EUVIMEDEUVIMED entdecken
Crossref

Agentic AI-Driven Autonomous Covert Communication in MISO-NOMA VLC Systems With Imperfect CSI

Yangbo Guo, Yuqing Zhou, Baofang Chang, Ruihong Jiang, Tuanwei Tian, Lei Zhang, Hao Deng

IEEE Transactions on Cognitive Communications and Networking · 2026 · 12 · 9218-9230

DOI: 10.1109/tccn.2026.3705848 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Cognitive Communications and Networking, 2026. Verfasst von Yangbo Guo und 6 weiteren Autor:innen. Die Arbeit nennt 43 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Key R&D Projects in Henan Province, First Batch of Leading Projects for AI + Education of Henan Provincial Department of Education.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This paper proposes an agentic artificial intelligence (AI) framework for autonomous covert communication in multiple-input single-output non-orthogonal multiple access (MISO-NOMA) visible light communication (VLC) systems. We conceptualize the transmitter as an autonomous Alice-agent that learns to establish covert links with a legitimate user (Bob) while remaining undetectable to an adversarial eavesdropper (Willie), despite lacking prior knowledge of the eavesdropper’s exact channel state. To maintain stealth, the agent dynamically leverages artificial noise (AN) and public user signals as covers, making real-time decisions on power allocation and beamforming under imperfect channel state information. The problem is formulated as a partially observable Markov decision process (POMDP), and a novel transformer-based dual-clip proximal policy optimization (TDC-PPO) algorithm is introduced as the agent’s decision-making engine. Specifically, the TDC-PPO agent utilizes a self-attention mechanism to enhance state representation and a dual-clipping strategy to ensure policy stability under adversarial conditions. Simulation results validate that the proposed agentic AI framework consistently achieves higher covert transmission rates than baseline deep reinforcement learning (DRL) methods while exhibiting superior adaptability and convergence across varying covertness constraints and quality-of-service requirements. This work validates the effectiveness of autonomous, learning-driven agents in governing complex, constraint-satisfying behaviors in secure wireless communications.

Förderung

  • National Natural Science Foundation of China
  • Key R&D Projects in Henan Province
  • First Batch of Leading Projects for AI + Education of Henan Provincial Department of Education
  • Science and Technology Research Project of Henan Province

Zitieren

APA

Guo, Y., Zhou, Y., Chang, B., Jiang, R., Tian, T., Zhang, L., & Deng, H. (2026). Agentic AI-Driven Autonomous Covert Communication in MISO-NOMA VLC Systems With Imperfect CSI. IEEE Transactions on Cognitive Communications and Networking, 12, 9218-9230. https://doi.org/10.1109/tccn.2026.3705848

Vancouver

Guo Y, Zhou Y, Chang B, Jiang R, Tian T, Zhang L, et al. Agentic AI-Driven Autonomous Covert Communication in MISO-NOMA VLC Systems With Imperfect CSI. IEEE Transactions on Cognitive Communications and Networking. 2026;12:9218-9230. doi:10.1109/tccn.2026.3705848

BibTeX
@article{Guo2026,
  author = {Guo, Yangbo and Zhou, Yuqing and Chang, Baofang and Jiang, Ruihong and Tian, Tuanwei and Zhang, Lei and Deng, Hao},
  title = {Agentic AI-Driven Autonomous Covert Communication in MISO-NOMA VLC Systems With Imperfect CSI},
  journal = {IEEE Transactions on Cognitive Communications and Networking},
  year = {2026},
  volume = {12},
  pages = {9218--9230},
  doi = {10.1109/tccn.2026.3705848}
}
Kostenlos starten →