Multiple-model Bernoulli filters—part II: A sequential Monte Carlo implementation

Abstract

The multiple-model (MM) Bernoulli filter and it's Gaussian mixture (GM) implementation has been proposed for joint maneuvering target detection and tracking. However, the GM-MM Bernoulli filter is not directly applied to nonlinear non-Gaussian models. This paper proposes a sequential Monote Carlo (SMC) implementation of the MM Bernoulli filter for nonlinear models, and compares it's filtering performance with the single model SMC Bernoulli filters. Simulation results have demonstrated that the proposed SMC-MM Bernoulli filter is superior to the single model SMC Bernoulli filters for maneuvering target tracking.