Scholarlink

<p>Terahertz waves and extremely large antenna arrays enable future 6G wireless communications in the near-field field. In the context of the extremely large-scale reconfigurable intelligent surface (XL-RIS) assisted wireless communication system, the pilot overhead for channel estimation becomes very large and channel modeling becomes significantly more complex. In this paper, we consider modeling the cascaded channel as a combination of hybrid-field and far-field channels. To decrease pilot overhead, we design a novel channel grouping estimation method and propose a super-resolution channel reconstruction strategy. To effectively recover the channel, we construct a dual-scale super-resolution network (DSRN), which can reconstruct the channel using channel features of different scales. Moreover, the DSRN can denoise in small-scale features, learn the sparse characteristics of the channel in large-scale features, and reverse the channel grouping process to convert small-scale features into large-scale features. The simulation results reveal that the suggested channel estimation scheme outperformed existing schemes.</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

Channel estimation for XL-RIS-aided Terahertz systems with dual-scale super-resolution network

  • 期刊论文

  • 发表日期：

    2026年6月30日

查看全文
  • Shitong Cheng,
  • Ziyan Liu,
  • Lihui Zhang,
  • Shenhua Zhao,
  • Banghai He

摘要

Terahertz waves and extremely large antenna arrays enable future 6G wireless communications in the near-field field. In the context of the extremely large-scale reconfigurable intelligent surface (XL-RIS) assisted wireless communication system, the pilot overhead for channel estimation becomes very large and channel modeling becomes significantly more complex. In this paper, we consider modeling the cascaded channel as a combination of hybrid-field and far-field channels. To decrease pilot overhead, we design a novel channel grouping estimation method and propose a super-resolution channel reconstruction strategy. To effectively recover the channel, we construct a dual-scale super-resolution network (DSRN), which can reconstruct the channel using channel features of different scales. Moreover, the DSRN can denoise in small-scale features, learn the sparse characteristics of the channel in large-scale features, and reverse the channel grouping process to convert small-scale features into large-scale features. The simulation results reveal that the suggested channel estimation scheme outperformed existing schemes.