On the achievable sum rate for MIMO interference channels

In this correspondence, we study some information theoretical characteristics of vector Gaussian interference channels. Resorting to the superposition code technique, a lower bound of the sum capacity for the vector Gaussian interference channel is obtained. Alternatively, orthogonal transmission via frequency division multiplexing is considered and we establish the concavity of sum rate as the bandwidth allocation factor for the vector channel case. Numerical examples indicate that the achievable sum rate via the superposition code compares favorably with orthogonal transmission: the lower bound obtained via the superposition code dominates the best achievable sum rate through orthogonal transmission. This improvement holds for all interference power levels, a sharp contrast to that of the scalar counterpart.