Fixed-interval smoothing for Markovian switching systems

A suboptimal approach to the fixed-interval smoothing problem for Markovian switching systems is examined. A smoothing algorithm is developed that uses two multiple-model filters, where one of the filters propagates in the forward-time direction and the other one propagates in the backward-time direction. A backward-time filtering algorithm based on the interacting multiple model concept is also developed. Results from a simulation example are given to illustrate the performance of the smoothing algorithm with respect to that of filtering. The example involves radar tracking of a Mach 1 aircraft.