A fine-resolution frequency estimator using an arbitrary number of DFT coefficients

A method for the frequency estimation of complex exponential signals observed under additive white Gaussian noise is presented. Unlike competing methods based on relatively few Discrete Fourier Transform (DFT) samples, the presented technique can generate a frequency estimate by fusing the information from all DFT samples. The estimator is shown to follow the Cramer–Rao bound with a smaller signal-to-noise ratio (SNR) gap than the competing estimators at high SNR.