OPTIMUM ARRAY DETECTION OF WEAK SIGNALS IN SPATIALLY CORRELATED NON-GAUSSIAN NOISE

The structure of the locally optimum array detector for a random signal embedded in non-Gaussian noise is derived. In this observation model the noise samples are assumed to be temporally independent at each array sensor, but correlated across the array at each time instant. Since the synthesis of the detector involves multivariate densities of the noise spatial samples, the transformation noise model, which is characterized only by the univariate probability density function and the covariance matrix, is considered. The performance is assessed via computer simulations by assuming for the first-order statistical properties of the noise the two-parameter generalized Gaussian family of distributions. A comparison with the performance of the locally optimum array detector synthesized for spatially independent noise is stated.