Localization in Wireless Sensor Networks

of thesis entitled Localization in Wireless Sensor Networks submitted by King-Yip Cheng for the degree of Master of Philosophy at The University of Hong Kong December 2006 Localization in wireless sensor networks is the process of determining the geographical positions of sensors. Only some of the sensors (anchors) in the networks have prior knowledge about their geographical positions. Localization algorithms use the location information of anchors and estimates of distances between neighbouring nodes to determine the positions of the rest of the sensors. In this work, modifications to Ad Hoc Positioning System (APS) [7] [8] are proposed to improve its performance in anisotropic networks. Only selected anchors instead of all anchors are included in the multilateration process. The nearest three anchors that form a convex hull embedding the sensor are used to localize the sensor. A heuristic-based Convex Hull Detection Method (CHDM) is used to detect whether the anchors form a convex hull embedding the sensor. Simulation results suggest that the modifications are considerably more accurate in anisotropic networks than the original APS. The CHDM is also applicable to localization systems based on proximitydistance map (PDM) [20]. The effects of the number and placement of anchors are also investigated. The performance of a PDM-based localization system is severely degraded if anchors are clustered together. A phased approach, MDS+CHDM, is proposed to alleviate the degradation. In the first phase, MDSMAP [11] is used to localize some nodes in the network as it shows less dependence on the number and placement of anchors. The localized nodes then become secondary anchors. The rest of the sensors are localized by PDM in the second phase. The phased approach is tested by extensive simulations. (Total words: 248) Signed King-Yip Cheng Localization in Wireless Sensor Networks