EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

RASCC: Resolvability-Aware Coded Semantic Overlay With Multimodal Side Information for Intelligent Covert Communications

Rong Lu, Min Jia, Shiyao Meng, Yihua Liu, Qing Guo

IEEE Open Journal of the Communications Society · 2026 · 7 · 13645-13660

DOI: 10.1109/ojcoms.2026.3745243 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Open Journal of the Communications Society, 2026. Verfasst von Rong Lu und 4 weiteren Autor:innen. Die Arbeit nennt 45 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, Fundamental Research Funds for the Central Universities.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Artificial-intelligence-native next-generation networks make covert delivery of task-relevant information increasingly important because encryption alone cannot conceal the existence of sensitive traffic. Existing covert semantic schemes mainly regulate transmit power or optimize against limited detectors, leaving payload-dependent waveform structure, unequal short-packet reliability, and heterogeneous post-training wardens insufficiently addressed. This paper proposes the Resolvability-Aware Coded Semantic Overlay (RASCC), which embeds a six-bit caption-innovation index into an always-on public image-semantic transmission. Public-image side information removes predictable caption semantics; the residual index supports at most 64 conditional alternatives, not unrestricted sentence recovery. Sparse unequal error protection protects the innovation bits, while one-time sign randomization with half-normal amplitudes, keyed orthogonal mixing, and constant-energy geodesic composition produce a payload-independent pre-overlay Gaussian carrier and a final waveform with exact per-packet energy. SNR-adaptive budgets and primal–dual learning regulate residual discrepancies. A conditional spherical-law analysis separates the Gaussian moment term from the non-Gaussian remainder and supplies a fading-aware KL upper bound. MS-COCO experiments combine multi-method comparisons with an image-disjoint, public-information-aware audit. At 6 dB, the spherical budget reduces analytical-warden AUC from 0.6279–0.6393 to 0.5292–0.5362 across three channels, at a retrieval cost. In AWGN, paired-caption Recall@1 (R@1) is 0.1509 versus 0.1061 for an image-only prior. These results quantify a model-dependent security–retrieval tradeoff, not universal undetectability.

Förderung

  • National Natural Science Foundation of China
  • Fundamental Research Funds for the Central Universities

Zitieren

APA

Lu, R., Jia, M., Meng, S., Liu, Y., & Guo, Q. (2026). RASCC: Resolvability-Aware Coded Semantic Overlay With Multimodal Side Information for Intelligent Covert Communications. IEEE Open Journal of the Communications Society, 7, 13645-13660. https://doi.org/10.1109/ojcoms.2026.3745243

Vancouver

Lu R, Jia M, Meng S, Liu Y, Guo Q. RASCC: Resolvability-Aware Coded Semantic Overlay With Multimodal Side Information for Intelligent Covert Communications. IEEE Open Journal of the Communications Society. 2026;7:13645-13660. doi:10.1109/ojcoms.2026.3745243

BibTeX
@article{Lu2026,
  author = {Lu, Rong and Jia, Min and Meng, Shiyao and Liu, Yihua and Guo, Qing},
  title = {RASCC: Resolvability-Aware Coded Semantic Overlay With Multimodal Side Information for Intelligent Covert Communications},
  journal = {IEEE Open Journal of the Communications Society},
  year = {2026},
  volume = {7},
  pages = {13645--13660},
  doi = {10.1109/ojcoms.2026.3745243}
}
Kostenlos starten →