Reliable Node Localization In Wireless Sensor Networks

Node localization is one of the most important technologies in wireless sensor networks and has appealed significant attention.Traditional node localization methods assume that the location reference provided by anchor node is reliable,so these methods are not adapt to some scenarios with unreliable anchors such as drifted anchors,malicious anchors,bogus anchors made by worm-hole attack and anchors that offer perverse beacons influenced by NLOS(Non Line Of Sight)environment.The node localization issue in these scenarios mentioned above are called reliable localization.Focusing on the reliable localization,this dissertation proposed four solution based on sufficient research and analysis from different perspectives.The major contributions of this work include as follows:(1)Take the drifted anchors into account,a centralized robust localization algorithm RI-MDS(RSSI based Iterative Multidimensional Scaling)is proposed.RI-MDS is an iterative algorithm combined the metric-MDS(Multidimensional Scaling)and nonmetric-MDS.After received the RSSI information measured by each node,the backbone server produces dissimilarity matrix using metric-MDS and the relative position of each node is determined based on nonmetric-MDS.Finally the relative coordinate of each node is converted to absolute coordinate suing affine transformation and maximum likelihood estimation.In addition to the error tolerance ability provided by RI-MDS,drifted anchors can be recognized by comparing the relative location matrix produced by two consecutive operations of RI-MDS.And the re-localization process can be fulfilled by promoting some normal nods as temporal anchors.Simulations and physical experiments show that the localization accuracy of RI-MDS is pretty high,Furthermore,RI-MDS can verifies the node location effectively and its’ localization accuracy still satisfies the requirement of application even after the re-localization process.(2)In order to fulfill reliable localization in large scale wireless sensor networks with both drifted anchors and malicious anchors,a reliable node localization algorithm based on distributed virtual-force is proposed including cooperative localization scheme called LVA and location verification scheme called VLVA.This whole set of schemes present the mutual location observation of certain pair of nodes as virtual-force,and integrate the relative distances and bearings constraints of nodes elegantly.LVA produces significantly accurate location of node by two refinements with different granularities and VLVA filters out unreliable anchors by recognizing the gap of resultants.After filtering these unreliable anchors,the re-localization is conducted by promoting some normal nodes as temporal anchors based on localization accuracy to recalibrate the location of normal nodes in the presence of insufficient of anchors.Extensive experiments are conducted to evaluate validity and practicability of the set of schemes.(3)By considering the problem that traditional localization methods ignore the judgement of anchor location so that cannot meet the requirement of reliable localization,a reliable localization framework called RLF is proposed.RLF can be served as an underlayer service for both of the range-based localization method and the range-free localization method to ameliorate these traditional methods applicable in reliable localization.A distributed reputation model with direct reputation value and third-party reputation value is introduced in RLF and a credibility updating scheme is also adopted to ensure the consistence of mutual node observation.RLF can filter out the unreliable anchors and fulfill the re-localization efficiently using the lightweight reputation model.In addition,the grading hop estimation scheme designed to improve the accuracy of reputation value for range-free localization methods is also useful to improve localization accuracy.(4)In order to solve the node location indetermination problem caused by signal NLOS propagation in indoor environment,an indoor localization method based on RTT(Round Trip Time)and AOA(Arrival of Angle)using a lightweight grid-based clustering(LGBC)algorithm is proposed.LGBC produces a location candidate collection based on all the received location reference,and determines the final location of certain node via a grid clustering method.The LGBC algorithm does not depend on any prior information of indoor environment and possesses significant flexibility.The simulation results show that LGBC algorithm has low time complexity and small computational overhead.Furthermore,it outperforms the other method in terms of localization accuracy whether the NLOS signal distribution is uniform or Gaussian and exponential.Focusing on anchor location verification and filtration,four solutions are proposed from different perspectives to implement the reliable localization in WSN with unreliable anchors.