Learning-Based Optimal Detector for Uplink Multiuser Massive MIMO Systems with One-Bit ADCs

Massive Multi-Input Multi-Output (MIMO) technology has emerged as a crucial advancement in wireless communication systems, promising improved reliability, higher transmission rates, and reduced power consumption. This paper explores the utilization of low-resolution Analog-to-Digital Converters (ADCs) in conjunction with massive MIMO systems to conserve energy and mitigate costs associated with high-resolution ADCs. However, employing low-resolution ADCs introduces challenges in accurately estimating transmitted signals at the receiver. We considered a one-bit model for our proposed method and solved a relaxed Maximum Likelihood (ML) problem, initially encountering with relatively high detection error. To reduce the error, the Recursive Least Squares (RLS) method is used to estimate the error, which is added to the initial detection using a trainable parameter, $\alpha$, derived from a neural network and improve estimation accuracy. Simulation results demonstrate significant improvements in Symbol Error Rate (SER) compared to existing methods, showcasing the effectiveness and feasibility of our proposed approach for one-bit massive MIMO systems.