Fusion of possible biased local estimates in sensor network based on sensor selection

The paper addresses the problem of estimation fusion in sensor network, in the presence of possible biased local estimates. A sensor selection-based fusion scheme is presented to deal with this problem, which seek to select a subset of sensors to be fused, for the purpose of achieving a better estimation performance. Firstly, we introduce an optimization criterion over any given subset of sensors based on the similarity measure among local estimates. Secondly, we invoke the cross entropy (CE) method to solve the resulting combinatorial optimization problems. We also explore the efficiency and performance of the proposed approach via simulation experiments, compared with other recently proposed techniques.