A Weighted Evidence Combination Approach for Target Identification in Wireless Sensor Networks

Information fusion using evidence theory in wireless sensors networks has been used extensively to identify targets because it offers the advantage of handling uncertainty. But the classical Dempster's combination rule cannot deal with highly conflicting information because it often generates counterintuitive results. In this paper, a new weighted evidence combination approach is proposed to solve this problem. First, two measures, i.e., a new contradiction measure of each body of evidence (BOE) and a probabilistic-based dissimilarity measure between two BOEs, are introduced to estimate the value of weight of each sensor. Then, when combining conflicting information, reasonable results can be produced by using weighted average of BOEs and Dempster's rule. Our experimental results showed that the proposed method has better performance in convergence than the existing methods.