A Secure Localization Method in Wireless Sensor Network

Transcription

1 Vol. 1(3), Ot. 2015, pp A Seure Loliztion Metho in Wireless Sensor Network Mrym Mirzei, Mohmm Rotmili n Mehi Khlili Dept. of Computer n Informtis Engineering, Pym Noor University, Tehrn, Irn *Corresponing Author's E-mil: mirzei_2011@yhoo.om Astrt W ireless sensor network is ompose of hunres or thousns of sensor noes tht ommunite n work together to perform speifi tsk or tsks. Informtion sense y eh noe sent to se sttion, whih is lle sink n onnete to the AC power. In mny ses, the lotion of sense t is importnt for eision-mking. Loliztion is use for this purpose. Loliztion is one of the tehniques use in wireless sensor networks. Rel pplitions of wireless sensor networks often re fe with vriety of hrmful interferenes tht hve signifint impt on the effiieny of loting. In this pper, seure n roust loliztion lgorithm for wireless sensor networks to reue the impt of hostility ttkers s externl ttks n fult n error in networks elements s internl ttks is presente. The propose metho onsists of two steps whih in the first step, mliious nhor noes re etete n noes trust vlues re lulte. In the next step, Tylor series lest squre metho is use to estimte the oorintes of the sensor noes. Simultion results will show tht the propose lgorithm is effiient n roust. Keywors: wireless sensor network, loliztion, mliious nhor noe, lest squre I. Introution Wireless sensor networks (WSNs) re se on sensor tehnology, wireless ommunitions, tiny emee evies, n istriute omputing. These networks re exhnging t with the environment y sensors n perform t mnipulting n gthering. WSNs re wiely use in environment monitoring, trget trking, militry pplition, isster mngement n et [1-3]. The noe loliztion tehnology is neee in WSN pplition, espeilly when lotion informtion is neessry. WSN noes loliztion is etermining oorintes of norml noes se on nhor noes oorintes n onfine reltion etween nhor n norml noes. The oorinte or lotion of norml noe is unknown. On the other hn, Anhor noes n otin their lotion vi glol positioning system (GPS) moules or mnully. In mny typil loliztion lgorithms [4-6] ssume tht nhor noes lotion informtion re quite properly without ny interferene y verse ftors or internl ttks n norml noes n use nhor informtion sfely. However, in rel hostile situtions, some mliious noes my enter into the sensor network without uthoriztion in orer to sotge. They re trying to introue themselves s enign nhor noes or ttk on other nhor noes in orer to fore them to elre wrong lotion [7]. In ition to mliious nhor noes, fulty nhors shoul e etete for voiing errors in loliztion. Erroneous istne estimtion or erroneous oorinte uses irreplele fult in norml sensor noes loliztion. In this se, some methos shoul e pplie to eliminte or reue hrness effets rete y mliious or fulty nhor noes n ensure seure wireless sensor network loliztion. We ll mliious nhor noes the oth fulty n mliious nhor noes. In this pper, we propose roust n seure loliztion lgorithm in orer to solve the prolem of mliious nhor noes existene in loliztion. In the propose metho, eh nhor noe sks other nhors their lotions n eision out other nhor noes tht re enign or mliious y tringultion loliztion metho. The sensor noe is informe out nhor noe eision in orer Artile History: JKBEI DOI: /11031 Reeive Dte: 11 Jul Aepte Dte: 20 Ot Aville Online: 05 Nov

2 to lolize itself. The propose metho will e isusse riefly lter. The reminer of the pper is orgnize s follows: Setion 2 introues relte works on seure loliztion lgorithms. Setion 3 presents the network moel, ttk moel, n relte efinitions. Setion 4 provies the etils of propose metho. Setion 5 presents the simultion results. Setion 6 onlues the pper. II. Relte Works A Alfro et l. [10] onsier the loliztion seurity of sensor noes uner limite trust nhor noes. It proposes three lgorithms to enle the sensor noes to etermine their positions, ut it woul fil when the mliious nhor noes re in olluing onitions. Colluing onition is se where the mliious nhor noes n etet whether other nhor noes re the sme type, n eh pir of olluing mliious nhor noes n revise the mesure istne etween them y hnging their elre lotions. Liu et l. [11] introue two seure loliztion lgorithms. One is ttk-resistnt minimum men squre estimtion, whih exlues mliious nhor noes y the onsisteny hek. The other is voting-se lotion estimtion. The lgorithms re iffiult to work for the mliious nhor noes in olluing onitions. Zhu et l. [12] present n ttk etetion moule whih n etet ompromise eons n provie loliztion servie in terms of oune estimtion error y seure loliztion moule, ut it minly onentrtes on the one-hop loliztion. Liu et l. [13] propose seure loliztion mehnism tht etets mliious nhor noes liming fke positions. It uses reunnt nhor noes inste of norml noes in the sensing fiel to verify mliious nhors. The metho relies on entrlize se sttion for the etetion. Li et l. [14] present seure sheme Biltertion whih is erive from multiltertion. It lultes the weight of nhor noes n eies whih nhor noes re mliious. After ignoring the oorintes use y ompromise noes, it uses the verge vlue of the let nite positions s the estimte lotion of the sensor noe, ut it minly fouses on the one-hop loliztion. Yu et l. [15] propose seure shem lle BRSL n try to etet n thwrt mliious nhor noes. In this pper et reputtion system is use in orer to etermine nhor noes trust vlues. If trust vlue of n nhor is less tht lulte threshol, this nhor etets s mliious. The Tylor series lest squre metho is use in orer to loliztion. The wekness of this metho is tht it relies on severl noes eployment reputtion whih my e impossile in mny situtions. III. PRIMARY DEFINITIONS All Network Moel: The network with two noes type (nhor n sensor) is onsiere. The nhor noes re equippe with speil equipments n know their positions. The sensor noes whih their positions shoul e etete, estimte their lotions y mesuring istnes to neighoring nhor noes y RSSI. All the noes re istriute rnomly in 2 omintion environment. The ommunition rnge is enote s R in eh noe n they n lulte rnges of their one-hop neighors. The omin error e follows Gussin istriution (N(u,λ2)) tht men u is zero n vrine λ is restrite: ei <=emx (1) The mximum physil error emx is otine experimentlly. In multihop loliztion, eh nhor noe rosts messge tht rries its elre position to its one-hop neighors. Then, the messge is propgte in the network in ontrolle flooing mnner. When sensor noe otins three or more nhor messges, the sensor noe n estimte its lotion y the loliztion lgorithm [8]. Attk moel: Assume tht WSN is in hostile environment whih mens there re mliious ttkers. The ttkers ttk nhor noes in orer to fore them to elre flse positions. When 213

3 n nhor noe is ttke n elres flse position is lle mliious nhor. The nhors tht elre rel positions re lle enign nhor noes. When sensor noe M gets enough mesurement istnes mi (i=1,2,,k), where k 3,to nhor noes Ai, system of the Eulien equtions n e set up: Xm X1 2=m1 Xm X2 2=m2 Xm Xk 2=mk Where Xm=[xm,ym]T is M s oorintes tht nee to e estimte n Xi=[xi,yi]T is nhor noe Ai s elre position. If the nhor noe A1 is ttke, it will eome mliious nhor noe A 1 with fke oorintes. When M utilizes A 1 to ompute its position, its estimte position M will evite fr from its physil position, n its lotion ury will e very low. IV. PROPOSED METHOD A) Mliious Anhor Noes Detetion In orer to etet mliious nhor noes, initilly, ll nhor noes rost their lotion informtion mong the network. Eh nhor noes estimte its lotion with other three pre nhor noes lotion informtion y tringultion metho. The nhor noe estimtes its lotion oring to ifferent three prs. The otine lotions re ompre with rel lotion. If ifferene etween otine lotion n rellotion eome more thn emx, t lest one of these three involve nhor noes re mliious. By evluting other otine lotions, the mliious nhor noes etets esily. We onsier trust vlue for eh nhor noes to voi involving fulty or untruste nhor noes in loliztion proess. If eferene etween estimte lotion y three nhor noes n rellotion is more thn pre-efine threshol, the trust vlue of these three nhors erese. On the other hn, if the eferene etween estimte lotion y three nhor noes n rellotion is less thn pre-efine threshol, the trust vlue of these three nhors inrese. Fig.1. Tringultion metho In tringultion metho lotion of noe is otine from the following formul y three known position = x 2x x + x + y 2y y + y = x 2x x + x + y 2y y + y = x 2x x + x + y 2y y + y (2) 214

4 ( ) ( ) ( ) ( ) = 2x x x + x x + 2y y y + y y = x x x x x 2y y y y y ( ) ( ) ( ) x x y y v = ( ) ( ) ( ) x x y y v = 2 v ( x x ) v ( x x ) ( y y ) ( x x ) ( y y ) ( x x ) v y ( y y ) x = ( x x ) y = ) (3) (4) (5) B) Loliztion Proess X = 1 k i 1 X K i. Firstly, lulte the entroi oorintes X =(x,y ) of k nhor noes, tht is, ( ) = Seonly, expn the funtion f( x) X X i 2 Therefore, ΔX ( ) C Δx Δy = in Tylor series t X, n ignore the high-orer terms. = n e otine suh s: ( ) 1 T T T T ΔX C = A W WA A W WB (6) Where: Tru_T1 0 Tru_T2 W = , 0 Tru_T K 1 2 B =... k 1C 2C KC, xc x A = 1C yc y 1C 1 1 xc x 2C yc y 2C xc x KC yc y KC K K T. Thirly, is lulte y formul 7, n juge whether the itertion termintion onition <=ᶯ is stisfie, where ᶯ is prior-efine threshol. If <=ᶯ, we stop the itertion proess. Otherwise, we set X =X + X n go to the seon step. Finlly, repet the seon step n the thir step until the itertion termintion onition is stisfie or the mximum itertion numer is rehe. The finl output X is the estimte oorintes of sensor noe M. The loliztion lgorithm is in elow. 215

5 2 2 = Δx C + Δy C (7) Algorithm. Noes loliztion lgorithm for i=1:1:length(engingnoes) W(i,i)=engingnoes(i).Tru_TA; % lultion of W en for i=1:1:noes+mnoes % lultion of XC sumx=0; sumy=0; for k=1:1:length(engingnoes) sumx=engingnoes(k).x+sumx; sumy=engingnoes(k).y+sumy; en X=sumx/length(engingnoes); Y=sumy/length(engingnoes); XC.x=X; XC.y=Y; steps=0; % prevent of unlimite while loop rep=1; % repet lines etween while until rep equl to 0 while (rep~=0) steps=steps+1; for k=1:1:length(engingnoes) (i,k)=istne1(s(i),engingnoes(k)); (k)=istne1(engingnoes(k),xc); en for k=1:1:length(engingnoes) B(k)=(i,k)-(k); en for k=1:1:length(engingnoes) A(1,k)=(XC.x-engingnoes(k).x)/(k); A(2,k)=(XC.y-engingnoes(k).y)/(k); en A=A'; B=B'; eltxc=(a'*w'*w*b)'/(a'*w'*w*a); =sqrt(eltxc(1)^2+eltxc(1)^2); A=A'; B=B'; if >=n XC.x=eltXC(1)+XC.x; XC.y=eltXC(2)+XC.y; elseif <n steps>40 rep=0; % this line uses stop while en en news(i).x=xc.x; news(i).y=xc.y; en After the text eit hs een omplete, the pper is rey for the templte. Duplite the templte file y using the Sve As ommn, n use the nming onvention presrie y your onferene for the nme of your pper. In this newly rete file, highlight ll of the ontents n 216

6 import your prepre text file. You re now rey to style your pper; use the sroll own winow on the left of the MS Wor Formtting toolr. V. Simultion results To emonstrte the effetiveness of the propose metho in this setion, we o omprison etween this metho n some of the leing methos in the fiel of seure loliztion. We ompre the propose metho with BRSL n Biltertion tht re roust n effetive loliztion lgorithms. Mtl is the simultion tool use in this pper whih is pplie in mny sientifi ppers. Initilly network size is efine s two omintions squre 200*200. Numer of ll noes, enign nhor noes n mliious nhor noes re 200, 20 n 10 respetively. The noes istriution strtegy is rnom. Other network prmeters re liste in tle 2. Figure 2 illustrtes smple noes eployment where lue yles, green strs n re strs re emonstrte norml sensor noes, enign nhor noes n mliious nhor noes respetively. A) Simultion Prmeters TABLE I. INITIAL PATAMETERS Prmeter Vlue Initil Energy 0.5J Numer of ll Noes 200 Network Are Flt 2D Network Dimension 200*200 Sink Position Noes Deployment Control Pkets Emx Pre-efine Threshol (100,100) Rnom 12 it Fig. 2. Noes rnomly eployment 217

7 The mjor prmeter tht shoul e mesure in loliztion tehniques is mount of loliztion error in ifferent situtions. Initilly the stte is onsiere tht numer of mliious nhor noes inreses from 1 to 10. The simultion result of this sitution is shown in figure 3 whih the error rte of the propose metho is muh less thn the other two methos. Figure 4 emonstrtes results of omprisons etween errors with ifferent stnr evition. The error of eh lgorithm is inrese y stnr evition inrese. Aoring to these simultion results we n onlue tht in omprison with BRSL n Biltertion, the propose metho is more effiient. Fig. 3. Compre etween error rtes with ifferent numer of mliious nhor noes Fig. 4. Compre etween error rtes with ifferent stnr evition VI. Conlusion In wireless sensor network, for sening informtion, the lotion of sense t is importnt for eision-mking. Thus, loliztion eomes n importnt issue for this purpose. In this pper, to reue the impt of hostility ttkers s externl ttks n lso fult n error in networks elements s internl ttks in wireless sensor networks, we propose seure n roust loliztion lgorithm, in whih, first, mliious nhor noes etete noes trust vlues lulte. Then, to estimte the oorintes of the sensor noes, Tylor series lest squre metho use. The otine results of simultion inites tht the propose lgorithm not only stisfies effiieny n roustness, ut lso ereses error rtes in the stnr evition n the numer of mliious nhor noes, ompre to BRSL n iltertion. Moreover, the propose lgorithm le reserhers to lustering struture s the future work. 218

8 Referenes [1] J. Alowiz, A. Chen, n L. Zhng, Reursive position estimtion in sensor networks, in Proeeings of the 9th Interntionl Conferene on Network Protools, pp , [2] H.A. Oliveir, E.F. Nkmur, A.A.F. Loureiro, n A. Boukerhe, Direte position estimtion: reursive loliztion pproh for wireless sensor networks, in Proeeings of the 14th IEEE Interntionl Conferene on Computer Communitions n Networks, pp , Sn Diego, Clif, USA, [3] T. He, C. Hung, B.M. Blum, J. A. Stnkovi, n T. Aelzher, Rnge-free loliztion shemes for lrge sle sensor networks, in Proeeings of the 9th Annul Interntionl Conferene Moile Computing n Networking (MoiCom 03),pp.81 95, ACM Press, Sn Diego, Clif, USA, [4] M. L. Sihitiu n V. Rmuri, Loliztion of wireless sensor networks with moile eon, in Proeeings of the 1st IEEE Interntionl Conferene on Moile A-Ho n Sensor Systems, pp , Fort Luerle, Fl, USA, Otoer [5] X. L. Guo, R. J. Feng, Y. F. Wu, n J. W. Wn, Gri-Sn-Bse multi-hop Loliztion lgorithm for wireless sensor networks, in Proeeings of IEEE Sensors Conferene, pp , Wikolo, Hwii, USA, [6] D. Niulesu n B. Nth, DV se positioning in ho networks, Teleommunition Systems, Volume 22, pp ,2003. [7] K. Yu, Y. J. Guo, n M. Heley, TOA-se istriute lolistion with unknown internl elys n lok frequeny of sets in wireless sensor networks, IET Signl Proessing, Volume 3, No. 2, pp , [8] J. Wn, N. Yu, R. Feng, Y. Wu, n C. Su, Loliztion refinement for wireless sensor networks, Computer Communitions, Volume 32, pp , [9] N. Yu, L. Zhng, n Y. Ren, BRS-Bse Roust Seure Loliztion Algorithm for Wireless Sensor Networks, Interntionl Journl of Distriute Sensor Networks, Volume 2013, Artile ID , 9 pges. [10] J.W. Wn, X.L. Guo, N. Yu, Y.F. Wu, n R.J. Feng, Multi-hop loliztion lgorithm se on gri-snning for wireless sensor networks, Sensors, Volume 11, No. 4, pp , [11] Y. Shng, H. Shi, n A.A. Ahme, Performne stuy of loliztion methos for -ho sensor networks, in Proeeings of the IEEE Interntionl Conferene on Moile A-Ho n Sensor Systems, pp , Otoer [12] S.Y. Wong, J.G. Lim, S.V. Ro, n W.K.G. Seh, Multihop loliztion with ensity n pth length wreness in nonuniform wireless sensor networks, in Proeeings of the IEEE 61st Vehiulr Tehnology Conferene, Volume 4, pp , Stokholm, Sween, June [13] Q.J. Xio, B. Xio, J.N. Co, n J.P. Wng, Multihop rnge free loliztion in nisotropi wireless sensor networks: pttern-riven sheme, IEEE Trnstions On Moile Computing, Volume 9, No. 11, pp , [14] J. Hwng, T. He, n Y. Kim, Deteting phntom noes in wireless sensor networks, in Proeeings of the 26th IEEE Interntionl Conferene on Computer Communitions (INFOCOM 07), pp , My [15] R.J. Feng, X.L. Guo, N. Yu, n J.W. Wn, Roust multihop loliztion for wireless sensor networks with unrelile eons, Interntionl Journl of Distriute Sensor Networks, Volume 2012, ArtileID ,13 pges,