Innovation filter and its application to the IMM algorithm using Zhou model

The interacting multiple model (IMM) algorithm is now a more efficient algorithm for tracking targets, however, the weight probabilities of the algorithm are more sensitive to the measurement noise. In this paper, the authors introduce an innovation filter (IF) to the algorithm to overcome this shortcoming. The inherent mechanism between the IF and IMM algorithm is revealed from theoretical analysis and verified by Monte Carlo simulations. It is different from Blom´s IMM algorithm, as the Zhou model is used instead of the CA model. The Zhou model is much more efficient for tracking high maneuvering targets. However, while the targets are moving with constant velocity, it will lose accuracy. The authors show that the new method makes the two approaches, the IMM algorithm and Zhou model, more efficient