Statistical eigenvalue transmission over MIMO relay channels

The paper considers the multiple-access two-hop relay network, due to more practical environment, the source nodes only know the statistical channel state information. We adopt a general jointly-correlated MIMO channel model, which does not require separable spatial correlations at the source nodes and relay node. Based on this model, we firstly derive the closed-form tight upper bound for the ergodic capacity through the matrix permanents. Then we employ the convex optimization techniques and a simple iterative water-filling algorithm to obtain the a low complexity power allocation solution. Simulation displays the upper bound is tight and results are in agreement with the derivation.