Noise-Aware Model Inversion Eavesdropping Attack in Semantic Communication System for Video Transmission
IEEE Transactions on Communications · 2026 · 74 · 12705-12719
DOI: 10.1109/tcomm.2026.3721643 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Communications, 2026. Verfasst von Xin Wang und 9 weiteren Autor:innen. Die Arbeit nennt 48 Literaturangaben. Gefördert von: National Natural Science Foundation of China.
Abstract (Originalsprache)
Semantic communication systems have emerged as a promising alternative to conventional bit-level transmission due to their superior error tolerance and efficiency. However, this paradigm is vulnerable to model inversion attacks in which adversaries can recover source data from intercepted semantic signals. While existing research has focused on image transmission, the security of video semantic communication over time-varying channels remains critically under-examined. To address this gap, this paper introduces the Noise-Aware Eavesdropping Attack (NAEA), a framework tailored for video semantic communication. NAEA leverages a dedicated noise classifier to identify channel conditions, which in turn guides an inverse network with a noise attention mechanism to accurately reconstruct video content. Furthermore, to counteract this threat, we propose a steganography-based defense to safeguard semantic features. Experimental results demonstrate that NAEA effectively compromises video semantic communication systems under realistic channel conditions, while our proposed defense proves highly effective in mitigating this threat.
Förderung
- National Natural Science Foundation of China
Zitieren
APA
Wang, X., Liu, X., Zhang, J., Gulomsafdarov, A., Xie, X., Zhang, K., Cai, Q., Liu, J., Wang, K., & Li, K. (2026). Noise-Aware Model Inversion Eavesdropping Attack in Semantic Communication System for Video Transmission. IEEE Transactions on Communications, 74, 12705-12719. https://doi.org/10.1109/tcomm.2026.3721643
Vancouver
Wang X, Liu X, Zhang J, Gulomsafdarov A, Xie X, Zhang K, et al. Noise-Aware Model Inversion Eavesdropping Attack in Semantic Communication System for Video Transmission. IEEE Transactions on Communications. 2026;74:12705-12719. doi:10.1109/tcomm.2026.3721643
BibTeX
@article{Wang2026,
author = {Wang, Xin and Liu, Xiulong and Zhang, Jiuwu and Gulomsafdarov, Abdulnazar and Xie, Xin and Zhang, Kaixuan and Cai, Qixuan and Liu, Jiao and Wang, Ke and Li, Keqiu},
title = {Noise-Aware Model Inversion Eavesdropping Attack in Semantic Communication System for Video Transmission},
journal = {IEEE Transactions on Communications},
year = {2026},
volume = {74},
pages = {12705--12719},
doi = {10.1109/tcomm.2026.3721643}
}
EUVIMED