Scholarlink

<p>In 5G multi-frequency mobile communication systems, the trade-off between adaptability, resilience, and compatibility with existing infrastructure has been achieved with multi-input multi-output orthogonal frequency division multiplexing (MIMO-OFDM). However, in virtually all configurations of a MIMO-OFDM system, the major challenge remains finding an efficient algorithmic method for estimating the value of a sufficiently accurate channel between a transmitter and receiver; in fact, much of the performance depends on the accuracy of this estimation. This paper describes how deep learning and machine learning models provide insight into channel estimation. The proposed hybrid compressive sensing framework for estimating sparse channels combines the principles of BSAMP with the concepts of deep learning or ML models. To enhance efficiency, a DNN will predict the sparsity pattern of the channel and subsequently RL will tune the BSAMP algorithm parameters. The proposed hybrid can significantly improve the accuracy of the overall systems, especially for complex time-varying communication channel patterns at massive MIMO and improve the speed of recovering channel state information.The simulation results exhibited that DNN-BSAMP technique performs better in terms of bit error rate and means square error, throughput over conventional methods BSAMP, SAMP, OMP, and LS. The overall design and simulation had been executed in MATLAB.</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

A hybrid model-driven deep learning and block-sparse adaptive matching pursuit framework for robust 5G massive MIMO channel estimation

  • 期刊论文

  • 发表日期：

    2026年8月3日

查看全文
  • Sivangi Ravikanth,
  • Rallabandi Revathi,
  • P. Kiran Kumar,
  • Jillella Venkateswara Rao,
  • Sakam Nagi Reddy,
  • Srinivasa Reddy Kurukuntla

摘要

In 5G multi-frequency mobile communication systems, the trade-off between adaptability, resilience, and compatibility with existing infrastructure has been achieved with multi-input multi-output orthogonal frequency division multiplexing (MIMO-OFDM). However, in virtually all configurations of a MIMO-OFDM system, the major challenge remains finding an efficient algorithmic method for estimating the value of a sufficiently accurate channel between a transmitter and receiver; in fact, much of the performance depends on the accuracy of this estimation. This paper describes how deep learning and machine learning models provide insight into channel estimation. The proposed hybrid compressive sensing framework for estimating sparse channels combines the principles of BSAMP with the concepts of deep learning or ML models. To enhance efficiency, a DNN will predict the sparsity pattern of the channel and subsequently RL will tune the BSAMP algorithm parameters. The proposed hybrid can significantly improve the accuracy of the overall systems, especially for complex time-varying communication channel patterns at massive MIMO and improve the speed of recovering channel state information.The simulation results exhibited that DNN-BSAMP technique performs better in terms of bit error rate and means square error, throughput over conventional methods BSAMP, SAMP, OMP, and LS. The overall design and simulation had been executed in MATLAB.