A two stage hybrid algorithm for estimation of non-stationary sinusoidal signal parameter

In this paper we consider the problem of estimation of amplitude, phase and frequency of a time-varying sinusoidal signal corrupted by noise. The frequency is estimated using Linear Predictor Approach and amplitude, phase is estimated using ADALINE approach. It is a two stage algorithm, estimated frequency of first stage is used to estimate amplitude and phase in second stage. Moreover, the learning parameter is tuned iteratively for faster convergence. From the proposed method a numerically robust and low complexity algorithm is derived for tracking the parameters of real sinusoid. Simulation results illustrate the good performance of the proposed method.