Multirate Interacting Multiple Model Algorithm Combined with Particle Filter for Nonlinear/Non-Gaussian Target Tracking

Abstract

In this paper, we propose a new interacting multiple model (IMM) algorithm combined with particle filter for nonlinear/non-Gaussian systems, which adopts the multirate technique to improve the computational efficiency. The interacting multiple model (IMM) algorithm is specially designed to track accurately targets, and the particle filter is aim to deal with nonlinear/non-Gaussian problems. But the problem of a particle filter is its expensive computation, especially when it is introduced into the IMM algorithm. Here, the multirate technique is to solve this problem and not making the performance of the algorithm bad. The experimental results show the multirate IMMPF (IMM particle filter) works as well as IMMPF with much lower computation load.