Information weighted consensus filtering with improved convergence rate

In this paper, we propose a novel method to improve the accuracy of information weighted consensus filtering (IWCF). Recently, the IWCF was introduced for the distributed estimation in the scenario of target tracking using a camera network. In contrast to other consensus methods, the IWCF can overcome the naivety and redundancy problems which exist in many real applications. Furthermore, the IWCF can converge to the central filtering method when the consensus step tends to the infinity. However, the IWCF algorithm uses a deterministic consensus weight to achieve the average consensus, which affects its accuracy when only a finite number of consensus steps is performed. This paper tries to improve the accuracy of the IWCF by using the Metropolis weights, which can guarantee the convergence of average consensus and has faster convergence rate. Finally, a target tracking experiment using the distributed camera network is demonstrated to show the advantages of the proposed method.