UDC 621.37
ESTIMATING THE FREQUENCIES OF POLYHARMONIC SIGNAL BY REDUCING THE ORDER OF A PARAMETRIC MODEL
M. B. Kagalenko, engineer, dept. of radiotechnical systems, RSREU, Ryazan; This email address is being protected from spambots. You need JavaScript enabled to view it.
We propose an algorithm for estimating multiple frequencies in a sampled signal. This algorithm first constructs a parametric model with excessive number of frequencies, and then performs the optimal reduction of its order. The aim of this work is to compute the accurate estimates of the frequencies using a limited number of signal’s samples. At the initial stage, we construct a parametric model of the signal that is a weight function on a unit circle. The poles of rational weight function are determined by the roots of a polynomial, the coefficients of which are given by a singular vector of the matrix constructed from signal’s samples. We use the least squares method to determine the weights corresponding to poles. At the second stage, we use Adamyan-Arov-Krein theorem to compute the optimal in the Hankel norm approximation to the model that has an order chosen by a user. We demonstrate using Monte-Carlo simulations that the algorithm has better stability in the presence of uncorrelated white noise than a previously published related algorithm
Key words: Wiener – Levinson algorithm, Adamyan – Arov – Krein theorem, Prony algorithm, leastsquares method, Pisarenko algorithm.