Algorithms for instantaneous frequency estimation: a comparative study

This paper examines the problem of instantaneous frequency (IF) estimation for Frequency Modulated (FM) signals imbedded in white Gaussian noise. It reviews currently available techniques and in addition proposes some new ones based on a modelling of the signal phase as a polynomial. Both linear least-squares techniques and Maximum Likelihood (ML) techniques are investigated for estimating the polynomial coefficients. It is seen that the linear least squares approach is efficient (i. e. unbiased and meets the Cramer-Rao bounds) for high SNR while the ML scheme is efficient for a much larger range of SNR. Theoretical lower variance bounds are given for estimating the polynomial coefficients and are compared with the results of simulations. Guidelines are given as to which estimation method should be used for a given signal class and Signal to Noise Ratio (SNR) level.