Gradient-Based Solution for Hybrid Precoding in MIMO Systems

Abstract

The combination of baseband and analog precoding for multiple-input multiple-output (MIMO) systems is considered in this paper which is referred to as hybrid precoding. The system capacity, as a design criterion, is maximized subject to unit modulus constraints on the elements of the analog precoder (phase shifters), and a total power constraint. This is a non-convex problem due to the product of the analog and baseband precoder variables. The proposed technique suggests computing non-trivial complex derivatives of the objective and constraints, analytically, in order to develop an iterative gradient-based sequential optimization algorithm to solve the non-convex problem. Promising simulation results show that the solution of the proposed algorithm is sufficiently close to the optimal (full-baseband) precoder solution, regardless of the channel characteristics.