Scholarlink

<p>Deep learning (DL), a branch of artificial intelligence (AI), has established important success across various fields like speech recognition, image classification, and natural language processing. Motivated by this success, recent years have seen a growing interest in applying DL to wireless communication problems—particularly channel estimation and channel state information (CSI) feedback for next-generation (6G) systems. In contrast to the rule-based approaches, DL-based approaches are learning-based approaches that are based on inductive reasoning, which bring a new set of challenges, such as model selection, network architecture design, and training data acquisition. In this paper, the following three challenges are discussed, and a novel DL-based approach for wireless channel estimation and CSI feedback based on a Multi-Head Self-Attention Gated-Dilated Convolutional Neural Network (MSGCNN-6G) is proposed. The proposed architecture is designed to capture the local and global channel dependency effectively for robust estimation in a complex and dynamic wireless environment. We conduct several case studies and numerical experiments to evaluate the proposed method. The results demonstrate the superiority of MHSA-GDCNN in terms of estimation accuracy and feedback efficiency, highlighting its potential for 6G wireless communication systems. Then, the proposed MSGCNN-6G is implemented, and the performance metrics like Detection Success Rate, MSE, and NMSE are analyzed.</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

Advanced Channel Estimation and CSI Feedback for 6G Communication Systems Using a Multi-Head Self-Attention Gated-Dilated Convolutional Neural Network

  • 期刊论文

  • 发表日期：

    2026年7月11日

查看全文
  • S. Parvathi,
  • B. Kiruthika,
  • M. Sathesh,
  • Bhuvaneswari Balachander,
  • M. Raja

摘要

Deep learning (DL), a branch of artificial intelligence (AI), has established important success across various fields like speech recognition, image classification, and natural language processing. Motivated by this success, recent years have seen a growing interest in applying DL to wireless communication problems—particularly channel estimation and channel state information (CSI) feedback for next-generation (6G) systems. In contrast to the rule-based approaches, DL-based approaches are learning-based approaches that are based on inductive reasoning, which bring a new set of challenges, such as model selection, network architecture design, and training data acquisition. In this paper, the following three challenges are discussed, and a novel DL-based approach for wireless channel estimation and CSI feedback based on a Multi-Head Self-Attention Gated-Dilated Convolutional Neural Network (MSGCNN-6G) is proposed. The proposed architecture is designed to capture the local and global channel dependency effectively for robust estimation in a complex and dynamic wireless environment. We conduct several case studies and numerical experiments to evaluate the proposed method. The results demonstrate the superiority of MHSA-GDCNN in terms of estimation accuracy and feedback efficiency, highlighting its potential for 6G wireless communication systems. Then, the proposed MSGCNN-6G is implemented, and the performance metrics like Detection Success Rate, MSE, and NMSE are analyzed.