Maximum Likelihood Frequency Estimation of Periodic Signals with Slowly Time-Varying Amplitude and Phase

Abstract

This paper applies the Expectation-Maximisation (EM) algorithm for computing the maximum likelihood (ML) estimate of the fundamental frequency (constant parameter) of a complex multiharmonic signal measured in noise. Most importantly, the individual harmonics have time-varying complex amplitudes, conditional mean estimates of these are obtained using a fixed interval Kalman smoother.