Scholarlink

In millimeter wave (mm Wave) and massive multiple input multiple output (MIMO) systems, beam selection can enhance channel capacity and reduce bit error rate. However, existing beam selection methods for MIMO systems rely on traditional optimization techniques, which may not be feasible for real-time data transmission. Therefore, this paper proposes a vision-aided beam prediction method based on a three-dimensional convolutional neural network (3D CNN) and efficient channel attention (ECA) mechanism. First, 3D CNN is used to extract features from image data. Then, ECA assigns larger weights to critical image features, enhancing the representation capability of specific regions in the image. Finally, a multilayer perceptron (MLP) is employed to predict the optimal beam index. Experimental results on real-world data demonstrate that this method significantly improves prediction accuracy and stability compared to existing methods.

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

Vision-Assisted Beam Prediction with 3D Convolutional Network and Efficient Channel Attention Mechanism

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Shaohui Pan,
  • Zhuoran Cai,
  • Yu Wang

摘要

In millimeter wave (mm Wave) and massive multiple input multiple output (MIMO) systems, beam selection can enhance channel capacity and reduce bit error rate. However, existing beam selection methods for MIMO systems rely on traditional optimization techniques, which may not be feasible for real-time data transmission. Therefore, this paper proposes a vision-aided beam prediction method based on a three-dimensional convolutional neural network (3D CNN) and efficient channel attention (ECA) mechanism. First, 3D CNN is used to extract features from image data. Then, ECA assigns larger weights to critical image features, enhancing the representation capability of specific regions in the image. Finally, a multilayer perceptron (MLP) is employed to predict the optimal beam index. Experimental results on real-world data demonstrate that this method significantly improves prediction accuracy and stability compared to existing methods.