Expectation-maximization (EM) algorithm for instantaneous frequency estimation with Kalman smoother

In this paper, we propose an expectation-maximization (EM) algorithm based approach for instantaneous frequency (IF) estimation in a Kalman smoother framework. We formulate time-varying AR (TVAR) model as a state-space model and describe EM algorithm for model parameter estimation. This is used with Kalman smoother for IF estimation. We show that our scheme EMIF shows best performance among the other existing adaptive algorithms like RLS and LMS. Performance analysis is reported on a class of FM signals.