Toward Reliable Semantic Communications: A Fast One-Step Channel-Adaptive Denoising Diffusion Method
IEEE Transactions on Networking · 2026 · 34 · 5911-5926
DOI: 10.1109/ton.2026.3706774 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Networking, 2026. Verfasst von Wei Du und 1 weiteren Autor:innen. Die Arbeit nennt 48 Literaturangaben. Gefördert von: Chinese Universities Scientific Fund, Key Research and Development Projects of Shaanxi Province, Fundamental Frontier Research Projects of Shaanxi Provincial Agriculture Department.
Abstract (Originalsprache)
Semantic communication (SC) aims to deliver the underlying meaning of raw information, in which deep-learning-based joint source-channel coding (DeepJSCC) is usually utilized to enable end-to-end communications. DeepJSCC-based SC frameworks perform well in wireless image transmission tasks, particularly under channel conditions with low signal-to-noise ratios (SNRs). However, existing methods still cannot adaptively eliminate different levels of noise, which limits the adaptability and effectiveness of DeepJSCC-based SC frameworks. To this end, we propose a one-step channel-adaptive denoising diffusion model in the SC framework (SC-CAD2M), which can learn denoised semantic features under uneven noise distributions and rapidly restore the original signal. Specifically, we design the noise-adaptive mask (NAM) module to adaptively calibrate semantic features across various SNR conditions. By bridging the forward diffusion process with the inevitable noise injection in wireless communications, we derive the optimal sampling step to accelerate signal denoising. Moreover, we incorporate the NAM and CAD2M modules into the DeepJSCC-based SC framework, which are capable of adaptively predicting and eliminating a wide range of noise levels from received semantic features. Extensive simulations on the STL-10 dataset demonstrate the superior performance and adaptability of the SC-CAD2M framework for image transmission tasks under varying SNR conditions and fading channels.
Förderung
- Chinese Universities Scientific Fund
- Key Research and Development Projects of Shaanxi Province
- Fundamental Frontier Research Projects of Shaanxi Provincial Agriculture Department
- Qinchuangyuan Innovation and Entrepreneurship Talent Project
Zitieren
APA
Du, W., & Yang, B. (2026). Toward Reliable Semantic Communications: A Fast One-Step Channel-Adaptive Denoising Diffusion Method. IEEE Transactions on Networking, 34, 5911-5926. https://doi.org/10.1109/ton.2026.3706774
Vancouver
Du W, Yang B. Toward Reliable Semantic Communications: A Fast One-Step Channel-Adaptive Denoising Diffusion Method. IEEE Transactions on Networking. 2026;34:5911-5926. doi:10.1109/ton.2026.3706774
BibTeX
@article{Du2026,
author = {Du, Wei and Yang, Bo},
title = {Toward Reliable Semantic Communications: A Fast One-Step Channel-Adaptive Denoising Diffusion Method},
journal = {IEEE Transactions on Networking},
year = {2026},
volume = {34},
pages = {5911--5926},
doi = {10.1109/ton.2026.3706774}
}
EUVIMED