An Effcent Recover Algorthm for Coverage Hole n WSNs Song Ja 1,*, Wang Balng 1, Peng Xuan 1 School of Informaton an Electrcal Engneerng Harbn Insttute of Technolog at Weha, Shanong, Chna Automatc Test an Control Insttute Harbn Insttute of Technolog,Harbn, Chna * songja@htwh.eu.cn Abstract. Falures are nevtable n wreless sensor networks, an t s mportant to etect an patch the coverage hole. In ths paper, we propose an effcent recover algorthm base on mnmum stance reunant noes. B numercal algorthm, the snk noe gets the centre of the coverage hole, an then wakes up a reunant noe, whch has the mnmum stance to the centre, to recover the hole. Smulaton results emonstrate that, b choosng approprate number of reunant noes, ths algorthm wll have great recover accurac an coverage qualt, also acheve the purpose of prolongng the lfeccle of WSNs. Kewors: coverage hole; fault management; WSNs 1 Introucton Falures are nevtable n wreless sensor networks (WSNs) ue to the nhosptable envronment an unattene eploment. Furthermore, the constrane energ, storage capact an computatonal ablt make the stuaton worse. In practcal applcaton, fault sensor noes are one of the most mportant sources of faults. A noe s fragle an has lmte resources, whch makes t eas to become fault. So there appears some coverage hole, whch wll en the lfe of WSNs an leave much more energ resources waste n the networks [1] []. Noe Self Detecton b Hstor ata an Neghbors (NDHN) algorthm [3] an Fault Detecton Technque base on Clusterng (FDTC) [4] are propose to etect the fault noes. In ths paper, we target the problem of recoverng the coverage hole. Relate Work There are man reunant noes eploe ranoml n WSNs, so t s possble to solate fale noes rectl an choose some reunant noes to restructure a new Supporte b the Natonal Scence Nature Founaton of Chna uner Grant No 611706 ASP 013, ASTL Vol. 18, pp. 5-9, 013 SERSC 013 5
Proceengs, The n Internatonal Conference on Avance Sgnal Processng route to transmt the nformaton. Man exstng approaches use ths technque to hanle smple fault stuaton. Mart et al. [5] propose a technque to etect fault neghbor an choose another new neghbor nstea. Usng moble noe to repar coverage hole s another most effectve technques n fault recover, whch can solve har fault. In the paper [6], Gulng Wang propose an algorthm that the snk noe etermne reunant noe s locaton an controlle t to move to the proper place to repar the hole. 3 Localze Fault Recover Algorthms 3.1 Sesson Scenaro The algorthms n ths paper are evelope uner the followng assumptons: 1) Actve noes are structure an eploe n avance; reunant noes are eploe ranoml; ) All sensor noes are statonar after eploment, wth sensng range R s ; 3) All sensor noes have the same structure an functon, an ther status s equal n an wa; 4) There s onl one snk noe, an t s eploe outse the network, whch has unrestrcte energ; 5) Ever noe has ts locaton nformaton. Rs N3 Nr3 Nr4 N4 X N1 N Lo Lo Nr1 N5 N6 Nr (1) () (3) (4) Fg. 1. Sesson scenaro Fg.1 llustrates the man role urng a recover sesson, whch starts when the snk noe etects a eath noe N1 through NDHN technque [3]. The coverage hole cause b N1 s showe b the shaow X (1). Accorng to the locaton nformaton of N1 s neghbors, such as N N3 N4 N5 an N6, we can obtan the rregular shape of X. Then seek a foursquare to cover the whole area of X, whch the square area shoul be mnmum (). After that, t s mportant to seek a mnmum crcle to cover the foursquare, an the centre of the crcle wll be the optmum locaton of a substtute noe, marke Lo (3). Furthermore, the ajacent reunant noes calculate the stances to Lo, an transform the stances to the snk noe. Fnall, accorng to the prncple of mnmum stance, snk noe chooses reunant noes Nr1 as the substtute noe, an wake t up to jon the network (4). 6
An Effcent Recover Algorthm for Coverage Hole n WSNs 3. Dervaton of Recover Proceure The proceure of recoverng a coverage hole can be explane as follows. Denote R N ) as a boune close set of n m that contans actve noes ( 1, N Nn, an reunant noes N r, Nr,... Nrm N,... as N ( x, )( 1, n) ( x l, l L n R N ) spot ) = an N ( x, m) ( are compute b equaton (1). r r, an enote r r 1. Denote ther coornates )( = 1,. Denote an arbtrar r as the stance from L to N, whch = ( xl x ) + ( l ) (1) We also enote L x, ) as the optmal locaton of recover noe. An enote o( lo lo as the stance from 3.3 Algorthm N to L, whch are compute b equaton (). r o = ( xlo xr) + ( lo r) () 1) NDHN algorthm wll be operate n the cluster, an then the snk noe makes a ecson whether there has a fault noe N ; ) Calculatng the stance from an arbtrar spot L to N j, neghbor noe of N, b equaton (1). When ever r s satsfe wth r R s, t ncates that the arbtrar spot L s n the coverage hole ; 3) Searchng for the extreme x-coornates of all arbtrar spots n the coverage hole, marke x an x mn, an the extreme -coornates, marke an mn ; 4) Choosng the greater value of x t an square S h, an etermnng four enponts of t as the length of the se of one S h at the same tme, marke P x', ' ; P x', ' ), P x', ' ), P x', ' ) an ) 1( mn mn crcle of ( mn 3( 4( mn 5) Choosng a crcle to cover S h completel, an the centre of the mnmum Cmn wll be the optmal locaton of replacement noe, marke L o. The raus C mn an the coornate of L o wll be calculate; 6) Calculatng the value of b equaton (). Snk noe actves noe jon the network an sense parameters, whch has the mnmum value of orers the rest of the reunant noes to sleep. N r to, an 7
Proceengs, The n Internatonal Conference on Avance Sgnal Processng 4 Smulaton an Dscussons MATLAB s use to perform the smulatons. It contans 4 noes n a square regon of sze 140 100m, n whch actve noes are structure an eploe n avance an 10 reunant noes are eploe ranoml. We use the NDHN technque to etect the voltages of the noes. Once the voltages are lower than.v, noes n the sstem wll stop transmttng ata an be regare as fault practcall. Then MDRN technque s trggere to recover the coverage hole. Fg.. Recover accurac Fg.3. Qualt of recover Because the number of reunant noes wll affect the recover result serousl, we choose 10, 0, 30 an 50 reunant noes respectvel to recover 1 fault noe, an then to recover fault noes. In ever stuaton, we operate MDRN algorthm 30 tmes an calculate the average value. The recover accurac of MDRN s showe n Fg., an the qualt of recover s showe n Fg.3. 5 Concluson Ths paper presents a technque MDRN for noe fault recover n fault management of WSNs. Accorng to the smulaton result, t s feasble an effectve for treatng coverage hole cause b fault noe. Furthermore, the metho coul be serve as a support for sensor networks energ balancng an optmzaton. In the future, we wll explore the technque of reucng the number of reunant noes, an mprovng the recover accurac when treatng the stuaton of multple fault noes. References 1. Aklz, I.F.,Welan Su, Sankarasubramanam, Y., Carc, E.: A surve on sensor networks. J. Communcatons Magazne, IEEE, 10-114(00).. Peng Yu, Song Ja, Peng Xuan. Surve of Fault Management Framework n Wreless Sensor Networks[J]. Journal of Electronc Measurement an Instrument,1-10(009) 3. Peng Yu, Song Ja, Peng Xuan. A Self Detecton Technque n Fault Management n WSN. C. In:011 Internatonal nstrumentaton an measurement technolog conference (IMTC011), 1755-1758(011) 4. Song Ja, Peng Yu, Peng Xuan. Fault Detecton Technque base on Clusterng n WSN. J. Chnese Journal of Scentfc Instrument. 14-0(01) 8
An Effcent Recover Algorthm for Coverage Hole n WSNs 5. S. Mart et al., Mtgatng Routng Msbehavor n Moble A Hoc Networks. C. In:ACM MobCom 0, Boston, MA, 000 6. Gulng Wang, Guohong Cao, La Porta, T., Wensheng Zhang.Sensor relocaton n moble sensor networks. C. In: 4th Annual Jont Conference of the IEEE Computer an Communcatons Socetes. pp. 30-31. Proceengs IEEE. INFOCOM 005 9