Recent advances in rapid frequency estimation

There are two main types of frequency estimator-the traditional, based on FFT's, and the more modern, based on linear filtering. Statistical and computational efficiency usually have to be traded off against each other. Typically the filter-based estimators are statistically grossly inefficient. This paper discusses the extension of the Quinn-Fernandes (filter-based, statistically optimal) technique to complex-valued signals, and the use of zero-padding to improve the statistical efficiency of the Fourier transform interpolative (FTI) estimator to statistical near-optimality.