No abstract

The drop in cost and reduction in size of sensor nodes has eased the development of wireless sensor networks (WSNs) applications. However, the noise and disturbing nature of most sensing environments require accurate algorithms that can overcome these difficulties. Nodes' localization is one of the basic activity a WSN can perform to make other network's functionalities, such as routing easy to tackle. Nowadays there exists many localization methods, however many pose computational and/or accuracy issues. Centroid is a localization algorithm by which an unknown node's coordinates are estimated as the centroid of anchors' coordinates. Its implementation is simple but it has a high error rate. In this paper, two methods are proposed to enhance the centroid localization algorithm. The first, Linear Weighting Centroid (LWC) uses the distance between the anchor and the unknown nodes to linearly weight each anchor's coordinates. The second, the Neighbor Weighting Centroid (NWC) uses the number of intersect nodes between an unknown node and its neighbor anchors to estimate the degree of proximity of the anchor nodes. Both methods assign larger weights to closer anchors and lesser weights to remote anchors to improve centroid accuracy while keeping computation almost at the same level. Simulation is used to study the performance of both mechanisms. The results show that for a large WSN, both methods localize unknown nodes with better position accuracy than centroid, with LWC performing better than NWC.

No abstract