Estimation of quasiperiodic signal parameters by means of dynamic signal models

The problem of estimating the parameters of a quasiperiodic signal consisting of a sum of a given number of sinusoidal signals with known frequencies but unknown time-varying amplitudes and phases superimposed by some additive noise sources is treated in this paper. Different estimation techniques that solve the problem are presented in this paper. The reason to do that is twofold: (1) to present the derivation of a new estimator for quasiperiodic signals, which is based on Kalman filtering theory and a random walk model, and to illustrate its structure and parametrization; and (2) to compare the new estimator with another Kalman filter-based estimator that uses an “oscillator model” and with the more classical running recursive discrete Fourier series method