Side-Information-Aided Deep Unfolding Network for Hybrid-Field Channel Estimation of Wideband Terahertz Ultra-Massive MIMO Systems
IEEE Transactions on Communications · 2026 · 74 · 10889-10905
DOI: 10.1109/tcomm.2026.3708500 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Communications, 2026. Verfasst von Sijia Wu und 3 weiteren Autor:innen. Die Arbeit nennt 52 Literaturangaben. Gefördert von: Natural Science Foundation of Jilin Province, National Natural Science Foundation of China.
Abstract (Originalsprache)
Terahertz (THz) ultra-massive multiple-input multiple-output (UM-MIMO) systems are promising to fulfill the increasing demands of the sixth generation (6G) wireless communications. However, accurate channel estimation in the THz band remains a significant challenge due to the inherent characteristics of hybrid near-field and far-field propagations. Additionally, the performance of traditional compressed sensing (CS)-based channel estimation methods degrades considerably in the UM-MIMO case. In this work, a side-information-aided dynamic threshold network (SIDT-Net) is proposed for channel estimation in wideband THz UM-MIMO system, which implements the high-dimensional sparse reconstruction with limited pilot overhead by integrating side information and learnable threshold with CS-based deep unfolding networks. The application of side information allows the SIDT-Net to send large volumes of information between adjacent phases, substantially augmenting the estimation accuracy and network performance. The SIDT-Net also employs the multi-scale feature extraction method and residual-aided attention mechanism to adaptively emphasize informative features. Simulation results indicate that the proposed SIDT-Net achieves higher estimation accuracy than existing CS methods and deep learning algorithms, with an average normalized mean square error (NMSE) reaching -20.5 dB when the signal-to-noise ratio (SNR) is 10 dB. It also provides a good trade-off between accuracy and complexity.
Förderung
- Natural Science Foundation of Jilin Province
- National Natural Science Foundation of China
Zitieren
APA
Wu, S., Jiang, H., Zheng, H., & Qu, L. (2026). Side-Information-Aided Deep Unfolding Network for Hybrid-Field Channel Estimation of Wideband Terahertz Ultra-Massive MIMO Systems. IEEE Transactions on Communications, 74, 10889-10905. https://doi.org/10.1109/tcomm.2026.3708500
Vancouver
Wu S, Jiang H, Zheng H, Qu L. Side-Information-Aided Deep Unfolding Network for Hybrid-Field Channel Estimation of Wideband Terahertz Ultra-Massive MIMO Systems. IEEE Transactions on Communications. 2026;74:10889-10905. doi:10.1109/tcomm.2026.3708500
BibTeX
@article{Wu2026,
author = {Wu, Sijia and Jiang, Hong and Zheng, Huiming and Qu, Liangdong},
title = {Side-Information-Aided Deep Unfolding Network for Hybrid-Field Channel Estimation of Wideband Terahertz Ultra-Massive MIMO Systems},
journal = {IEEE Transactions on Communications},
year = {2026},
volume = {74},
pages = {10889--10905},
doi = {10.1109/tcomm.2026.3708500}
}
EUVIMED