Robust Node Localization Based on Distributed Weighted-multidimensional Scaling in Wireless Sensor Networks

Abstract

This paper focuses on the robustness of node localization in various topological and sparse network. By taking account of the number of 1-hop neighboring nodes, the node position accuracy and the ranging errors, we introduce concepts of node relative localization error and relative reliability, and then propose a robust node localization algorithm based on distributed weighted-multidimensional scaling. It adaptively chooses those neighboring nodes with high relative reliability to join in the node position refinement according to local node density and their relative localization errors within 2 hops, and adopts a weighting scheme proportional to the relative reliability which emphasizes the lowest relative error within the sensor networks. For received signal strength based range measurements, extensive simulation shows that this algorithm can prevent large localization errors from spreading through the networks. Compared with dwMDS (G), this algorithm can decrease iterative times by one half and gain about 5% smaller localization errors in sparse node density or anisotropic topologies.