On a Mutual Information and a Capacity Bound Gap of Pilot-Aided MIMO Channels
The ergodic mutual information of MIMO channels in the case where the receiver has only partial knowledge of the fading coefficients is difficult to compute, let alone the capacity. Therefore, previous work has essentially focused on handling lower and upper bounds which are relevant for approaching the region of achievable rates. The authors use the bounds derived originally. The main idea consists in treating the additional noise induced by the channel estimation error as if it were independent Gaussian noise. The resulting bounds have already been used extensively, e.g., for devising precoding and training strategies that regard the mutual information lower bound as a utility to be maximized and extensions to multiple user scenarios have been considered.