Estimators for instantaneous phase

Abstract

Mean-squared error versus signal-to-noise ratio (SNR) of four estimators of instantaneous phase are compared. The simulator produces a noisy waveform with noise-free signal having zero instantaneous phase. The zero-phase assumption is without loss of generality when reference waveforms are known a priori. Estimators one and two are implemented using the conventional instantaneous phase and frequency integration techniques without regard to noise on the signal. Estimators three and four are adaptive estimators which utilize an approximate Kalman filter and a measure of instantaneous envelope. The adaptive estimators are found to be more accurate by at least an order of magnitude with the Kalman method being the best.