Large N Analysis of Amplify-and-Forward MIMO Relay Channels with Correlated Rayleigh Fading

In this correspondence the cumulants of the mutual information of the flat Rayleigh fading amplify-and-forward MIMO relay channel without direct link between source and destination are derived in the large array limit. The analysis is based on the replica trick and covers both spatially independent and correlated fading in the first and the second hop, while beamforming at all terminals is restricted to deterministic weight matrices. Expressions for mean and variance of the mutual information are obtained. Their parameters are determined by a nonlinear equation system. All higher cumulants are shown to vanish as the number of antennas n goes to infinity. In conclusion the distribution of the mutual information I becomes Gaussian in the large n limit and is completely characterized by the expressions obtained for mean and variance of I . Comparisons with simulation results show that the asymptotic results serve as excellent approximations for systems with only few antennas at each node. The derivation of the results follows the technique formalized by Moustakas et al. in [1]. Although the evaluations are more involved for the MIMO relay channel compared to point-to-point MIMO channels, the structure of the results is surprisingly simple again. In particular an elegant formula for the mean of the mutual information is obtained, i.e., the ergodic capacity of the two-hop amplify-and-forward MIMO relay channel without direct link. Index Terms MIMO relay channel, amplify-and-forward, replica analysis, random matrix theory, large antenna number limit, cumulants of mutual information, correlated channels.