Maximum-Likelihood Direction Finding Under Elliptical Noise Using the EM Algorithm

Unlike subspace-based solutions of direction-of-arrival (DOA) estimation under non-Gaussian noise, where the only optional difference with the Gaussian case is the scatter/covariance matrix estimation method, maximum-likelihood (ML)-based DOA solutions need a different treatment under the non-Gaussianity assumption. In this letter, we derive a particular ML-based DOA solution, called the expectation-maximization (EM) estimator, under the wide class of complex elliptically symmetric (CES) distributions.