Beamforming Design for Dynamic Metasurface Antennas-Based Massive Multiuser MISO Downlink Systems

Abstract

This study addresses a massive multiuser multiple-input single-output downlink system with a base station equipped with dynamic metasurface antennas (DMAs) serving single-antenna users. Our goal is to jointly optimize the transmit precoder and configurable DMA coefficients to maximize system performance. This task is challenging due to the complex interdependencies between these parameters. The established solution, AO-RMO, employs alternating optimization (AO) to iteratively optimize the transmit precoder using the weighted minimum mean squared error algorithm and Riemannian manifold optimization (RMO) for the DMA coefficients. While AO-RMO performs reasonably well, it exhibits high time and computational complexity due to RMO. To address this issue, we propose a computationally efficient alternative: a projected gradient descent (PGD)-based algorithm for determining the DMA coefficients. Additionally, for practical reconfiguration of DMA coefficients with discrete values, we introduce the cross-entropy optimization (CEO)-based algorithm. Simulation results demonstrate that when continuous DMA coefficients are used, our proposed AO-PGD algorithm achieves nearly comparable weighted sum rate performance to the AO-RMO algorithm but with significantly reduced computational complexity and running time. When discrete DMA coefficients are applied, the proposed AO-CEO algorithm surpasses other benchmark algorithms in various system configurations.

Keywords

• Alternating optimization (AO)
• dynamic metasurface antenna (DMA)
• massive multiuser multiple-input single-output (MU-MISO)
• precoder