Autocorrelation-based algorithm for single-frequency estimation

A novel iterative algorithm for estimating the frequency of a single complex sinusoid in the complex white Gaussian noise is proposed. Inspired by the works of Fitz, L&R, and the ILP approaches, the algorithm is based on the repeated use of an autocorrelation-based frequency estimator. This approach is different from the iterative linear prediction (ILP) algorithm, yet produces very comparable performance. Like the ILP algorithm, the proposed estimator has the reduced threshold, and its performance is close to that of the ML estimation and uniform across the frequency range of -π-π to ππ. In addition, the proposed estimator demonstrates an asymptotic error variance of 0.14 dB above the CRB for N=1024N=1024 and 0.62 dB for N=48N=48, which is better than that of the ILP algorithm.