Title:In-Memory Massive MIMO Linear Detector Circuit with Extremely High Energy Efficiency and Strong Memristive Conductance Deviation Robustness

Abstract:The memristive crossbar array (MCA) has been successfully applied to accelerate matrix computations of signal detection in massive multiple-input multiple-output (MIMO) systems. However, the unique property of massive MIMO channel matrix makes the detection performance of existing MCA-based detectors sensitive to conductance deviations of memristive devices, and the conductance deviations are difficult to be avoided. In this paper, we propose an MCA-based detector circuit, which is robust to conductance deviations, to compute massive MIMO zero forcing and minimum mean-square error algorithms. The proposed detector circuit comprises an MCA-based matrix computing module, utilized for processing the small-scale fading coefficient matrix, and amplifier circuits based on operational amplifiers (OAs), utilized for processing the large-scale fading coefficient matrix. We investigate the impacts of the open-loop gain of OAs, conductance mapping scheme, and conductance deviation level on detection performance and demonstrate the performance superiority of the proposed detector circuit over the conventional MCA-based detector circuit. The energy efficiency of the proposed detector circuit surpasses that of a traditional digital processor by several tens to several hundreds of times.