LAU-Net: a lightweight attention U-Net framework for channel estimation in IRS-aided communications

View at publisher Summarise with VIRAT
Authors: Jaya Singh, Faisel Mushtaq, Kuldeep Singh, Ghanshyam Singh
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Accurate channel estimation remains a major challenge in intelligent reflecting surface (IRS)-aided wireless systems because of cascaded links and the large number of passive elements. To overcome this limitation, we propose a lightweight attention U-Net (LAU-Net) framework designed for IRS-assisted multi-user communications. The network combines lightweight convolutional blocks with attention gates that highlight informative channel features while suppressing noise. This design enables efficient feature fusion and produces high-precision channel recovery with reduced complexity. Simulation results demonstrate that LAU-Net consistently outperforms least squares (LS) and CNN-based deep residual networks (CDRN) over a broad range of SNR values and IRS configurations. These improvements make LAU-Net a practical and efficient solution for channel estimation in future 6G networks.

Fields of Study & Subjects

Advanced Wireless Communication Technologies Millimeter-Wave Propagation and Modeling Optical Wireless Communication Technologies

Publication details

Journal / source
International Conference on Optical and Wireless Technologies (OWT)
Pages
34–34
Publication date
30 March 2026
Language
EN