A Research On Localization Technique Of Wireless Sensor Networks

Wireless sensor networks is self-organized network composed of lots of sensorsdistrusted in a monitored area, the sensors communicate with each other according to a certainprotocol. The locations of the sensors play an important role in a variety of the applications ofwireless sensor networks, since the measured data obtained by the sensors becomemeaningless if the locations of the sensors are unknown. In addition, the locating informationis helpful for the design of network layer and application layer. All in all, localizationbecomes an indispensable issue in wireless sensor networks.This dissertation focuses on the research of localization of wireless sensor networks.Through in-depth and systematic study of existing localization algorithms, this paper made agood attempt in the following aspects:According to a detailed error analysis, we found out that, DV-hop algorithm does notperform well within anisotropic network, because the inaccuracy of the calculation of theaverage distance per hop leads to the inaccurate estimation of the distance between theunknown nodes and the anchors. Thus, this paper presented an improved DV-hop algorithm tocalculate the average distance per hop for each pair of anchors, which reflect the localtopology more objectively. In addition, the connectivity between the unknown node and thereference node is transformed into geometric constraints, narrowing down the potential areawhere the unknown node is supposed to be.Through simulation, we found out that, once the hop count between an unknown nodeand an anchors is settled, the larger the distance between the unknown node and the anchor is,the more the unknown node’s neighbors whose hop to the anchor is smallest is, and vice versa.Thus, it is reasonable to accept that, there is a relationship between the distance from theunknown node to the anchor and the distribution of the neighbors of the unknown nodes. Inorder to derive this relationship, this paper proposed a novel localization algorithm usingexpectation of node-to-anchor distance. As for the reference node and the unknown node, thealgorithm respectively classifies their neighbors into3sets according to the neighbors’ hopcount to the anchor, then, the communication range of the reference node and the unknownnode is divided into different areas respectively. Finally, the distance between the unknown node and the anchor is calculated by utilizing probability theory and geometric analysis. Thealgorithm makes use of the hop count information of the unknown node and its neighbors, aswell as the hop count information of the reference node and its neighbors, to establish aquantitative relationship between the distance from the unknown node to the anchor and theexpectation of the distance from the unknown node to the anchor. The algorithm enhances theaccuracy of distance estimation, reducing the positioning error.