Consensus-based joint target tracking and sensor localization

TL;DR

This paper introduces a distributed algorithm that combines target tracking and sensor self-localization using consensus-based Kalman filtering, optimizing drift parameters without extra data exchange and demonstrating effectiveness in various network configurations.

Contribution

It extends consensus-based Kalman filtering to jointly estimate target states and sensor positions in a distributed network, with an online cost minimization approach for drift parameters.

Findings

Effective in tree and cyclic networks

Works for linear and nonlinear sensors

Requires minimal additional resources

Abstract

In this paper, consensus-based Kalman filtering is extended to deal with the problem of joint target tracking and sensor self-localization in a distributed wireless sensor network. The average weighted Kullback-Leibler divergence, which is a function of the unknown drift parameters, is employed as the cost to measure the discrepancy between the fused posterior distribution and the local distribution at each sensor. Further, a reasonable approximation of the cost is proposed and an online technique is introduced to minimize the approximated cost function with respect to the drift parameters stored in each node. The remarkable features of the proposed algorithm are that it needs no additional data exchanges, slightly increased memory space and computational load comparable to the standard consensus-based Kalman filter. Finally, the effectiveness of the proposed algorithm is demonstrated through simulation experiments on both a tree network and a network with cycles as well as for both linear and nonlinear sensors.