Intelligent Detection for RIS-Assisted MIMO Systems: A First-and-Second Momentum Approach

Abstract

Reconfigurable intelligent surface (RIS)-assisted massive multiple-input multiple-output (MIMO) is a potential technology for providing high-quality service in future communication systems. To further enhance system performance, in this letter, we propose a novel deep learning (DL) -based symbol detector, namely the DL-based first-and-second momentum detector (DFSM-Det). Specially, in each network layer, DFSM-Det utilizes the outputs from multiple previous layers to incorporate the mechanisms based on the double momentum to progressively refine symbol estimation layer by layer. Simulation results show that DFSM-Det achieves remarkably better performance than the existing DL-based detection schemes, particularly in highly loaded scenarios.