Distributed Variational Measurement Update for Extended Target Tracking with Random Matrix

The Gamma Gaussian inverse Wishart (GGIW) extended target model leads to an elegant implementation of the Bayesian estimation for the extended target tracking. In this paper, we proposed the distributed variational Bayesian measurement update for the GGIW implementation. Firstly, a centralized multi-sensor measurement update algorithm for GGIW model was derived using variational Bayesian inference. Then, we proposed the distributed variational measurement update based on the alternating direction method of multipliers (ADMM) algorithm. Furthermore, we introduced a regularization term representing a constraint of the prior information and the optimality of the estimation. We presented a heuristic solver which divided the formulated optimization problem into three sub-problems and solved alternately in the framework of ADMM. The simulation results verify the effectiveness and show the comparable performance to that of the centralized algorithm with faster convergence.