Maximum ratio combining with arbitrary correlated generalized Ricean branches

We study maximum ratio combining with correlated generalized Ricean branch fadings. The L received signals are represented as vectors in a 2m-dimensional signal space where m is the fading severity, and 2m is assumed to he an integer. Then, the channel model is a non-zero mean L/spl times/2m Gaussian random matrix. We discuss the optimality of maximum ratio combining when the branch fadings are correlated. A novel derivation of the multivariate moment generating function of noncentral chi-square distributed and correlated random variables is presented. The average bit-error performance evaluation is illustrated for constant, exponential and Gaussian branch signal-to-noise ratio correlation models.