Frequency estimation of sinusoids in colored noise

A maximum a posteriori estimator of complex sinusoids in colored Gaussian noise is proposed. The unknown noise covariance matrix and the complex amplitudes of the sinusoids are of no interest, and therefore are not estimated. The observed data vectors are decomposed into signal and noise components. From the probability density functions of these components, using Bayes´ rule, the marginal a posteriori density of the frequencies, and the proposed estimator are found. Computer simulations show excellent results.