Maximum likelihood estimation of superimposed complex sinusoids in white Gaussian noise by reduced effort coarse search (RECS)

An efficient implementation for the maximum likelihood estimator for the frequencies of superimposed complex sinusoids in white Gaussian noise is presented. The algorithm uses one-dimensional searches and classical optimization techniques and produces data that can be used for model order selection. Its convergence and statistical properties are verified through simulation. The method does not require polynomial rooting, frequent manipulation of large matrices, or the use of approximate constraints