Low-Resolution MMSE Equalization for Massive MU-MIMO Systems

研究成果: Article同行評審

2 引文 斯高帕斯(Scopus)

摘要

Finite-alphabet equalization that represents a spatial equalization matrix with low-resolution coefficients is a promising technique to reduce the power consumption, processing delay, and circuit area of baseband processing in the context of all-digital massive multiuser multiple-input–multiple-output uplink systems. However, to minimize the performance loss caused by a coarse-resolution spatial equalization matrix, its coefficients must be carefully designed in the minimum mean-square error sense to achieve the desired bit error rate (BER) performance, which unfortunately constitutes an NP-hard optimization problem. To tackle this problem, we first reformulate the finite-alphabet equalization design problem as an unconstrained optimization on a smooth Riemannian manifold. Then we propose an algorithm based on Riemannian manifold optimization (RMO) to solve the reformulated problem. Based on simulation results, the proposed 2-bit RMO-assisted equalizer outperforms its state-of-the-art counterparts while maintaining the same asymptotic complexity. In addition, the proposed 2-bit RMO-assisted equalizer exhibits a loss of only 1.47 dB at a BER of 10−4 compared to an unquantized linear minimum mean squared error equalizer.

原文English
頁(從 - 到)8164-8169
頁數6
期刊IEEE Transactions on Vehicular Technology
72
發行號6
DOIs
出版狀態Published - 2023 6月 1

All Science Journal Classification (ASJC) codes

  • 航空工程
  • 電氣與電子工程
  • 電腦網路與通信
  • 汽車工程

指紋

深入研究「Low-Resolution MMSE Equalization for Massive MU-MIMO Systems」主題。共同形成了獨特的指紋。

引用此