Maneuvering target tracking via model transition hypotheses

In this paper, a concept for designing a tracking filter is proposed. It is composed of generation and evaluation of target model transition hypotheses and estimation of target state. Since the hypotheses increase exponentially, two realizable suboptimal filters are proposed based on hypothesis reduction techniques. It is also discussed that IMM and VD filters can be considered as suboptimal filters based on the proposed concept. Simulation results show that the Kalman filter restricted by the MAF hypothesis which is one of the proposed suboptimal filters shows the best performance.