IMM tracking of maneuvering targets in the presence of glint

A problem of handling multiple types of discrete uncertainties (modes) in a dynamic system is addressed. The dynamic model for a target that exhibits maneuvers (one type of discrete uncertainty) and radar measurements that are perturbed by glint noise (another type of discrete uncertainty) is an example of such hybrid system with two independent mode sets. It is shown how an interacting multiple model (IMM) estimator can be efficiently designed for such systems. Doing the mixing step of the IMM in a number of layered stages, yields some computational savings when the mode sets are independent. Computer simulations are provided for two examples of maneuvering targets in the presence of glint. The results show that the IMM estimator with layered sets of models for glint and maneuver yields better performance (in terms of position and velocity root mean square error) compared with an IMM which handles maneuvers via multiple models and glint via nonlinear "score function" based filters.