Maximum likelihood parameter estimation under impulsive conditions, a sub-Gaussian signal approach

In this paper we present an alternative to the Gaussian and Cauchy distributions for modeling stochastic signals. The proposed model has the same impulsiveness as the Cauchy density, but it is derived as a sub-Gaussian process, i.e., a variance mixture of Gaussian random variables.We proceed to use the derived model in the problem of signal parameter estimation through the use of multisensor data. Both the data and noise are assumed to be stochastic. The main problem of interest is the estimation of the DOA and statistics of the signal. A maximum likelihood algorithm is presented for the solution of this problem, and a pseudo-maximum-likelihood separable solution approach is derived. Finally, simulations are presented to demonstrate the robustness of the proposed algorithm.