Precoder Design for User-Centric Network Massive MIMO: A Symplectic Optimization Approach

TL;DR

This paper introduces a symplectic optimization-based precoder design for user-centric massive MIMO systems, reducing computational complexity and outperforming traditional methods like WMMSE.

Contribution

It develops a novel precoder design method using symplectic optimization to avoid matrix inversion in UCN massive MIMO systems.

Findings

Outperforms WMMSE precoder in simulations

Reduces computational complexity of precoder design

Simplifies implementation by reducing precoder dimension

Abstract

In this paper, we utilize symplectic optimization to design a precoder for user-centric network (UCN) massive multiple-input multiple-output (MIMO) systems, where a subset of base stations (BSs) serves each user terminal (UT) instead of using all BSs. In UCN massive MIMO systems, the dimension of the precoders is reduced compared to conventional network massive MIMO. It simplifies the implementation of precoders in practical systems. However, the matrix inversion in traditional linear precoders still requires high computational complexity. To avoid the matrix inversion, we employ the symplectic optimization framework, where optimization problems are solved based on dissipative Hamiltonian dynamical systems. To better fit symplectic optimization, we transform the received model into the real field and reformulate the weighted sum-rate (WSR) maximization problem. The objective function of the optimization problem is viewed as the potential energy of the dynamical system. Due to energy dissipation, the continuous dynamical system always converges to a state with minimal potential energy. By discretizing the continuous system while preserving the symplectic structure, we obtain an iterative method for the precoder design. The complexity analysis of the proposed symplectic method is also provided to show its high computational efficiency. Simulation results demonstrate that the proposed precoder design based on symplectic optimization outperforms the weighted minimum mean-square error (WMMSE) precoder in the UCN massive MIMO system.