Uplink Precoding Design for Coherent Paired Users in Multi-cell MIMO Systems

This paper investigates the uplink precoding design in a multi-cell multiple-input multiple-output network. Unlike the existing cooperative schemes that only share channel state information among base stations, we schedule users into pairs, enabling the information symbols sharing by device-to-device transmission and cooperation in symbol precoding. Our work aims to promote uplink capability and reduce interference by optimizing the precoders. An optimization problem is formulated by maximizing the sum rate in the constraints of the maximum transmit powers of users. Since the problem is non-convex and difficult to solve, we employ the spherical reparameterization to convert the power-constrained problem into an unconstrained optimization and propose a weighted minimum mean square error iterative algorithm to optimize the precoders. The spherical-based scheme can improve the sum rate performance while taking a high complexity. Then, a complexity-friendly scheme is developed via the Lagrangian method, where auxiliary variables are introduced to obtain closed-form solutions rather than gradient searching in the spherical-based algorithm. Simulation results demonstrate that both the proposed spherical-based and Lagrange-based schemes improve the sum rate efficiently, outperforming the existing scheme based on singular value decomposition.