A New Particle Filter Tracking Algorithm For Doa Sensor Systems

In this paper, we present new a bearings-only tracking algorithm based on particle filtering (PF) and a state-space discretization technique for DOA sensor systems. The discretization stage is used to generate the proposal distribution used in the PF algorithm. More specifically, we first partition the state space into finite cells, and use the cell centers to represent the "discretized" state space. Next, we calculate the weight for each cell based on the target initial distribution, and then propagate only a small number of cell center, whose weights are above certain threshold, through the target dynamic system. At each iteration, the cells with large weights represent the high probability area in the target distribution. The samples (or particles) are drawn from this area, which is served as the proposal distribution in the new particle filter. We apply this algorithm to a bearings-only tracking problem. As shown in the theory and indicated by our simulations, this proposal renders more support from the true posterior distribution, thereby significantly improves the estimation accuracy compared to particle filter algorithms.