Improved Multiple Model Particle Filter for Maneuvering Target Tracking in the Presence of Delayed Measurements

In this paper, a particle filter is designed to deal with a class of nonlinear systems where the measurements undergo multi-step random delay due to the limited transmission capability of data link. First, a discrete-time variable governed by a Markov chain is defined to describe the measurement random delay. Considering the target maneuver, the multi-model method is adopted where another discrete-time variable is introduced to model the system as a jump Markov system. Second, the two newly defined variables are used to augment the original state vector, and then a hybrid system containing two kinds of discrete-time components is obtained. Finally, the hybrid system is estimated in the particle filtering framework and the simulation results show the proposed filter can effectively deal with the systems where the target maneuver and measurement delay occur simultaneously.