Improved Correction Localization Algorithm Based on Dynamic Weighted Centroid for Wireless Sensor Networks

Size: px

Start display at page:

Download "Improved Correction Localization Algorithm Based on Dynamic Weighted Centroid for Wireless Sensor Networks"

Gloria Heath
5 years ago
Views:

1 Sesors & Trasducers, Vol. 76, Issue 8, August 204, pp Sesors & Trasducers 204 by IFSA Publishig, S. L. Improved Correctio Localizatio Algorithm Based o Dyamic Weighted Cetroid for Wireless Sesor Networks Xuejiao Che, 2 Pa Wag Departmet of Commuicatio, NaJig College of Iformatio Techology, NaJig, 20023, Chia 2 School of Software, NaJig Uiversity of Posts & Telecommuicatios, NaJig, 20023, Chia chexj@jcit.c, wagpa@jupt.edu.c Received: 2 July 204 /Accepted: 2 August 204 /Published: 3 August 204 Abstract: For wireless sesor etwork applicatios that require locatio iformatio for sesor odes, locatios of odes ca be estimated by a umber of localizatio algorithms. However, precise locatio iformatio may be uavailable due to the costrait i eergy, computatio, or terrai. A improved correctio localizatio algorithm based o dyamic weighted cetroid for wireless sesor etworks was proposed i this paper. The idea is that each achor ode computes its positio error through its eighbor achor odes i its rage, the positio error will be trasform to distace error, accordig the distace betwee ukow ode ad achor ode ad the achor ode s distace error, the dyamic weighted value will be computed. For each ukow ode, it ca use the coordiate of achor ode i its rage ad the dyamic weighted value to compute it s coordiate. Simulatio results show that the localizatio accuracy of the proposed algorithm is better tha the traditioal cetroid localizatio algorithm ad weighted cetroid localizatio algorithm, the positio error of three algorithms is decreased alog radius icreasig, where the decreased tred of our algorithm is sigificat. Copyright 204 IFSA Publishig, S. L. Keywords: Wireless sesor etworks (WSN), Cetroid localizatio, Weight, Correctio localizatio.. Itroductio With the developmet of electroics ad wireless commuicatios techologies, sesor odes have a tred of miiaturizatio, cheap ad smart. Each ode has a low-power processor, u-replaced battery power, limited computatio capacity ad a radio trasceiver for commuicatio. Nodes are resposible for gaiig data ad sedig to sik ode, thousads of odes to form a large wireless sesor etwork to moitor the huge terrais []. I wireless sesor etworks (WSN), the locatio i formatio is crucial, whe a abormal evet occurs. The sesor ode detectig the evet eeds the locatio iformatio to locate the abormal evet ad report to the sik ode. Therefore, the locatio iformatio is usually embedded i the report message geerated by the sesor ode. Without locatio iformatio, wireless sesor etworks ca t work properly. Therefore, locatio awareess has become a importat feature for may wireless sesor etworks applicatios. It provides locatio iformatio to each idividual ode i the etwork over which services like evet reportig, routig, data aggregatio ad may other higher level services ca be built [2]. Examples of such applicatios iclude target trackig, locatio-aided routig, ad safety protectio, et al. Because of cost 34

2 Sesors & Trasducers, Vol. 76, Issue 8, August 204, pp ad eergy costraits, ot all odes may have bee equipped with GPS [3] (Global Positioig System). Therefore, localizatio systems for WSN usually employ a small set of odes which kow their ow positio (called achor odes), ad the achor odes usually radomly distribute i the etwork, the ukow odes will use these achor odes localizatio to discover their coordiate. Without the locatio iformatio, the raw data would ot be useful for may applicatios [4]. Also the localizatio system s architecture iflueces the outcome of a localizatio system. It plays a more importat role especially whe we have mobile odes or/ad mobile achors i the etwork. The architecture of a localizatio system has a sigificat impact o its scalability, its ability to preserve user locatio privacy, its ease of deploymet, ad its accuracy. 2. Related Work I the past several years, a umber of localizatio protocols have bee proposed. We classify the existig localizatio algorithms ito two categories: the statioary achor localizatio algorithms ad the mobile achor localizatio algorithms. The deploymet ad umber of achor odes could greatly ifluece localizatio accuracy [5]. However, the more achor odes are, the larger the cost of deploymet etwork is. Oce all the odes are located, achor odes will be ot so importat. So some researcher used a mobile achor ode (AN) dyamic movig i the etwork to assist locatio, which ca reduce cost of computatio ad commuicatio. Furthermore, sice the AN ca move to blid areas where static achor odes do ot cover, it may commuicate with all the odes directly, which could ehace localizatio accuracy [6]. Statioary achor locatio algorithm use statioary achor iformatio for localizatio, which ca be classified as rage-based ad rage-free. The rage-based algorithm uses absolute poit-to-poit distace or agle estimates for calculatig the locatio, which are relatively precise but require additioal hardware ad their cost is relatively high [5], Commo approaches for distace or agle estimatio iclude received sigal stregth idicator (RSSI) [7]; time of arrival (TOA) [8]; time differece of arrival (TDOA) [9] ad agle of arrival (AOA) [0], TOA is based o the rage estimatios by the sigal arrival time, while TDOA relies o the differece i time betwee two arrived sigals. Agle of arrival (AOA) to estimate ode positio is proposed i [0], the priciple of AOA is to detect the origiators of sigals accordig to the agle of arrival, ad the to calculate the positios of odes by meas of triagulatio. Maximum likelihood estimatio (MLE) is a alterative used i AHLoS system (Ad-Hoc Localizatio System) [], whose aim is to miimize the differeces betwee the measured distaces ad estimated distaces to determie the positio of odes. However, equippig all sesor odes with a GPS receiver is ot a realistic solutio because it icreases the cost, size ad eergy cosumptio of the sesor odes [2]. While producig fie graied locatios, rage-based protocols remai cost-ieffective due to the cost of hardware for radio, soud, or video sigals, as well as the strict requiremets o time sychroizatio ad eergy cosumptio. Rage-based approaches ca obtai more accurate measuremets, but they require complex ad expesive hardware [3]. The rage-free scheme eables sesors to lear their locatio iformatio without the aid of rage estimates, it is suitable for sesor positioig due to its cost-effectiveess. A amorphous positioig algorithm, DV-Hop, addressed i [4], employs offlie hop-distace estimatios to improve locatio estimates via eighborig iformatio exchages. I [5], DV-Hop localizatio algorithm based o improved average hop distace ad estimate of distace cosidered the feature of coicidece or part of the overlap which existed i the path from ukow odes to beaco odes ad the path betwee achor odes, it improved average hop distace ad used the error to correct. I [6], a improved localizatio algorithm for DV-Hop based o trusty degree used estimatig the value of trusty degree to filter appropriate average distace per hop. I [7], a improved DV Hop localizatio algorithm for wireless sesor etworks used the agle to compute the distace of ukow odes ad achor odes, the agle ABC formed by three eighbor odes A, B ad C i the path of ukow odes ad beaco odes, ad give distace data amog the odes. This agle ABC was estimated with a overlappig degree of B s eighbor ode sets collectig with A ad C, respectively. I [8], a improved DV-Hop positioig algorithm based o agle threshold was proposed, which ca set the agle threshold to filter out the achor odes ad improve DV-Hop s positioig capability. New ode localizatio scheme virtual beacos-eergy ratios localizatio (VB-ERL) ad its refiemets for the WSN are preseted i [9]. I the scheme, the mobile ode moves i the surveillace field based o the Gauss-Markov mobility model ad periodically broadcasts the iformatio packets. Each static ukow ode receives the virtual beacos ad eergy i its sesig grage, ad estimates its locatio by fidig the itersectio of a set of hyperspheres. A two-objective evolutioary approach based o topological costraits for ode localizatio i wireless sesor etworks was proposed i [20], it takes cocurretly ito accout durig the evolutioary process both the localizatio accuracy ad certai topological costraits iduced by coectivity cosideratios. A directio-based localizatio scheme (DLS) was proposed i [2], whose mai goal is for each sesor to determie its directio rather tha its absolute positio, DLS cosiders multiple messages received for a sesor to determie its directio ad achor deploymet 35

3 Sesors & Trasducers, Vol. 76, Issue 8, August 204, pp strategy to improve the estimated correctess i directio of the sesor withi the commuicatio rage of the sik. Liear-regressio-based weighted cetroid localizatio algorithm i wireless sesor etwork was proposed i [22], it improved weighted cetroid localizatio algorithm by the use of hops betwee odes, cetralizatio ad the calibrated the odes' positio by the use of liear regressio. Fially, the algorithm corrected the positio through certai parameters. The more achor odes are, the larger cost of deploymet etwork is. Oce all the odes are located, achor odes will be ot so importat. So some researcher used a mobile achor ode (AN) dyamic movig i the etwork to assist locatio, which ca reduce cost of computatio ad commuicatio. Furthermore, sice the AN ca move to blid areas where static achor odes do ot cover, it may commuicate with all the odes directly, which could ehace localizatio accuracy [6]. I [23], they proposed a localizatio scheme usig a mobile achor. Each achor, equipped with the GPS, moves i the sesig field ad broadcasts its curret positio periodically. The sesor odes that obtai the iformatio are able to compute for heir locatios. To provide localizatio i etworks where lad mark desity is low, hop-based techiques propagate locatio aoucemets through out the etwork. Mobile achor localizatio algorithms also compute ode-to-ode distaces by RSSI, TOA, TDOA, ad AOA. Zhag et al. [24] proposed very low eergy cosumptio wireless sesor localizatio for dagerous eviromets with sigle mobile achor ode. But this algorithm caot esure each ode receives three o-colliear achor coordiates. Koutsoikolas et al. [25] studied the problem of path plaig for mobile achor to reduce localizatio error. Zhag et al. [26] proposed a rage-free localizatio scheme usig mobile achor odes. Whe ruig oce, this algorithm oly located a part of odes. I order to icrease localizatio efficiecy, the movemet mode of the MN eeds to be improved. Kuag et al. Virtual beacos-eergy ratios localizatio (VB-ERL) scheme usig the Gauss-Markov mobility model was proposed i [26], which was fully distributed ad did ot eed itersesor commuicatio. The Gauss-Markov mobility model provided movemet patters which might be expected i the real-world for the mobile achor ode, which took full advatage of the correlatio betwee the curret velocity ad locatio of the AN ad its future velocity ad locatio. 3. Network Model There are a set of achor odes ad a set of sesor odes i a WSN. A fixed umber of achor odes are placed with the regios of coverage overlapped ad serve as referece poits, broadcastig periodic achor sigals. The sesor odes are distributed radomly i the sesig field ad receive messages from achor odes. The mai resposibility of the achor odes is to sed out beaco sigals to help the sesor odes to locate themselves. Each sesor ode listes for a fixed time period ad collects the RISS iformatio of all beaco sigals from adjacet achor odes. I this eviromet, it is assumed that [22]. ) The etwork is a static radomly deployed etwork. It meas a large umber of sesor odes are radomly deployed i a two-dimesioal geographic space, formig a etwork ad these odes do ot move ay more after deploymet. 2) There exists oly oe Sik ode, which is deployed at a relative static place outside the WSNs. 3) There are N static achor odes, which their positios are kow through GPS or by other meas such as pre cofiguratio, ad M ukow odes, ad there is a mobile achor ode (AN). 4) The radio propagatio is perfectly spherical ad the trasmissio rages for all radios are idetical. 5) The locatio of each achor ode is kow, amely, the locatio of ode i is p(i)=(x i, y i ). 6) The Euclidea distace of arbitrary two odes (i, j) is 2 2 d i j= ( xi xj) + ( yi yj) () The sesig area has bee divided may circles which have the same radius r, each ode has the same maximum trasmissio distace R, ad 0 r R. 4. Improved Correctio Localizatio Algorithm Based o Dyamic Weighted Cetroid for Wireless Sesor Network 4.. Weighted Cetroid Algorithm The mai idea of the weighted cetroid algorithm is determie the weightig factor accordig to the differet reflectio o ukow odes from achor odes, it reflects the degree of ifluece of each achor ode for cetroid positio, which is usually accordig to the received sigal stregth (RSSI) betwee achor odes ad ukow odes. The specific approach is trasform the RSSI to distace whe the ukow ode received the iformatio of achor ode ad form a fuctio by distace, the it computes the weighted of each achor ode, last, usig these iformatio to locate the ukow ode coordiate based o cetroid algorithm. For traditioal cetroid algorithm, the coordiate of ukow ode j could be computed the followig equatio: xi yi i= i= x = ; y =, (2) 36

4 Sesors & Trasducers, Vol. 76, Issue 8, August 204, pp where is the umber of achor odes. Assume that the distace betwee achor ode i ad ukow ode u is d iu, weight value is w iu, the coordiate of ukow ode u through weighted cetroid method is show as: wx xu= = wij iu i iu i i= i= ; yu i= i= wy, (3) wij where is the umber of achor odes, w iu =, if diu the ukow ode u is ot directly commuicate with achor ode i, w iu is zero. I wireless sesor etworks, RSSI ragig is greatly affected by evirometal factors, such as oise, obstacles, multi-path reflectios ad artificial malicious attackig. The distace betwee achor ode ad ukow ode is always ot the real distace ad will lead to a large positio error, so it is ecessary to itroduce other correctio methods Improved Correctio Localizatio Based o Dyamic Weighted Cetroid Localizatio Algorithm The basic idea of our algorithm is as follow: each achor ode i oe hop rage of ukow ode j is to ru self-correct, ad trasforms the error of selfcorrect to the error of distace, each achor ode computes its trust degree accordig to fuctio of distace, the it uses weighted cetroid localizatio algorithm to gai the coordiate of ukow ode u. Assume that ukow ode u has achor odes i its rage, d iu is the distace betwee ukow ode u ad achor ode i. For achor ode i, it ca gai its estimate coordiate by Formula (2), assume that the estimate coordiate is ( x ', i y ' i),so the error of localizatio of achor ode i ca be computed as: The the weight is defied as the followig: w ui = = (7) Ti dui d _ error _ i dui So the coordiate of ukow ode u based o correctio weighted cetroid algorithm is show i the followig equatio: x xu= = wui w i i= d _ error _ i diu i= d _ error _ i diu ; yu i= i= ui yi, (8) where is the umber of achor odes i the rage of ukow ode u. 5. Simulatio Results This sectio provides a detailed quatitative aalysis comparig the performace of our scheme with traditioal cetroid algorithm ad weighted cetroid algorithm. The average of localizatio value i 0 times is tests the stability. I our experimets, the deploymet area is a square plae of 500 m by 500 m. There are 70 achor odes ad 00 ukow odes, the radius of odes are respectively 40 m 50 m ad 60 m. The average of localizatio error i 0 times is tests the stability. We ca see from Fig. to Fig. 2, the positio error of our algorithm is decreased sigificatly compared with traditioal cetroid algorithm ad weighted cetroid algorithm. With the icrease of achor odes, the positio error of three algorithms is gradually decreased. Uder the same coditios, the average positio accuracy of our algorithm is improves 43.2 % tha the traditioal cetroid algorithm, ad improves 3.25 % tha weighted cetroid algorithm. ex _ i = xi x' i, (4) ey _ i = yi y' i where (x i, y i ) is the coordiate of achor ode i, ( x ', i y ' i) is the estimate coordiate of achor ode accordig to other achor odes i its rage. It ca compute the error distace by Formula (5) 2 2 d _ error _ i = ex _ i + ey _ i (5) We defie the trust degree is T i, it ca be computed through Formula (6) Ti = (6) d _ error _ i Fig.. The relatioship betwee achor umber ad localizatio error. 37

5 Sesors & Trasducers, Vol. 76, Issue 8, August 204, pp ode s distace error, the dyamic weighted value will be computed. For each ukow ode, it ca use the coordiate of achor ode i its rage ad the dyamic weighted value to compute it s coordiate. Simulatio results show that the localizatio accuracy of the proposed algorithm is better tha the traditioal cetroid localizatio algorithm ad weighted cetroid localizatio algorithm, the positio error of three algorithms is decreased alog radius icreasig, where the decreased tred of our algorithm is sigificat. Fig. 2. The relatioship betwee achor umber ad localizatio error. We ca see from Fig. 3 to Fig. 5, the positio error of three algorithms is decreased alog radius icreasig. O the whole, the improved weighted cetroid algorithm is decreased tha traditioal cetroid algorithm ad 3.8 tha weighted cetroid algorithm. Fig. 4. The relatioship of localizatio error for weighted cetroid localizatio algorithm. Fig. 3. The relatioship of localizatio error for traditioal cetroid localizatio algorithm. 6. Coclusios A umber of algorithms have bee proposed for the localizatio problem i wireless sesor etworks. Yet precise locatio iformatio may be uavailable due to the costrait i eergy, computatio, or terrai. A improved correctio localizatio algorithm based o dyamic weighted cetroid for wireless sesor etworks was proposed i this paper. The idea is that each achor ode computes its positio error through its eighbor achor odes i its rage, the positio error will be trasform to distace error, accordig the distace betwee ukow ode ad achor ode ad the achor Fig. 5. The relatioship of localizatio error for improved weighted cetroid localizatio algorithm. Ackowledgemet We ackowledge the support of Six Top Level Talet Fuds Jiagsu Provice, Natioal Developmet ad Reform Commissio Security Fuds - "Mobile Iteret security audit system", we sicerely thak the aoymous reviewers for their costructive commets ad suggestios. 38

6 Sesors & Trasducers, Vol. 76, Issue 8, August 204, pp Refereces []. I. F. Akyildiz, S. Weilia, Y. Sakarasubramaiam, E. Cayirci, A survey o sesor etworks, IEEE Commuicatios Magazie, 40, 8, 2002, pp [2]. Bi Yag, Jiwu Xu, Jiahog Yag, Mi Li, Localizatio algorithm i wireless sesor etworks based o semi-supervised maifold learig ad its applicatio, Cluster Computer, 3, 200, pp [3]. B. Hofma-Wellehof, H. Lichteegger, J. Collis, Global Positioig System: Theory ad Practice, Spriger - Verlag, 993. [4]. N. Patwari, A. O. Hero, M. Perkis, et al., Relative locatio estimatio i wireless sesor etworks, IEEE Trasactios o Sigal Processig, 5, 8, 2003, pp [5]. A. Savvides, C. C. Ha, M. B. Strivastava, Dyamic fie-graied localizatio i Ad-Hoc etworks of sesors, i Proceedigs of the 7 th Aual Iteratioal Coferece o Mobile Computig ad Networkig, Rome, Italy, 200, pp [6]. Zhog Zhi, Luo Da-Yog, Liu Shao-Qiag, Fa Xiao-Pig, Qu Zhi-Hua, A Adaptive Localizatio Approach for Wireless Sesor Networks Based o Gauss-Markov Mobility Model, Acta Automatica Siica, 36,, 200, pp [7]. H. Wu, C. Wag, N. Tzeg, Novel self-cofigurable positioig techique for multi-hop wireless etworks, IEEE/ACM Trasactios Network, 3, 2005, pp [8]. N. B. Priyatha, A. Chakraborty, H. Balakrisha, The cricket locatio-support system, i Proceedigs of the ACM Iteratioal Coferece o Mobile Computig ad Networkig (MOBICOM), August 2000, pp [9]. Y. B. Ko, N. H. Vaidya, Locatio aided routig (LAR) i mobile ad hoc etworks, Wireless Networks, 6, 4, 2000, pp [0]. A. Savvides, C. C. Ha, M. B. Strivastava, Dyamic fie-graied localizatio i ad-hoc etworks of sesors, i Proceedigs of the ACM Iteratioal Symposium o Mobile Ad Hoc Networkig ad Computig (MOBIHOC), August, 200, pp []. D. Niculescu, B. Nath, Ad hoc positioig system (APS) usig AoA, i Proceedigs of the IEEE INFOCOM, the Aual Joit Coferece of the IEEE Computer ad Commuicatios Societies, March, 2003, pp [2]. B. F. Oliveira, etc., Localizatio i time ad space for wireless sesor etworks: A efficiet ad lightweight algorithm, Performace Evaluatio, 66, 2009, pp [3]. Zhag Mig, A Improved Locatio Algorithm for DV-HOP based o Trusty Degree Average Hop Distace i Wireless Sesor Networks, Sesors & Trasducers, Vol. 23, Issue 7, July 203, pp [4]. D. Niculescu, B. Nath, DV based positioig i ad hoc etworks, Telecommuicatio Systems, 22,, 2003, pp [5]. She Mig-Yu, Zhag Yi, DV-Hop localizatio algorithm based o improved average hop distace ad estimate of distace, Applicatio Research of Computers, 28, 2, 20, pp [6]. Shi Tig-Ju, Sag Xia, Xu Li-Jie, Yi Xi-Chu, A Improved Localizatio Algorithm for DV-Hop Based o Trusty Degree, Microeletroics & Computer, 25, 0, 2008, pp [7]. Zhag Xiao Log, Xie Hui Yig, Zhao Xiao Jia, Improved DV Hop localizatio algorithm for wireless sesor etworks, Computer Applicatios, 27,, 2007, pp [8]. Yag Lei, Zhag Zheg-Bao, Xie Gui-Hai, Mig Liag, Wag Xi-Feg, Improved DV-Hop Positioig Algorithm Based o Agle Threshold, Computer Egieerig, 34, 20, 2008, pp [9]. Kuag Xig-Hog, Shao Hui-He, Feg Rui, A New Distributed Localizatio Scheme for Wireless Sesor Networks, Acta Automatica Siica, 34, 3, 2008, pp [20]. Massimo Vecchio, Roberto Lpez-Valcarc, Fracesco Marcelloi, A two-objective evolutioary approach based o topological costraits for ode localizatio i wireless sesor etworks, Applied Soft Computig, 20, pp. -. [2]. Sheg-Shih Wag, Kuei-Pig Shih, Chih-Yug Chag, Distributed directio-based localizatio i wireless sesor etworks, Computer Commuicatios, 30, 2007, pp [22]. Hu Yu, Yao Weizhao, Liear-regressio-based weighted cetroid localizatio algorithm i Wireless Sesor Network, Procedia Egieerig, 5, 20, pp [23]. K.-F. Ssu, C.-H. Ou, H. C. Jiau, Localizatio with mobile achor poits i wireless sesor etworks, IEEE Trasactioso Vehicular Techology, 54, 3, 2005, pp [24]. R. B. Zhag, L. L. Zhag, Y. B. Feg, Very low eergy cosumptio wireless sesor localizatio for dager eviromets with sigle mobile achor ode, Wireless Persoal Commuicatios, 47, 4, 2008, pp [25]. D. Koutsoikolas, S. M. Das, Y. C. Hu, Path plaig of mobile ladmarks for localizatio i wireless sesor etworks, Computer Commuicatios, 30, 3, 2007, pp [26]. Zhag Zheg Yog, Su Zhi, Wag Gag, Yu Rog, Mei Su-Liag, Localizatio i wireless sesor etworks with mobile achor odes, Joural of Tsighua Uiversity (Sciece ad Techology), 47, 4, 2007, pp (i Chiese). 204 Copyright, Iteratioal Frequecy Sesor Associatio (IFSA) Publishig, S. L. All rights reserved. ( 39

WSN Node Localization Regularization Algorithm Based on Quasi Optimal Criterion Parameter Selection

WSN Node Localization Regularization Algorithm Based on Quasi Optimal Criterion Parameter Selection Sesors & rasducers Vol. 23 Special Issue July 203 pp. 94-98 Sesors & rasducers 203 by IFSA http://www.sesorsportal.com WSN Node Localizatio Regularizatio Algorithm Based o Quasi Optimal Criterio Parameter