Deep Reinforcement Learning-Based Robust ISAC in RIS-Assisted V2I Networks
IEEE Transactions on Cognitive Communications and Networking · 2026 · 12 · 9732-9748
DOI: 10.1109/tccn.2026.3709865 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Cognitive Communications and Networking, 2026. Verfasst von Lidong Ma und 2 weiteren Autor:innen. Die Arbeit nennt 41 Literaturangaben. Gefördert von: National Natural Science Foundation of China.
Abstract (Originalsprache)
Integrated sensing and communication (ISAC) empowered by reconfigurable intelligent surfaces (RIS) is a promising enabler for next-generation vehicle-to-infrastructure (V2I) systems. However, dynamic vehicular environments remain challenging due to high mobility, signaling overhead, and environmental uncertainty. To address these issues, we propose a robust ISAC framework for RIS-assisted V2I networks. A vehicular kinematic model and a radar cross-section (RCS) model provide physical priors for mobility and scattering, while an extended Kalman filter (EKF) exploits these priors with echo measurements to recursively predict vehicle states. In addition, a deep reinforcement learning (DRL) module is incorporated to adapt system behavior under rapidly varying conditions. Building on this architecture, we further develop a DRL-based tracking-aware configuration and optimization (DRL-TACO) algorithm, which enables cooperative EKF–DRL operation to refine state predictions and dynamically configure the RIS for real-time beamforming. Simulation results demonstrate that the proposed approach achieves superior robustness under stochastic EKF-input perturbations and scenario randomness, faster convergence, and higher tracking accuracy than conventional EKF- and optimization-based baselines, underscoring its potential for practical deployment in ISAC-enabled vehicular systems.
Förderung
- National Natural Science Foundation of China
Zitieren
APA
Ma, L., Xu, C., & Wen, M. (2026). Deep Reinforcement Learning-Based Robust ISAC in RIS-Assisted V2I Networks. IEEE Transactions on Cognitive Communications and Networking, 12, 9732-9748. https://doi.org/10.1109/tccn.2026.3709865
Vancouver
Ma L, Xu C, Wen M. Deep Reinforcement Learning-Based Robust ISAC in RIS-Assisted V2I Networks. IEEE Transactions on Cognitive Communications and Networking. 2026;12:9732-9748. doi:10.1109/tccn.2026.3709865
BibTeX
@article{Ma2026,
author = {Ma, Lidong and Xu, Chen and Wen, Miaowen},
title = {Deep Reinforcement Learning-Based Robust ISAC in RIS-Assisted V2I Networks},
journal = {IEEE Transactions on Cognitive Communications and Networking},
year = {2026},
volume = {12},
pages = {9732--9748},
doi = {10.1109/tccn.2026.3709865}
}
EUVIMED