ML (Maximum Likelihood) Method for Frequency Estimation of Real Sinusoids in Noise

摘要

The DFT-based maximum likelihood method for frequency estimation of complex sinusoids as proposed in two previous articles is extended to treat the case of real data. Improvements on estimation precision and computation efficiency are obtained by imposing an equal-frequency constraint on the pair of complex sinusoids which corresponds to a real sinusoid and by utilizing a single-step interpolation in conjunction with a coarse finite search over the DFT data for the maximum of the likelihood function. Simulation results demonstrate these improvements. (Author)