A new localization algorithm in complex environment for wireless sensor networks

Localization is one of the supporting technologies in wireless sensor networks. Many techniques have been proposed that attempt to localize in a distributed manner. But the border problem (nodes near the edge of the sensor field are poorly positioned) is not solved perfectly. In this paper we propose a novel localization algorithm with the following merits. First, we use classification method to classify the nodes into two categories - border nodes and non-border nodes. Then we localize the border nodes based on mere connectivity information (i.e., hop counts only), and deal with the non-border nodes based on improved weighted centroid algorithm. The simulation results show that the border problem is solved well and the average localization error will be reduced to about a quarter and 10% in the complex environment with fewer anchor nodes compared to the centroid algorithm and diffusion algorithm.