Scholarlink

<p>In this paper, we propose an enhanced orthogonal approximate message passing network (OAMP-Net) algorithm, called intelligent detection with sparse computations (IDSC), to significantly reduce the computational complexity of OAMP-Net for massive multiple-input multiple-output (MIMO) systems. While OAMP-Net delivers excellent detection performance, its practical deployment is limited by the high computational cost arising from frequent matrix inversions. To overcome this challenge, the proposed IDSC algorithm adopts a matrix approximation technique to reduce both the dimension and frequency of matrix inversions, effectively eliminating computationally intensive operations. Simulation results and complexity analysis demonstrate that IDSC achieves detection performance comparable to that of OAMP-Net, while greatly reducing computational overhead. To further validate its practical feasibility, a corresponding VLSI architecture is developed and implemented using the TSMC 90-nm CMOS process.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Intelligent Detection with Sparse Computations for Massive MIMO Systems

  • 期刊论文

  • 发表日期：

    2026年1月26日

查看全文
  • Mao-Hsu Yen,
  • Hoang-Yang Lu,
  • Hsin Chen,
  • Chia-Hsun Li

摘要

In this paper, we propose an enhanced orthogonal approximate message passing network (OAMP-Net) algorithm, called intelligent detection with sparse computations (IDSC), to significantly reduce the computational complexity of OAMP-Net for massive multiple-input multiple-output (MIMO) systems. While OAMP-Net delivers excellent detection performance, its practical deployment is limited by the high computational cost arising from frequent matrix inversions. To overcome this challenge, the proposed IDSC algorithm adopts a matrix approximation technique to reduce both the dimension and frequency of matrix inversions, effectively eliminating computationally intensive operations. Simulation results and complexity analysis demonstrate that IDSC achieves detection performance comparable to that of OAMP-Net, while greatly reducing computational overhead. To further validate its practical feasibility, a corresponding VLSI architecture is developed and implemented using the TSMC 90-nm CMOS process.