Low-Complexity Weighted Sum-Rate Maximization of MIMO MAC with Finite-Alphabet Inputs

This paper investigates the linear precoding design for multiple-input multiple-output (MIMO) multiple access channel (MAC) with finite-alphabet inputs. Since the mutual information (MI) and its gradient-minimum mean-square error (MMSE) matrix lack closed-form solutions, traditional MI/MMSE based precoding scheme suffers from a prohibitive burden of computation. To address this problem, we first derive an asymptotically optimal lower bound of MI, which is more computationally efficient. Based on the lower bound, we derive a suboptimal precoding scheme that maximizes the weighted sum-rate (WSR) of MIMO MAC with finite-alphabet inputs. Furthermore, a low-complexity precoding algorithm is proposed to obtain the numerical results of precoding matrices. Numerical results confirm a higher efficiency of the proposed precoding scheme compared with traditional MI/MMSE based precoding strategies.