A novel autocorrelation approach for frequency estimation of a sinusoid

Abstract

Based on the linear prediction (LP) property of sinusoidal signals, a novel autocorrelation approach for a single-tone sinusoidal signal emerged in Gaussian white noise is proposed. Firstly, a new closed-form autocorrelation expression is derived that it contributes more to frequency estimation than conventional autocorrelation expressions, then, inspired by RIM estimator, we propose a novel estimator using multiple autocorrelation lags. Computer simulations show that the accuracy of the proposed approach is closer to the Cramer-Rao bound(CRB) especially for lower signal noise radio(SNR) and has lower computation complexity via comparing with several autocorrelation-based estimators.