from other devices such as IEEE 802.11and microwave are inevitable. A series of ex-periments to evaluate the impact of IEEE 802.11and microwave ovens on the IEEE 802.15.4transmission are carried out in [16]. The au-thors considered XBee 802.15.4 developmentkit having two XBee modules. Table 5 showsthe effects of microwave oven on the XBeeremote module. When the microwave oven is
ON, the packet success rate and the standarddeviation (Std.) is degraded to 96.85% and3.22% respectively. However, there is no losswhen the XBee modules are taken two metersaway from the microwave oven.
3.2 Heartbeat Driven MAC (H-MAC) Proto-colA Heartbeat Driven MAC protocol (H-MAC)is a TDMA-based protocol originally proposedfor a star topology WBAN. The energy ef-ficiency is improved by exploiting heartbeatrhythm information in order to synchronize thenodes. The nodes do not need to receive peri-odic information to perform synchronization.The heartbeat rhythm can be extracted fromthe sensory data and hence all the rhythmsrepresented by peak sequences are naturallysynchronized. The H-MAC protocol assignsdedicated time slots to each node to guaran-tee collision-free transmission. This protocol issupported by an active synchronization recov-ery scheme supported by two resynchroniza-tion schemes.
Although H-MAC protocol reduces extra en-ergy cost required for synchronization, it doesnot support sporadic events. Since the TDMAslots are dedicated and are not traffic adap-tive, H-MAC protocol encounters low spec-tral/bandwidth efficiency in case of low traffic.
6
TABLE 5Co-existence Test Results between IEEE 802.15.4 and Microwave Oven [16]
Microwave Status Packet Success Rate (Mean) Packet Success Rate (Std.)
ON 96.85% 3.22%
OFF 100% 0%
For example, a blood pressure node may notneed a dedicated time slot while an endoscopepill may require a number of dedicated timeslots when deployed in WBAN. But the slotsshould be released when the endoscope pillis expelled. The heartbeat rhythm informationvaries depending on the patient condition. Itmay not reveal valid information for synchro-nization all the time. One of the solutions is toassign the time slots based on the node’s traf-fic information and to receive synchronizationpackets when required, i.e., when a node hasdata to transmit/receive.
3.3 Reservation-based Dynamic TDMA (DT-DMA) Protocol
A Reservation-based Dynamic TDMA Protocol(DTDMA) is originally proposed for normal(periodic) WBAN traffic where slots are allo-cated to the nodes having buffered packetsand are released to other nodes when thedata transmission/reception is completed. Thechannel is bounded by superframe structures.Each superframe consists of a beacon - used tocarry control information including slot alloca-tion information, a CFP period - a configurableperiod used for data transmission, a CAP pe-riod - a fixed period used for short commandpackets using slotted-ALOHA protocol, anda configurable inactive period - used to saveenergy. Unlike a beacon-enabled IEEE 802.15.4superframe structure where the CAP durationis followed by CFP duration, in DTDMA pro-tocol the CFP duration is followed by CAPduration in order to enable the nodes to sendCFP traffic earlier than CAP traffic. In addition,the duration of an inactive period is config-urable based on the CFP slot duration. If thereis no CFP traffic, the inactive period will beincreased. The DTDMA superframe structureis given in Fig. 3.
Fig. 3. DTDMA superframe structure
It has been shown that for normal (periodic)traffic, the DTDMA protocol provides moredependability in terms of low packet droppingrate and low energy consumption when com-pared with IEEE 802.15.4. However, it doesnot support emergency and on-demand traffic.Although the slot allocation based on the trafficinformation is a good approach, the DTDMAprotocol has several limitations when consid-ered for the MICS band. The MICS band hasten sub-channels where each sub-channel has300 Kbps bandwidth. The DTDMA protocolcould operate on one sub-channel but cannotoperate on ten sub-channels simultaneously.In addition, the DTDMA protocol does notsupport the sub-channel allocation mechanismin the MICS band. This protocol can be fur-ther investigated for the MICS band by in-tegrating the sub-channel information in thebeacon frame. The new concept may be calledFrequency-based DTDMA (F-DTDMA), i.e., thecoordinator first selects one of the sub-channelsin the MICS band and then divides the selectedsub-channel in TDMA superframe(s) accordingto the DTDMA protocol. However the Fed-eral Communications Commission(FCC) hasimposed several restrictions on the sub-channelselection/allocation mechanism in the MICSband (see Section 4.4), which further createsproblems for the MAC designers.
7
3.4 PB-TDMA ProtocolThe performance of a Preamble-based TDMA(PB-TDMA, see Section 4.3) for WBAN hasbeen analyzed in [17]. The authors used NS-2 for extensive simulations where the wirelessphysical parameters were considered accordingto low-power Nordic nRF2401 transceiver andthe simulation area was 3 × 3 meters. For theperformance comparison, many other protocolssuch as S-MAC and IEEE 802.15.4 were alsosimulated in the same environment. Simulationresults showed that PB-TDMA protocol out-performed S-MAC and IEEE 802.15.4 protocolin terms of energy efficiency. The results arevalid for normal traffic only. The co-existenceof emergency and on-demand traffic is notconsidered in the simulation.
3.5 BodyMAC ProtocolA BodyMAC protocol is a TDMA-based pro-tocol where the channel is bounded by TDMAsuperframe structures with downlink and up-link subframes as given in Fig. 4. The downlinkframe is used to accommodate the on-demandtraffic and the uplink frame is used to accom-modate the normal traffic. There is no propermechanism to handle emergency traffic. Theuplink frame is further divided into CAP andCFP periods. The CAP period is used to trans-mit small size MAC packets. The CFP periodis used to transmit the normal data in a TDMAslot. The duration of the downlink and uplinksuperframes are defined by the coordinator.
Fig. 4. BodyMAC superframe structure
The advantage of the BodyMAC protocolis that it accommodates the on-demand traf-fic using the downlink subframe. However, incase of low-power implants (which should notreceive beacons periodically), the coordinatorhas to wake up the implant first and then
send synchronization packets. After synchro-nization, the coordinator can request/send datain the downlink subframe. The wake up pro-cedure for low-power implants is not definedin the BodyMAC protocol. One of the solutionsis to use a wakeup radio in order to wake uplow-power implants before using the down-link subframe. In addition the wakeup packetscan be used to carry control information suchas sub-channel (MICS band) and slot alloca-tion information from the coordinator to thenodes. Finally, the BodyMAC protocol uses theCSMA/CA protocol in the CAP period whichis not reliable for WBAN. This should be re-placed by slotted-ALOHA as done in DTDMA.
4 POWER-EFFICIENT MECHANISMS FORWBAN
Power-efficient mechanisms play an importantrole in the performance of a good MAC pro-tocol. These mechanisms are categorized intoLow-power Listening (LPL), Scheduled Con-tention, and TDMA mechanisms. The follow-ing sections briefly explain each mechanismwith examples.
4.1 Low-power Listening (LPL) MechanismIn a Low-power Listening (LPL) mechanism,nodes wake up for a short duration to checkthe channel activity without receiving any data.If the channel is idle the nodes go into sleepmode, otherwise they stay on the channel toreceive the data. This is also called channelpolling. The LPL is performed on regular basisregardless of synchronization among nodes.The sender sends a long preamble before eachmessage in order to detect the polling at thereceiving end.
The WiseMAC [18] protocol is based onthe LPL mechanism. In this protocol, a non-persistent CSMA and a preamble samplingtechnique is used to reduce idle listening. Thepreamble is used to alert the receiving node of apacket arrival. All the nodes in a network sam-ple the medium periodically. If a node samplesa busy medium, it continues to listen until itreceives data or the medium becomes idle. Fig.5 shows the WiseMAC concept.
8
Fig. 5. WiseMAC concept
In WBAN, the LPL mechanism has severaladvantages and disadvantages. The periodicsampling is efficient for high-traffic nodes andperforms well under variable traffic conditions.However, it is ineffective for low-traffic nodes,especially in-body nodes, where periodic sam-pling is not preferred due to strict power con-straints. Since the WBAN topology is a startopology and most of the traffic is uplink, usingLPL mechanism is not an optimal solution tosupport both in-body and on-body communi-cation simultaneously.
4.2 Scheduled-Contention MechanismIn a scheduled-contention mechanism, sched-uled and contention based schemes are com-bined to incur scalability and collision avoid-ance. In this mechanism, the nodes adapt acommon schedule for data communication.The schedules are exchanged periodically dur-ing a synchronization period. If two neighbour-ing nodes reside in two different clusters, theykeep the schedules of both clusters resulting inextra energy consumption.
The S-MAC [19] protocol is a good exampleof a scheduled-contention mechanism. S-MACis a power-efficient and a contention-basedMAC protocol designed for multi-hop WirelessSensor Networks (WSNs). In this protocol, thelow duty cycle mode is default operation ofall nodes. This protocol introduces the conceptof coordinated sleeping among neighbouringnodes. The node is active when it has data tosend otherwise its radio is completely turnedoff. The energy is reduced from all the sourcesof energy waste, i.e., idle listening, collision,overhearing and control overhead. A complete
cycle of listen and sleep is called frame. Eachframe begins with a wakeup period, which isused by nodes to exchange control informa-tion. The wakeup period is usually followedby a sleep period. If a node has data to sendwhile in the sleep mode, it must defer itstransmission until the next wakeup slot. Thestate of the channel is determined using phys-ical and virtual carrier sense mechanism. Foreach unicast frame transmission, Request toSend/Clear to Send (RTS/CTS) mechanism isfollowed. Broadcast frames are sent withoutusing RTS/CTS mechanism. If a node fails toaccess the medium, it turns off its radio untilthe Network Allocation Vector (NAV) is zero.Nodes maintain synchronization by sendingSYNCH packet. The size of SYNCH packet isvery short and includes information about nextsleep period. The listen period of a node isdivided into two parts when both SYNCH anddata packets are received at the same time. Fig.6 illustrates the timing relationship between areceiver and different senders. Sender 1 sendsa SYNCH packet only. Sender 2 sends a unicastdata packet only. Sender 3 sends both a SYNCHand a data packet.
Fig. 6. Timing relationship between a receiverand different senders. CS is carrier sense.
A scheduled-contention mechanism reducesidle listening using sleep schedules and per-forms well for multi-hop WSNs. However, con-sidering this mechanism for WBAN revealsseveral problems for low-power in-body/on-body nodes such as pacemakers and defibril-lator nodes/implants, which are not requiredto wake up periodically in order to exchangetheir schedules with other nodes. Furthermore,scheduled-contention mechanism may performwell for on-body applications but it does notprovide reliable solutions to handle sporadic
9
events including emergency and on-demandevents. Handling sporadic events (emergency)requires innovative solutions that allow in-body/on-body nodes to update the coordinatorwithin strictly limited amount of time.
4.3 TDMA MechanismIn a TDMA mechanism, the channel isbounded by a superframe structure that con-sists of a number of time slots allocated by abase-station or a coordinator. The time slots areallocated according to the traffic requirements,i.e., a node gets a time slot whenever it has datato send or receive. Otherwise, it goes into sleepmode. Although it performs well in terms ofpower consumption but consumes extra energydue to frequent synchronization.
The PB-TDMA protocol is based on theTDMA mechanism. In this protocol, the nodesare assigned specified slots for collision-freedata transmission. These slots are repeated infixed cycle. A complete cycle of these slotsis called a frame. In the PB-TDMA protocol,each TDMA frame contains a preamble and adata transmission slot as illustrated in Fig. 7.A node always listens to the channel duringpreamble and transmits in a data transmissionslot. The preamble contains a dedicated subslotfor every node. These subslots are used toactivate the destination node by broadcastingits ID before transmission. After receiving thepreamble, the destination node identifies thesource node. Each node turns off its radio whenit has no data to transmit. This mechanismavoids unnecessary power consumption of sen-sor nodes. The radio is turned on when thenode finds its ID in the preamble or when thenode has data to transmit.
Fig. 7. PB-TDMA superframe structure
As discussed in Section 2, the CSMA/CAprotocol is not a reliable protocol for WBANdue to unreliable CCA, traffic correlation, and
heavy collision problems. The alternative is toadapt a TDMA protocol that can solve theaforementioned problems in a power-efficientmanner. However, traditional TDMA protocolssuch as PB-TDMA have several problems, i.e.,preamble overhearing and limitation of han-dling sporadic events. Solving these problems(including many others) towards WBAN canaccommodate the heterogeneous WBAN trafficin a power-efficient manner. Furthermore, newtechniques are required to accommodate spo-radic events in a reliable way.
4.4 Comparison of LPL, Scheduled-contention and TDMA Mechanisms forWBAN
Table 6 presents the characteristics of the LPL,schedule-contention, and TDMA mechanismsfor WBAN [20]. The table shows that LPL andscheduled-contention are unable to accommo-date the heterogeneous WBAN traffic includ-ing sporadic events. Although it is possible todevelop new MAC protocols based on thesemechanisms, they will not be able to satisfyall the requirements. For example, LPL mech-anisms may perform well in case of periodictraffic but they are unable to accommodateaperiodic (unpredictable sporadic events) traf-fic and low duty cycle nodes. Furthermore, thescheduled-contention mechanisms are unableto cover in-body nodes, which do not requirefrequent synchronization or exchange of theirschedules. The TDMA mechanisms providegood solutions to the variable WBAN traffic.The slots can be assigned according to the traf-fic volume of a node. Although in traditionalTDMA protocols, nodes are required to syn-chronize at the beginning of each superframeboundary, but this approach can be optimizedfor nodes that do not require frequent syn-chronization. One way is to skip the synchro-nization control packets such as the beacon.The control packets (beacons) can be receivedwhen the low duty cycle nodes have data tosend or receive. A detailed comparison of MACprotocols based on LPL, scheduled-contention,and TDMA mechanisms for WBAN is given in
10
TABLE 6Comparison of LPL, Scheduled-contention, and TDMA mechanisms for WBAN
LPL Scheduled-contention TDMA
10 times less expensive than listen-ing for full contention period
Listening for full contention period Low duty cycle
Asynchronous Synchronous Synchronous-Fine grained timesynchronisation
Sensitive to tuning for neighbour-hood size and traffic rate
Sensitive to clock drift Very sensitive to clock drift
Poor performance when trafficrates vary greatly (Optimised forknown periodic traffic)
Improved performance with in-crease in traffic
Limited throughput and number ofactive nodes
Receiver and polling efficiency isgained at the much greater cost ofsenders
similar cost incurred by sender andreceiver
Require clustering
challenging to adapt LPL directly tonew radios like IEEE 802.15.4
Scalable, adaptive, and flexible Limited scalability and adaptabilityto changes on number of nodes
Unable to accommodate aperi-odic traffic (unpredictable sporadicevents) and low duty cycle nodesin WBAN. Very hard to satisfythe WBAN traffic heterogeneity re-quirements
Low duty cycle nodesdo not require frequentsynchronization/exchange ofschedules in WBAN. Hardto satisfy the WBAN trafficheterogeneity requirements
Low duty cycle nodes do not re-quire frequent synchronization atthe beginning of each superframe.Easy to satisfy the WBAN trafficheterogeneity requirements
Table 71.From the following table, it can be seen that
LPL-based protocols such as WiseMAC andBMAC [21] protocols are good for high trafficapplications while STEM [22] performs wellfor low traffic applications. Furthermore, STEMcan also accommodate the sporadic events byusing a separate control channel. However,increase in the traffic load decreases the prob-ability of handling sporadic events. Schedule-contention protocols such as SMAC and TMAC[23] are suitable for high traffic applications,PMAC [24] for delay sensitive applications,and DMAC [25] for priority-based applications(where the nodes have different priorities). Asmentioned earlier, TDMA-based protocols caneasily accommodate the heterogeneous WBANtraffic since they are adaptable to the trafficload, i.e., slots can be assigned according to thetraffic volume. However, traditional TDMA-based protocols such as FLAMA [26], LEACH[27], and HEED [28] are unable to satisfy theWBAN requirements as mentioned in the fol-
1. The protocols given in the table are not explained heredue to space limitation problems. The table is a result ofa comprehensive study of these protocols in the context ofWBAN
lowing table. In addition, Most of the exist-ing MAC protocols are designed for a sin-gle channel only, i.e., they do not operate onMulti-PHYs simultaneously. The MAC trans-parency has been a hot topic for the MACdesigners since different bands have differ-ent characteristics in terms of data rate, num-ber of sub-channels in a particular frequencyband/channel, and data prioritization. A goodMAC protocol for WBAN should enable reli-able operation on MICS, ISM, and UWB etcbands simultaneously. With regards to MICSband, the FCC has imposed several restrictions[29]. According to the FCC
Within 5 seconds prior to initiating a commu-nications session, circuitry associated with a med-ical implant programmer/control transmitter mustmonitor the channel (sub-channel) or channels (sub-channels) the MICS system devices intend to oc-cupy for a minimum of 10 milliseconds per channel.
In other words, the coordinator must per-form Listen-before-talking (LBT) criteria priorto establish a MICS communication sessionsThe sub-channels are solely assigned by thecoordinator, i.e., the implants cannot initiatea communication session except in case of anemergency event. Furthermore, the implants
11
TABLE 7Summary of existing MAC Protocols for WBAN
Power-efficientMecha-nisms
Protocols Channels Organization and Basic Op-eration
Advantages and Disadvan-tages
Adaptability to WBAN
LowpowerListening
WiseMAC 1 Organized randomly andoperation is based on listen-ing
Scalable and adaptive totraffic load, Support mobil-ity, low and high power con-sumption in low and hightraffic conditions, and lowdelay
Good for high traffic appli-cations, not suitable for lowduty cycle in-body/on-bodynodes
BMAC 1 Organized in slots and oper-ation is based on schedules
Flexible, high throughput,tolerable latency, and low-power consumption
Good for high traffic appli-cations
STEM 2 Organized randomly havingtwo sub-channels (control +data channel) and operationis based on wakeup sched-ules
Suitable for events based ap-plications
Good for periodic traffic es-pecially for low traffic ap-plications. Suitable to han-dle sporadic events dueto a separate control sub-channel. But hard to handlesporadic events when thetraffic load is high
Scheduled-contention
SMAC 1 Organized in slots and oper-ation is based on schedules
High transmission latency,loosely synchronized, lowthroughput
Good for high traffic appli-cations. Suitable for appli-cations where throughput isnot a primary concern suchas in-body medical applica-tions
TMAC 1 Organized in slots and oper-ation is based on schedules
Queued packets are sent ina burst thus achieve betterdelay performance, looselysynchronized
Good for high traffic ap-plications. Early sleep prob-lems allow the nodes toloose synchronization
PMAC 1 Organized in hybrid modeand operation is based onlistening
Adaptation to changesmight be slow, looselysynchronized, highthroughput under heavytraffic
Good for delay-sensitive ap-plications
DMAC 1 Organized in slots and oper-ation is based on schedules
better delay performancedue to Sleep schedules,loosely synchronized,optimized for dataforwarding sink
On-body nodes can be prior-itized according to their ap-plication requirements and adata tree can be built, wherethe WBAN coordinator canbe a cluster node
TDMA FLAMA 1 Organized in frames andoperation is based on sched-ules
Better end-to-end reliabilityand energy saving, smallerdelays, high reliability
Good for low-power appli-cations. Adaptable to hightraffic applications
LEACH 1 Organized in clusters andoperations is based onTDMA scheme
Distributed protocol, noglobal knowledge required,extra overhead for dynamicclustering
TDMA schedules should becreated by the WBANcoordinator. Clusterhead should not change(depending on minimumcommunication energy) asin the traditional LEACH
HEED 1 Organized in clusters andoperations is based onTDMA scheme
Good for energy efficiency,scalability, prolonged net-work lifetime, load balanc-ing
The WBAN coordinator actsas a cluster head. Unlike tra-ditional HEED, the WBANnetwork size is often de-fined (by the physician)
12
are not allowed to perform LBT which createsproblems to handle emergency events. Send-ing an emergency data regardless of LBT mayresult in heavy collision since the selected sub-channel may have data from another implant.The LBT restriction prevents MAC designers todevelop a reliable mechanism for emergencytraffic. One of the solutions is to use a wakeupradio or a control sub-channel dedicated tem-porarily to emergency traffic since the FCCdoes not allow the permanent dedication of asub-channel in MICS band. The sub-channel in-formation can be updated using control framessuch as the beacon frame. Fig. 8 illustrates anexample of a control sub-channel used to sendemergency data. It shows that five nodes (1, 2,5, 7 and 8) are transmitting normal data andone node (3) is transmitting emergency data.The remaining nodes (4, 6 and 9) are in sleepmode. Normal transmission requires beacon toallocate resources. While in emergency case,nodes are not required to wait for the beacon.The communication is initiated by the implantsand a control sub-channel can be used to sendthe emergency data/command as given in thefigure.
Fig. 8. An example of a control sub-channel foremergency traffic
5 CONCLUSIONS
This paper presented a comprehensive study ofexisting/proposed MAC protocols for WBANwith useful suggestions. Different power-efficient mechanisms such as LPL, schedule-contention, and TDMA mechanisms were an-alyzed and discussed in the context of WBAN.As discussed, the CSMA/CA protocol encoun-ters heavy collision and unreliable CCA prob-lems, hence the TDMA protocol was consid-
ered the most reliable and power-efficient pro-tocol for WBAN. However, existing TDMAprotocols have a number of limitations interms of synchronization overhead, dynamicslot assignment, and Multi-PHYs communica-tion. Therefore, a novel low-power MAC pro-tocol (probably based on a TDMA mechanism)is required to satisfy the traffic heterogeneityand correlation, MAC transparency, and reli-ability requirements. This study can be usedas a guideline (for the beginners) towards thedesign and development of a new low-powerMAC protocol for WBAN.
ACKNOWLEDGEMENT
This research was supported by the The Min-istry of Knowledge Economy (MKE), Korea,under the Information Technology ResearchCenter (ITRC) support program supervised bythe Institute for Information Technology Ad-vancement (IITA) (IITA-2009-C1090-0902-0019).
REFERENCES
[1] IEEE Std.802.15.4: Wireless medium access con-trol (MAC) and physical layer (PHY) specificationsfor low data rate wireless personal area networks(WPAN); IEEE: Piscataway, NJ, USA, 2006.
[2] Li, H.M.; Tan, J.D. Heartbeat driven MAC for bodysensor networks, In Proceedings of the 1st ACMSIGMOBILE international workshop on systems andnetworking support for healthcare and assisted liv-ing environments, San Juan, Puerto Rico, 2007; pp.25-30.
[3] Li, C.; Li, H.B.; Kohno, R. Reservation-based dy-namic TDMA protocol for medical body area net-works. IEICE Trans. Commun. 2009, 92, 387-395.
[4] Ullah, S.; Islam, S.M.R.; Nessa, A.; Zhong, Y.; Kwak,K.S. Performance analysis of preamble based TDMAprotocol for wireless body area network. J. Commun.Softw. Syst. 2008, 4, 222-226.
[5] Fang, G.; Dutkiewicz, E. BodyMAC: Energy efficientTDMA-based MAC protocol for wireless body areanetworks. In Proceedings of IEEE ISCIT 2009, In-cheon, Korea, 2009.
[6] IEEE 802.15.6, Technical Requirements Document;IEEE: Piscataway, NJ, USA, 2008.
[7] Li, H.M.; Tan, J.D. An ultra-low-power medium ac-cess control protocol for Body Sensor Network. InProceedings of 27th Annual International Conferenceof the Engineering in Medicine and Biology Society,Shanghai, China, 2005; pp. 2451-2454.
[8] Zhen, B.; Li, H.B.; Kohno, R. IEEE body area net-works and medical implant communications. In Pro-ceedings of the ICST 3rd International Conference onBody Area Networks, Tempe, AZ, USA, 2008.
13
[9] Ullah, S.; Kwak, D.; Lee, C.; Lee, H.; Kwak, K.S.Numerical Analysis of CSMA/CA for Pattern-BasedWBAN System. In Proceedings of Biomedical Engi-neering and Informatics, Tianjin, China, 2009.
[10] Timmons, N.F.; Scanlon, W.G. Analysis of the perfor-mance of IEEE 802.15.4 for medical sensor body areanetworking, In Proceedings of IEEE SECON, SantaClara, CA,USA, 2004; pp. 4-7, 16-24.
[11] Li, C.; Li, H.B.; Kohno, R. Performance evaluationof IEEE 802.15.4 for wireless body area network(WBAN). In Proceedings of IEEE International Con-ference on ICC Workshops, Dresden, Germany, 2009;pp.1-5, 14-18.
[12] Shah, R.C.; Yarvis, M. Characteristics of on-body802.15.4 networks, In Proceedings of 2nd IEEE Work-shop on Wireless Mesh Networks, Reston, VA, USA,2006; pp.138-139.
[13] Adler, R.; Flanigan, M.; Huang, J.; Kling, R.; Kushal-nagar, N.; Nachman, L.; Wan, C.Y.; Yarvis, M. IntelMote 2: An advanced platform for demanding sen-sor network applications. In Proceedings of DemoSession, San Diego, CA, USA, 2005.
[14] IEEE 802.11e Std ”Amendment to Part 11: WirelessLAN Medium Access Control (MAC) and PhysicalLayer (PHY) specifications: Medium Access ControlQuality of Services Enhancements; IEEE: Piscataway,NJ, USA, 2005.
[15] Cavalcanti, D.; Schmitt, R.; Soomro, A. PerformanceAnalysis of 802.15.4 and 802.11e for body sensornetwork applications. In Proceedings of 4th Interna-tional Workshop on Wearable and Implantable BodySensor Networks, Aachen, Germany, 2007.
[16] Chen, C.; Carlos P.R. Monitoring human movementsat home using wearable wireless sensors, In Proceed-ings of ISMICT 09, Montreal, QC, Canada, 2009.
[17] Ullah, S.; Higgins, H.; Islam, S.M.R; Khan, P.; Kwak,K.S. On PHY and MAC performance in body sensornetworks. EURASIP J. Wirel. Commun. Netw. 2009,doi:10.1155/2009/479512.
[18] El-Hoiydi, A.; Decotignie, J.D. WiseMAC: An ultralow power MAC protocol for the downlink of infras-tructure wireless sensor networks. In Proceedingsof the Ninth IEEE Symposium on Computers andCommunication, ISCC’04, Alexandria, Egypt, 2004;pp. 244-251.
[19] Ye. W.; Estrin, H.J. An energy-efficient MAC proto-col for wireless sensor networks, In Proceedings ofTwenty-First Annual Joint Conference of the IEEEComputer and Communications Societies, Miami,FL, USA, 2002; pp. 1567-1576.
[20] Chebbo, H. Literature Review of Energy EfficientMAC in WSN/BAN; IEEE: Piscataway, NJ, USA,2008
[21] Polastre, J.; Hill, J.; Culler, D. Versatile low powermedia access for wireless sensor networks. In Pro-ceedings of the 2nd International Conference onEmbedded Networked Sensor Systems, New York,USA, 2004; pp. 95-107.
[22] Schurgers, C.; Tsiatsis, V.; Srivastava, M.B., STEM:Topology management for energy efficient sensornetworks. In Proceedings of Aerospace Conference,Manhattan Beach, CA, USA, 2004; Volume 3, pp. 3-1099-3-1108.
[23] Dam, V.T.; Langendoen. K. An adaptive energy-efficient mac protocol for WSNs. In Proceedings of
the 1st ACM Conference on Embedded NetworkedSensor Systems (Sen-Sys), Los Angeles, CA, 2003; pp.171-180.
[24] Khan, N.P.; Boncelet, C., PMAC: Energy efficientmedium access control protocol for wireless sensornetworks. In Proceedings of IEEE Military Commu-nications Conference, Washington, DC, USA, 2006;pp.1-5, 23-25.
[25] Lu, G.; Krishnamachari, B.; Raghavendra, C. AnAdaptive Energy-Efficient and Low-Latency MACdor Data Gathering in Sensor Networks. In Pro-ceedings of the 4th International Workshop on Al-gorithms for Wireless, Mobile, Ad Hoc and SensorNetworks, Santa Fe, Mexico, 2004.
[26] Rajendran, V.; Garcia-Luna-Aveces, J.J.; Obraczka, K.Energy-efficient, application-aware medium accessfor sensor networks. In Proceedings of IEEE MobileAdhoc and Sensor Systems Conference, Washington,DC, USA, 2005; p. 8 ,630.
[27] Heinzelman, W.R.; Chandrakasan, A.; Balakrishnan,H. Energy-efficient communication protocol for wire-less microsensor networks, In Proceedings of the33rd Annual Hawaii International Conference onSystem Sciences, Maui Kula, HI, USA, 2000.
[28] Younis, O.; Fahmy, S., Heed: a hybrid, energy-efficient, distributed clustering approach for ad hocsensor networks. IEEE Trans. Mob. Comp. 2004, 3,366-379.
[29] Index of Bureaus. Available online:http://www.fcc.gov/Bureaus/ (accessed on 16August 2009).