Scholarlink

With the official commercialization of 5G-Advanced (5G-A), its potential to enhance data transmission speed, reduce network latency, and expand device connectivity has laid a solid foundation for innovative applications such as the Internet of Things (IoT) and autonomous vehicles. Accurate estimation of User Equipment (UE) location has become increasingly important. However, the accuracy of traditional positioning methods is insufficient in complex network environments. While some Artificial Intelligence (AI) techniques have improved positioning accuracy, they are often data-dependent and have limited generalization capabilities. To address these issues, this study constructs a graph structure based on China Mobile’s 5G Measurement Report (MR) data to describe the relationships between UEs and cells and introduces Graph Attention Networks (GAT) for UE position estimation. GAT dynamically integrates features from different nodes through the attention mechanism, achieving more accurate and stable position estimation in dense 5G-A network environments. Experimental results show that this algorithm maintains high accuracy and robustness under complex conditions, significantly improving the estimation accuracy of UE positions.

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

Graph Attention Networks Based User Equipment Position Estimation in 5G-Advanced Networks

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Zhiyong Liu,
  • Tong Liang,
  • Xingwei Zhou,
  • Bowei Pu,
  • Linyu Li,
  • Su Wang,
  • Xu Yin,
  • Yuhao Liu,
  • Fengjun Zhao

摘要

With the official commercialization of 5G-Advanced (5G-A), its potential to enhance data transmission speed, reduce network latency, and expand device connectivity has laid a solid foundation for innovative applications such as the Internet of Things (IoT) and autonomous vehicles. Accurate estimation of User Equipment (UE) location has become increasingly important. However, the accuracy of traditional positioning methods is insufficient in complex network environments. While some Artificial Intelligence (AI) techniques have improved positioning accuracy, they are often data-dependent and have limited generalization capabilities. To address these issues, this study constructs a graph structure based on China Mobile’s 5G Measurement Report (MR) data to describe the relationships between UEs and cells and introduces Graph Attention Networks (GAT) for UE position estimation. GAT dynamically integrates features from different nodes through the attention mechanism, achieving more accurate and stable position estimation in dense 5G-A network environments. Experimental results show that this algorithm maintains high accuracy and robustness under complex conditions, significantly improving the estimation accuracy of UE positions.