Sum-Rate Optimization for the MIMO IC under Imperfect CSI: a Deterministic Equivalent Approach
Abstract
A new method is proposed to determine precoding matrices that achieve local maxima of the expected sum rate in a multiple input multiple output interference channel (MIMO IC), in the realistic scenario where only partial channel state information (CSI) is available at the transmitters. Relying on a random matrix analysis of the capacity of large dimensional Ricean channels introduced in [1], the expected sum rate of the K-user MIMO IC is approximated by a deterministic equivalent to which an iterative gradient scheme is applied to find local maxima of the approximated sum rate.