A novel distributed fusion algorithm for multi-sensor nonlinear tracking

Abstract

The covariance intersection (CI), especially with feedback structure, can be easily combined with nonlinear filters to solve the distributed fusion problem of multi-sensor nonlinear tracking. However, this paper proves that the CI algorithm is sub-optimal, thus degrading the fusion accuracy. To avoid such an issue, a novel distributed fusion algorithm, namely Monte Carlo Bayesian (MCB) algorithm, is proposed. First, it builds a distributed fusion architecture based on the Bayesian tracking framework. Then, the Monte Carlo sampling is incorporated into this architecture to form a feasible solution to nonlinear tracking. Finally, the simulation results verify that our MCB algorithm advances the state-of-the-art distributed fusion of nonlinear tracking.

A nonlinear and non-Gaussian distributed fusion based on Rao-Blackwellized particle filtering

“…According to (2) and the expression of z k , we notice that the nonlinear part of target state is x n k = [d x,k , d y,k ] T , and the linear part is…”

“…As seen in this algorithm that, at time step k, each particle x Input: N ∈ N + , the prior probability p 0 of x n , the meanx l 0 and varianceP ll 0 of x l , the tracking model (2).…”

“…Note that, x n,(i) k+1|k is the prediction of the nonlinear part in the i-th particle, which is drawn from p(x n k+1|k |x (i) k|k ) by a prediction process in [11]. However, according to the SS model (2) and the procedure of Rao-Blackwellisation [7], [12], p(x n k+1|k |x…”

“…Lemma 1. Given the linear/nonlinear SS model (2), the correlation between x n k and x l k in the update step can be calculated as follows,…”

“…Distributed fusion [1], [2] is a fundamental structure for multi-sensor tracking networks. In this structure, Covariance Intersection (CI) [3] algorithm is normally used as a conventional fusion algorithm which directly combines estimations from sensors to form fusion results.…”