Region Based Localization In Wireless Sensor Networks

The development of Micro Electromechanical System brings a new research area,wireless sensor network(WSN), which is organized by multiple sensor nodes with sens-ing, computing and communicating ability. WSN logically integrates informationworld and physical world together and has universal military and civil applications.The knowledge of sensor node position is required by many WSN applications. How-ever, node localization has several issues.To reduce network budget, a small number of nodes carrying GPS receiver act asanchor nodes and other nodes localize themselves based on pairwise distance or anglebetween neighbor nodes. However, the measurements can be noisy due to inaccuratemeasurement equipments. In addition, both deployment terrian and accuracy ofanchor node position are important decision factors of localization performance. It isnecessary to propose a localization method which is capable of evaluating localizationaccraucy. On the other hand, the position accuracy of sensor nodes largely impactsthe performance of a typical WSN application, target tracking. This is due to the factthat target position information is decided by sensor node position. Most state-of-arttarget tracking algorithms assume node position is accurate and do not consider theimpacts of inaccurate node position.This dissertation aims to solve above problems and investigate WSN node lo-calization techniques. First, we proposed a method to evaluate and reduce nodeposition uncertainty. And then, a node localization algorithm based on feasible geo-graphic region is given. Finally, we discussed how to track targets with nodes withuncertain positions. These works are supported by the National Science Foundationof China"Information Processing and Transmission on Reconstructable MIMO-basedWireless Sensor Networks"(No.60572049) and"Research on Key Issues on CognitiveRadio Networks"(No.60602029), the Foundation of Hubei Provincial Key Labora-tory of Smart Internet Technology"Research on Key Issues on Cognitive Radio Net-works"(No.HSIT200605). The contributions of this dissertation include: 1. Node position uncertainty in WSN A concept"Feasible Geographic Re-gion(FGR)"is proposed to evalute node position uncertainty. A constraint setis first contructed by collecting geometric constraints all over the network. Andthen, the FGR of a sensor node is computed by projecting the feasibe regionof this constraint set onto a 2D plane. Approximation results are given usingsimple polygons. This region clearly re?ects the deviation between actual andestimated node position. After that, node position uncertainty can be reducedby increasing transmission power of sensor nodes, deploying new anchor nodesor norml sensor nodes.2. WSN node Localization based FGR A localization algorthim which aimsat processing noisy distance measurements, angle measurements or both of themis given. This distributed and iterative algorithm which finally converges toglobal computation resuls is designed base on network clustering and provides away to handle irregular deployment terrian and inaccurate anchor node position.3. Target tracking with uncertain node position A cooperative targettracking algorithm which assumes sensor nodes are bounded by FGR is proposedto investigate the impacts of uncertain node position to tracking performance.During tracking, the maximum target speed is used to predict possible targetposition at the time of next observation. The algorithm can distinguish multipletargets and give accurate tracking results.