Adaptive single-tone frequency estimation based on autoregressive model

A novel adaptive filter algorithm based on the linear prediction property of sinusoidal signals is proposed for estimating the frequency of a real tone in white noise, assuming that the noise power is known a priori. Using the least mean square algorithm, the estimator is computationally efficient and it provides explicit frequency measurements on a sample-by-sample basis. Convergence behavior and variance of the estimated frequency are derived. Computer simulations are included to corroborate the theoretical analysis and to demonstrate the capability of the method of tracking time-varying frequencies.