Implementation of the Adaptive CSMA-CA Algorithm Based on Traffic Load

Transcription

1 Internatonal Journal of Future Generaton Communaton and Networkng Vol.7, No.4 (204), pp Implementaton of the Adaptve CSMA-CA Algorthm Based on Traff Load Ln Zhu, Guangmng L * and We Lu Shool of Mehanal, Eletral and Informaton Engneerng, Shandong Unversty, Weha, , Chna zhulngshandong@gmal.om, gml@sdu.edu.n, luwe206.sw@63.om Abstrat IEEE s a standard desgned for low rate wreless personal area network. For ts MAC protool adopts the ompetton method to share wreless hannel, ollsons between endng deves are unavodable and lead to retransmssons, onsumng more energy. As traff load beomes heaver, ollsons and retransmssons are more serous. The fous of ths paper s to aheve dynam-parameter CSMA-CA algorthm based traff load n IEEE and measure the performane of modfed IEEE n some aspets (paket delvery rate, ollson rate, ollson rate and average MAC delay). Dfferent from prevous papers, t analyzes the performane of the algorthm n dret data transmsson and respetvely exeutes smulatons both nludng and exludng hdden-termnal (HT) problems. Keywords: IEEE ; CSMA-CA; dynam-parameter; dret data transmsson. Introduton In reent years, the onept of wreless sensor networks (WSNs) has gradually penetrated nto varous applatons suh as ntellgent transportaton, envronment montorng, and smart home for ts features of low power onsumpton, low ost, short transmsson dstane, moderate response tme and good survvablty. Servng as one of promsng anddates for WSNs, IEEE s a short-dstane wreless ommunaton standard. Low power onsumpton s the most mportant feature of the standard. Prevous work for IEEE an be lassfed roughly nto the followng ases. Varous performane metrs and expermental senes are desgned to evaluate the performane of IEEE []. Markov hans based analytal model are proposed to evaluate the slotted CSMA-CA algorthm n the MAC layer. MAC protool wth ACK frame transmsson [2], retransmssons [3] and throughput lmt [4] are studed wth the above-mentoned models. Bottleneks of IEEE are analyzed and dfferent mehansms are proposed to enhane ts performane [5]. A two-stage adaptve ontenton ontrol strategy s mplemented and evaluated on hardware to mprove transmsson effeny n ndret transmsson type [6], but the ommonly dret data transmsson and hdden-termnal problems are not onsdered. In ths paper, the nfluene of traff load on performane metrs of IEEE s study. The memorzed adaptve sheme s extended to dret data transmsson type and the performane of the adaptve algorthm s examned n two stuatons, onsderng or removng hdden-termnals. The rest of ths paper s organzed as follows: In Seton 2, we gve a bref desrpton of IEEE MAC layer. Next, n Seton 3, the adaptve CSMA-CA algorthm s ntrodued. In Seton 4, the analyss model of the modfed s gven. Then we present the ISSN: IJFGCN Copyrght 204 SERSC

2 Internatonal Journal of Future Generaton Communaton and Networkng Vol.7, No.4 (204) expermental setup and gve out the expermental results wth dsussons n Seton 5. Fnally, n Seton 6, we present the onluson. 2. A Bref Desrpton of IEEE MAC Layer Aordng to IEEE spefaton [7], the MAC sublayer provdes an nterfae between the serve spef onvergene sublayer (SSCS) and the PHY layer. Its man funtons nlude the followng aspets. The oordnator generates beaons perodally and deves synhronze wth the oordnator by trakng beaons. Assoaton and dsassoaton are appled to support self-onfguraton. Deves ontend hannel by the arrer sense multple aess wth ollson avodane mehansm (CSMA-CA). The relablty of the lnk between two peers s enhaned by varous mehansms, suh as frame aknowledgment, retransmsson and data verfaton. In ths subseton, we gve a desrpton of tems losely related to our performane study, nludng beaon-enabled mode, data transfer types. 2.. Baon-enabled Mode An IEEE network an operate n ether beaon-enabled mode or non beaon-enabled mode. In non beaon-enabled mode deves an send data at any tme based on CSMA/CA. Whle n beaon-enabled mode, deves must synhronze wth the oordnator and send data based on CSMA/CA at the boundares of tme unts alled bakoff slots. In IEEE standard, a superframe struture s used n beaon-enabled mode to manage perod beaon transmssons and data transmssons. The superframe s bounded by network beaons sent by the oordnator. It nludes three parts: a beaon transmsson part, an atve part and an natve part. At the begnnng of a superframe, a beaon shall be transmtted. Then the atve part shall ommene. Aordng to dfferent transmsson modes, the atve part s dvded nto a ontenton aess perod (CAP) and a ontenton-free perod. Durng the CAP, any deve wshng to ommunate ompetes wth other deves usng the slotted CSMA-CA mehansm. The CFP, f present, follows mmedately after the CAP and extends to the end of the atve part. Data transmssons wthn the CFP shall be ompleted n ertan slots alloated by the oordnator usng guaranteed tme slots (GTSs) mehansm. All transmssons must be suspended when the natve part arrves and deves n PAN enter a low-power mode for redung power onsumpton Data Transfer Types Dret data transmsson apples to all data transmssons ether from a deve to a oordnator, from a oordnator to a deve, or between two peers. In ths transfer type, data frames ontend hannel by the CSMA-CA algorthm. Indret data transmsson only apples to data transmssons from a oordnator to deves. In ths transfer type, all data frames sent by the oordnator wll be kept n the transaton lst and the oordnator notfes orrespondng deves to extrat pendng data by beaon frame. GTS data transmsson apples to data transmsson between a deve and ts oordnator regardless of transmsson dretons. It s desgned for low-lateny applatons or applatons requrng spef data bandwdth. In ths data transfer type, the oordnator may dedate portons of the atve superframe to some speal applatons. An adaptve ontenton ontrol strategy based on a two-stage approah has been ntrodued to mprove performane n ndret transfer [6]. In ths paper, a varous-parameter CSMA-CA algorthm based traff load s aheved n dret data transfer and a performane analyss s gven n detals. 28 Copyrght 204 SERSC

3 Internatonal Journal of Future Generaton Communaton and Networkng Vol.7, No.4 (204) 3. Adaptve Contenton Strategy 3.. IEEE CSMA-CA Mehansm In the beaon-enabled mode, the slotted CSMA/CA s used when deves aess hannel. Three ounters, BE, CW and NB are mantaned for hannel aess ontrol. BE defnes bakoff exponent used by deves to draw the random bakoff value. CW defnes the number of bakoff slots that the hannel needs to be sensed dle ontnuously before a transmsson. NB defnes the number of bakoffs that the CSMA-CA algorthm has attempted before aessng to the hannel. Then the slotted-csma algorthm works as follows. All assoated deves synhronze wth the pan oordnator. Before a new transmsson, BE, CW and NB are ntalzed to mamnbe, 2 and 0. Deves wth data frames to send frst baks off for a random number of slots wthn (2 BE -). Then after the random tme, the deve exeutes CCA, whh s short for lear hannel assessment. If two CCAs notfy that the hannel s dle, the deve wll send the data frame. If ether of CCAS notfes that the hannel s busy, BE and NB are nreased by one and CW s reset to 2.The above proedure s repeated untl the deve transmts the frame or NB arrves at mamaxcsmabakoffs Motvaton Sne IEEE s orgnally desgned for low-rate wreless personal area network, the system behaves well on low traff load and suffers a serous delne on heavy traff load. Based on the CSMA-CA algorthm desrbed n the prevous seton, an amount of ontenton overheads result from teratve bakoff operatons when the system traff load s heavy. It an be llustrated as follow. The bakoff wndow wll be reset to the mnmum value before a new transmsson. When network load s heavy, the serous ontenton ould not be resolved wthn a narrow bakoff wndow, whh leads to more ollsons. To mprove system performane under varous network loads, an adaptve CSMA-CA algorthm s needed. It s applable for wreless networks n whh traff loads do not hange fast. Through prevous experments and theoretal analyss, the values of MAC parameters have an mpat on the performane of the network [2]. When frames arrve at a moderate rate, deves wth a smaller bakoff wndow transmt the frames more qukly and have a hgher throughput. On the other hand, when frames arrve at a hgher rate, deves wth a larger bakoff wndow behave better for redung ollsons and mprovng transmsson effeny. Therefore, we make the parameter, ntal bak off wndow, hanges wth traff loads for a better performane Adaptve CSMA-CA Algorthm Flow Aordng to the standard CSMA-CA algorthm, the mean of parameter BE n suessful data delveres n a ertan nterval an reflet urrent traff load. Deves have a hgher probablty to aess the hannel wth a small bakoff wndow when network load s lght. As traff load beomes heaver, deves have a hgher probablty to try more tmes for hannel aess whh leads a large bakoff wndow. So we hoose the mean of BE n suessful transmssons durng a perod to predt the ntal bakoff wndow n the next perod. So the adaptve algorthm s mplemented n the followng flow, as shown n Fgure. The BE value for the suessful data delvery n the urrent superframe s reorded to predt the ntal value of the bakoff wndow for the next superframe. Then at the begnnng of the next superframe, the pan oordnator announes the ntal bakoff exponent value whh s Copyrght 204 SERSC 29

4 Internatonal Journal of Future Generaton Communaton and Networkng Vol.7, No.4 (204) defned as adaptvebe, to end deves va the beaon frame. When the next superframe starts, deves shall employ ths sutable BE mathng the urrent traff load n the CSMA-CA. In addton, when traff load beomes lghter, the adaptvebe should be dereased gently to avod a too wden bakoff wndow. New paket NB=0 CW=0 BE=max(MnBE,AverageBE) Watng for random [0,2 BE -] bakoff slots Perform CCA Channel dle? N Y CW=CW- AverageBE of system deves n a perod NB=NB+ CW=2 BE=mn(BE+,maMaxBE) CW=0? N N NB> mamaxcsma- Bakoffs? Y Reord BE n suessful transfers Y Dsard paket Y Reeve ACK? N 4. Performane Analyss Fgure. Adaptve CSMA-CA Algorthm Flowhart We now ntrodue analytal model. Some notatons used n ths part are lsted n Table. Table. Notatons n the Performane Analyss N Number of endng deves. L Data frame length. d ata L ACK frame length. ak The watng duraton between data frame and L d u rato n ACK frame. Paket arrval rate of an endng deve. G System output. The maxmum number of bakoff operatons m before delarng a hannel aess falure. The maxmum number of retres allowed after n a transmsson falure. In the analyss, we assume a beaon-enabled star topology wreless network wth N endng deves. The data frame length s assumed L bak-off slots and the arrval proess d a ta of frames follows a Posson dstrbuton wth a ertan arrval rate. In the followng part, we analyze G for the adaptve CSMA-CA. Here, we fous on the unsaturated ase n ACK mode. 30 Copyrght 204 SERSC

5 Internatonal Journal of Future Generaton Communaton and Networkng Vol.7, No.4 (204) Gven the aforementoned assumpton, the behavor of the system an be analyzed by Markov han models of a sngle deve and the hannel. In the followng part, the behavor of a sngle deve s analyzed usng a 3-D Markov han. Eah state s ndated by { r ( t ), s ( t ) j, b ( t ) k }, { b } or { b } T A C K. At a gven bakoff slot t, the stohast proess r( t ) represents the tmes of retransmsson for not reevng ACK frame, s( t) represents the transmsson stage and b( t ) represents the bakoff stage. Then the steady state probablty of beng n eah state s lsted as follow b lm P { r ( t ), s t j, b t k }, j, k t b lm P { T X T X } T r ( t ) t b lm P { A C K A C K } AC K r ( t ) t b lm P { r d le t, s t, b t } t () In ths model, p s the probablty that the hannel s dle and p s the probablty that the hannel s dle n next bakoff slot gven that hannel s dle n the urrent bakoff slot. By nspetng Fgure 2, eah steady-state probablty an be related by the followng Eqs. (2)-(6). n n n - (2) b ( - p ) b ( - p ) b ( - p ) b q b b d le d le, m, 0, m, - AC K AC K n ( q ) b W b n, j A C K, j, b [( ) ( ) ],, 2, j, 0 p b p b W b n j m, j, 0, j, j, j, p b W b,, j d le,, ( q ) b W b n, j AC K, j, k b [( ) ( ) ],, 2, j, k p b p b W b n j m, j, 0, j, j, j, k p b W b, j d le,, k (3) (4) p b, j,, j, 0 (5) b m AC K T (6) j, j, b b p b In Eq. (4), we gve the expresson of steady state b when k W, j, k. The expresson form of b when k W, j, k only nlude the frst part of Eq. (4). Copyrght 204 SERSC 3

6 Internatonal Journal of Future Generaton Communaton and Networkng Vol.7, No.4 (204) -p -,-,- p p,,- p /W,,0 /W,,,,(W -2),,(W -) TX -p p,2,- p -p /W 2 /W 2,2,0,2,,2,(W 2-2),2,(W 2 -) -p -p q ACK /W m -q p,m,- p /W m,m,0,m,,m,(w m -2),m,(W m -) -p -p -q /W /W p N,,- p n,,0 n,, n,,(w -2) n,,(w -) -p -p TX n /W m /W m ACK n p -p n,m,- p -p n,m,0 n,m, n,m,(w m -2) n,m,(w m -) Fgure 2. Markov Model of One Sngle Node for Slotted CSMA/CA We use the above assumpton to alulate some metrs whh determne the deve model. Aordng to the adaptve algorthm, the ntal bakoff wndow s dfferent from standard IEEE It s the same as the standard at the begnnng but hanges wth system traff load n the followng perods, and then we alulate the bakoff wndow n next perod based on traff load n the urrent perod by n w y W q p p b k, j, k j k m (7) W j j m 2 W j lo g ( 2 W ) 2 m m 2 lo g ( 2 W ) j m 2 (8) The probablty of a frame arrvng at the deve n a ertan bak off slot s p L. d a ta The probablty of a frame beng sent has a ollson wth other frames an be derved as N q p and the probablty p t t hannel s dle for two bakoff slots s equal to of a deve begnnng a transmsson gven that the n m n m w m n n p p p b p ( b b L b ( L L ) b ) t (9) t, j, d le, j, k d a ta T d u ra to n a k A C K j j k 32 Copyrght 204 SERSC

7 Internatonal Journal of Future Generaton Communaton and Networkng Vol.7, No.4 (204) For t s easy to get p whh s the probablty that the hannel beome dle n the next b bakoff slot gven that t s busy n the urrent bak off slot, followng form. p 2 ( L L ) b d ata ak p an be smplfed as the (0) p [ ( )] 2 ( ) p p p p p p ( L L ) () b To numerally obtan p, we take nto aount the jont dstrbuton of the N deves whh dedes the hannel state. In the prevous work [2], the hannel modeled by a dsrete-tme Markov han n Fg. 3 has been proposed and we use t for our analyss. d ata ak suess duraton dle,dle dle ak falure Let Fgure 3. Markov Model of the Channel b be the steady probablty that the hannel s dle. The hannel wll be d le,d le remanng n ths state wth the probablty f no deves begn a transmsson at the next bakoff slot. If only one deve begns transmsson, the hannel wll move to b wth su ess the probablty. If more than one deve begns transmsson at the same tme, the hannel wll move to b wth the probablty. Then,, are alulated by Eqs. (2-4), fa lu re and eah steady state of the hannel an be expressed by,. ( p ) N (2) t ( ) N N p p (3) t t ( ) N N p N p ( p ) (4) t t t The probablty p of the hannel remans dle for two onseutve bakoff slots s derved as the Eq. (5). p b ( b L b L b L b L b b ) d le,d le d le,d le d ata su ess d u rato n d u r ato n ak ak d ata falu re d le [( L )( ) ( L L ) ] d ata ak d u rato n p p p { ( L L ) [( L )( ) ( L L ) ]} ( L L 2 ) (6) d ata ak d ata ak d u rato n d ata ak (5) Copyrght 204 SERSC 33

8 Internatonal Journal of Future Generaton Communaton and Networkng Vol.7, No.4 (204) Aordng to Eqs. (), (5), and (6), the onstrants n the followng Eqs. (7, 8), the value of p an be alulated. d le,d le d ata su ess d u rato n d u rato n ak a k d ata falu re d le b L b L b L b L b b (7) As a result, metrs q, n m w y n n (8) b b L b L b d le, j, k d a ta T a k AC K j k p,, and eah state of two models are all avalable numerally. Base on the analyss above, the system throughput whh s defned as the rato of data transmsson perods to all tme that a transmsson proedure spends an be expressed as the followng equaton. G L b ( b L b L b L b L b b ) (9) 5. Results d ata su ess d le,d le d ata su ess d u rato n d u rato n ak ak d ata falu re d le 5.. Expermental Setup To measure the nfluene of traff load on the adaptve CSMA-CA, we evaluate the performane of adaptve CSMA-CA n NS2 smulator wth the followng metrs. Throughput: The rato of data transmsson perods to all tme that a transmsson spends. Paket delvery rate (PDR): The rato of pakets suessfully reeved to pakets sent n MAC layer. Ths metr does not dfferentate transmssons and retransmssons. Collson rate: The rato of pakets dropped by ollsons to total pakets sent n MAC layer. End-to-end MAC delay: The average transaton tme of passng a paket from the soure deve to the destnaton deve, nludng bakoff tme used n CSMA-CA and transmsson over all suessful end-to-end transmssons. General parameters are set as follow. A star topology wth a radus of 0 meters s used and there are sx deves n t. Transmsson range and arrer sense range an be set dfferently wthn a ertan range (5m-30m) for testng the effets of hdden-termnals. PHY operates at an over ar data rate of 250 kbps. The rado propagaton model adopted s two-ray ground refleton. In the smulaton, Posson traff s used and data frame length s assumed a fxed value of 90 bytes. Here BO and SO are both set 3 and the optonal natve part s not nluded n superframes. Traff Load hanges wth arrval rate of the Posson stream. Eah ndependent smulaton lasts 00 seonds and runs repeatedly wth dfferent random seeds Performane Analyss n Sene wthout HT Fgure 4 shows the effets of the traff load on the throughput, paket delvery rato, ollson rate and average MAC delay nludng HT respetvely. In Fgure 4(a), the throughput n Y-axs s normalzed to the physal-layer transmsson rate. Ths fgures show that the adaptve CSMA-CA algorthm almost has the same performane as IEEE standard when traff load s lght. However, t behaves muh better as traff load beomes heaver. Fgure 4(b) gves out ollson rate omparson. We fnd that the modfed algorthm has an obvous mprovement on ollson avodane. Fewer ollsons also ontrbute to better throughput. But ths urve s not qut smooth. Ths s beause deves have a double sze of the ontenton wndow va adaptvebe parameter when most of data frames suffer ollsons. Due to the gradent extenson of the ontenton wndow, deves have obvously fewer hanes of seletng the same bakoff slot sendng data n a qute short term. Thus, the adaptve CSMA-CA performane more effetvely as traff load beomes heaver. 34 Copyrght 204 SERSC

9 Paket Delvery Rato MAC delay(ms) Throughput Collson Rate Internatonal Journal of Future Generaton Communaton and Networkng Vol.7, No.4 (204) Fgure 4() and Fgure 4(d) respetvely shows paket delvery rato and MAC delay. Paket delvery rato dereases heaver as the traff load nreases when system deves adopt the standard CSMA-CA. Even the adaptve algorthm nreases ontenton wndow, t redues the average delay effently under a heaver traff load by avodng plenty of retransmssons IEEE std wthout HT IEEE wth adaptve CSMA wthout HT 0.6 IEEE std wthout HT IEEE wth adaptve CSMA wthout HT Traff Load(pkt/se) 0.95 (a) Traff Load(pkt/se) (b) IEEE std wthout HT IEEE wth adaptve CSMA wthout HT IEEE std wthout HT IEEE wth adaptve CSMA wthout HT Traff Load(pkt/se) () Traff Load(pkt/se) Fgure 4. System Performane of Adaptve CSMA-CA: (a) Throughput, (b) Collson Rate, () Paket Delvery Rate, (d) MAC Delay 5.3. Performane Analyss n Sene wth HT The hdden termnal problem (HTP) s ommon n the wreless ommunaton senaro usng CSMA sheme n MAC layer. It s losely related to the Transmsson Range (TX Range), Carrer Sensng Range (CS Range) and Interferene Range (IF Range) of deves n wreless networks. The TX Range s the range wthn whh a transmtted frame an be suessfully reeved. The transmsson ranges vary wth dfferent data rates, beause dfferent SNR requrements are desgned for varous data rates n PHY layer. The CS Range ndates the range wthn whh deves an detet a transmsson. The IF Range ndates the range wthn whh deves n reevng mode an be nterfered by a hdden transmtter. Hdden-termnals refer to the nodes wthn the IF range of the ntended destnaton and out of the CS of the soure. As Fgure 5 shows, deve C and D are out of the CS range of deve B but n the TX range of deve A. The gray area ndates the hdden area. For C and D are hdden to B, they may aess the hannel at the same tme and lead to a data ollson. RTS/CTS mehansm s used n IEEE 802. to redue HTP. The desgn of ths mehansm s (d) Copyrght 204 SERSC 35

10 Collson Rate Paket Delvery Rato Internatonal Journal of Future Generaton Communaton and Networkng Vol.7, No.4 (204) usng RTS/CTS frames to tell neghbor nodes the hannel wll be ouped for a ertan perod. It effetvely redues ollsons between hdden deves n But the RTS/CTS mehansm s not adopted by IEEE CSMA/CA. D A TX Range CS Range IF Range B C CS Range Fgure 5. Hdden-termnals In ths part, we evaluate performane of the adaptve CSMA-CA respetvely nludng hdden nodes. Fgure 6 shows the effets of adaptve CSMA-CA algorthm on aspet of ollson rate, paket delvery rato. The range of traff load n X-axs s set n a farly small range for t performane delnes too fast as traff load vares IEEE std addng HT IEEE wth adaptve CSMA addng HT IEEE std wthout HT IEEE wth adaptve CSMA wthout HT Traff Load(pkt/se) (a) 0.5 IEEE std addng HT IEEE wth adaptve CSMA addng HT IEEE std wthout HT IEEE wth adaptve CSMA wthout HT Traff Load(pkt/se) Fgure 6. System Performane of Adaptve CSMA-CA n Senes Inludng HT: (a) Collson Rate, (b) Paket Delvery Rate The adaptve algorthm has a lttle better performane ompared wth the standard n Fgure 6. However we an fnd that the adaptve CSMA-CA algorthm annot mprove bad effets effently. Collsons n senaros wth HT an be attrbuted to two aspets. Frst, deves send data frame no matter whether the hdden nodes are transmttng data and CSMA-CA s neffetve for ths ollsons. Seond, deves send data wth other deves sgnfantly n the sensng range when t detets the hannel dle for the shortage CSMA-CA. Aordng to omparson wth the performane wthout HT n the same traff load, we fnd that ollsons aused by the frst aspet are more serous than ollsons by the seond aspet. (b) 36 Copyrght 204 SERSC

11 Internatonal Journal of Future Generaton Communaton and Networkng Vol.7, No.4 (204) The adaptve CSMA-CA algorthm mproves system performanes by untargeted expendng ontenton wndow to redue ollsons. But t fals to dstngush effets aused by traff load and HT. When data exhanges between HTs are heavy, the adaptve algorthm annot work effetvely. 6. Conluson In ths paper, we aheve dynam-parameter CSMA-CA algorthm based traff load estmaton and analyzes ts performane n dret transmsson. It behaves well n senaros exludng HT whle mproves performanes lttle nludng HT. For a better property, feedbak mehansm amng at HT under dfferent traff load needs to be further studed. Aknowledgements Ths work n part was supported by Sene and Tehnology Development Projet of Shandong under Projet No.20GSF2007. Addtonal support s provded by a grant from Independent Innovaton Foundaton of Shandong Unversty. The authors would express great grattude to Dr. Zunhua Guo and Dr. Chengyou Wang for ther help n the aadem assstant. Speal thanks also go to revewers for ther useful reommendatons. Referenes [] G. Lu, B. Krshnamahar and C. S. Raghavendra, "Performane evaluaton of the IEEE MAC for low-rate low-power wreless networks", Proeedngs of 23rd IEEE Internatonal Performane, Computng, and Communatons Conferene, (2004) Aprl 5-7, Phoenx, Unted states. [2] S. Wjetunge, U. Gunawardana and R. Lyanapathrana, "Impat of MAC parameters on the performane of IEEE MAC protool wth ACK frame transmsson", Proeedngs of Australasan Teleommunaton Networks and Applatons Conferene, (20) Otober 3-November 3, Aukland, New Zealand. [3] A. Fard, M. R. Palattella, A. Lozano, M. Dohler, G. Bogga, L. A. Greo and P. Camarda, "Comprehensve evaluaton of the IEEE MAC layer performane wth retransmssons", IEEE Transatons on Vehular Tehnology, vol. 59, no. 8, (200), pp [4] T. J. Lee, H. R. Lee and M. Y. Chung, "MAC throughput lmt analyss of slotted CSMA/CA n IEEE WPAN", IEEE Communatons Letters, vol. 0, no. 7, (2006), pp [5] J. Ms, S. Shaf and V. B. Ms, "Avodng the bottleneks n the MAC layer n low rate WPAN", Proeedngs of th Internatonal Conferene on Parallel and Dstrbuted Systems Workshops, (2005) July 20-22, Fukuoka, Japan. [6] H. W. Tseng, A. C. Pang, J. Chen and C. F. Kuo, "An adaptve ontenton ontrol strategy for IEEE based wreless sensor networks", IEEE Transatons on Vehular Tehnology, vol 58, no. 9, (2009), pp [7] "IEEE Std ", IEEE Standard for Teleommunatons and Informaton Exhange Between Systems - LAN/MWAN - Spef Requrments - Part 5: Wreless Medum Aess Control (MAC) and Physal Layer (PHY) Spefatons for Low Rate Wreless Personal Area Networks (WPANs), New York, Unted states, IEEE Standard, (2003). Authors Ln Zhu, she reeved Bahelor degree from Shandong Unversty, Weha, Chna n 20. Now she s studyng at Shool of Mehanal, Informaton Engneerng n Shandong Unversty, Weha, Chna. Her researh nterests fous on wreless sensor networks. Copyrght 204 SERSC 37

12 Internatonal Journal of Future Generaton Communaton and Networkng Vol.7, No.4 (204) Guangmng L, he reeved Bahelor degree and Master degree from Shandong Unversty n 996 and 999 respetvely. He was engaged n researh about fber opt sensors n Texas A&M Unversty from 2000 to Then he enrolled at The Unversty of Hong Kong and earned Ph.D. degree n Now he s workng n Shandong Unversty as an assoate professor and has publshed over 30 papers n journals and nternatonal onferenes about Optal networks and WSNs. We Lu, he reeved Bahelor degree from Shandong Unversty, Weha, Chna n 20. Now he s studyng at Shool of Mehanal, Eletral and Informaton Engneerng n Shandong Unversty, Weha, Chna. Hs researh nterests fous on wreless sensor networks. 38 Copyrght 204 SERSC