LOW-COMPLEXITY SIGNAL DETECTION FOR 5G MASSIVE MIMO USING STEEPEST DESCENT AND SYMMETRIC GAUSS-SEIDEL

Abstract

MASSIVE multiple-input multiple-output (M-MIMO) is one of the most promising technologies to meet 5G requirements and networks beyond. By equipping a base station (BS) with a large number, claimed to be tens to hundreds, of antennas, several devices can be run simultaneously without significant inter-user interference. M-MIMO has several benefits over traditional MIMO, especially in terms of energy efficiency and spectral efficiency. There are several difficulties in implementing M-MIMO; one of them is signal detection at the uplink of M-MIMO. Linear signal detection algorithms, such as the minimum mean-square error (MMSE), can help achieve high efficiency but require complex matrix inversions with high computational complexity. Therefore, numerous detection algorithms have been suggested to approximately achieve the MMSE performance, but their convergence rate is slow. In this work, a new method based on Steepest Descent and Symmetric Gauss-Seidel (SDSGS) method is proposed to reduce the complexity of linear detection for uplink M-MIMO systems. To speed up the convergence, the Steepest Descent (SD) algorithm has been used to ensure a fast direction for the subsequent Gauss-Seidel (GS) iteration. In addition, an effective and computationally less-complex initial estimation is being used to achieve the desired performance with a smaller number of iterations. To further improve the performance, a new hybrid iteration is proposed. Compared to the conventional GS method, our proposed method improves efficiency and leads to faster convergence. We analyze the computational complexity of the various approaches and provide numerical simulations to verify the effectiveness of our purpose detector.