Instantaneous phase shift estimation methods

Abstract

The paper presents novel simulation results from testing several different instantaneous phase shift estimation methods. The authors have optimised the methods' parameters and have designed the numerical algorithms capable of tracking the time-varying phase shift between two signals changing in the range /spl plusmn/180/spl deg/ with frequency of up to 2 kHz. The required estimation error has been assumed to be no greater then /spl plusmn/0.1/spl deg/. Special attention has been paid to improvement of the methods' resistance to noise. The following algorithms have been compared: 1) analytic signal method without and with smoothing post-filters, 2) short-time Fourier transform method 3) short-time Wigner transform method, and 4) (N)LMS adaptive correlator. Standard low-pass FIR filters as well as adaptive (N)LMS and Kalman linear prediction filters have been used as post-filters in the analytic signal method.