Abstract:

The node selflocalization is one of the supporting technologies in wireless sensor networks. A distributed cluster localization scheme (linear programming-cluster localization scheme, LP-CLS) is introduced, which is based on linear programming. First, the authority of node is defined as the criterion for voting cluster and the distance between neighbor nodes is estimated by the relation between received signal strength indicator and communication range. Then, according to linear programming, the relative coordinates of nodes in a cluster are calculated out by the relevant cluster node. Finally, starting from sink node, coordinates between neighbor clusters are combined and the whole relative coordinates are converted to absolute coordinate. The simulation results indicate that LP-CLS outperforms Convex significantly in the aspect of localization accuracy and algorithm complication whether nodes are placed regularly or randomly.