基于消息传递的机载雷达组网航迹融合

Airborne radar network track fusion requires three sub-problems: target tracking, data association and track management, which are coupled with each other and therefore require a joint solution. In this paper, we propose a track fusion method for airborne radar network using massage passing, which jointly solves the three sub-problems of target tracking, data association and track management. Firstly, the joint probability density function of airborne radar network track fusion is established, and its factorised form is converted into a factor graph. The statistical model of the target kinematic state is a conjugate exponential, and the mean-field approximation is used to obtain a simple message passing. The data association contains one-to-one constraints, and belief propagation is used. The target visibility state is also approximated by belief propagation to obtain better approximation results. Finally, the posterior probability densities can be approximated by a closed-loop iterative framework to effectively deal with the coupling problem between target tracking, data association and track management. Simulation results show that the proposed algorithm outperforms the multiple hypothesis tracking algorithm and the joint probability density association algorithm.