NONLINEAR STATE SMOOTHING FOR DISCRETE DYNAMIC-SYSTEMS WITH PAST HISTORIES

A state smoothing scheme is presented for dynamic systems with past histories. The state in the future is a given (linear or non-linear) function of the disturbance noise and both the present and N - 1 past discrete values of the state. An observation in the present is a given function of the observation noise and both the present and N - 1 past discrete values of the state. The proposed scheme is based upon multiple hypothesis testing and the Viterbi algorithm of information theory. Simulation results, some of which are presented, have shown that the proposed scheme performs well.