A Switch-Constrained Multiple Model Smoothing Algorithm for Maneuvering Target Tracking

In most existing multiple model (MM) smoothing algorithms for maneuvering target tracking, the mode evolution process is usually described by a Markov chain, which implicitly assumes that the target mode may switch at all time steps. This assumption contradicts the fact that the target motion mode will persist for at least a certain period of time. In this paper, a fixed-lag smoothing algorithm, named switch-constrained interacting multiple model smoothing (SC-IMM-S), is proposed based on state augmentation. The SC-IMM-S employs two strategies to accurately describe the mode evolution process. First, model switching is constrained to occur no more than once within three consecutive time steps, which provides a deterministic description that the target mode will not switch continuously. Second, the probability of model transiting to itself is set to an extremely large value to characterize the fact that mode remaining unchanged is the main trend of mode evolution. Simulation results show that the performance of the SC-IMM-S algorithm is superior to that of traditional Markov-chain-based MM smoothing algorithms.