Distributed Joint Robust Filter for WSNs-Based Target Localization Against Non-Gaussian Noises

Abstract

The demands for high localization accuracy in wireless sensor networks (WSNs) have led to the development of various innovative algorithms. This article proposes a distributed joint robust filter (DJRF) algorithm for addressing fluctuant localization errors caused by the measurement noises and ill-conditioned matrix solution. Firstly, the distributed solution is established after considering the non-Gaussian noises, drifted anchor node coordinates as well as unknown point-to-point distances. Secondly, since the outliers may seriously degrade the localization performance, the robustness filter integrated with virtual observation is used to estimate the point-to-point distances. Thirdly, the variable separation method that can weaken the influence of ill-conditioned matrices is proposed for mobile target localization without generating large errors. The numerical results show that the proposed DJRF combines the merits of distributed solution and robustness filter, which effectively deal with various uncertainties under an acceptable amount of computation. Furthermore, the DJRF experimental result has a 0.26 m average error under non-Gaussian noises and increases to a 0.33 m average error in a more actual scenario. The DJRF algorithm uses outlier elimination and a robust filter to suppress its divergent trend, which can further verify the superiority compared with relevant localization algorithms.