Reinforcement Learning For Secure and Efficient Next-Generation Wireless Communications: Exploring RIS-Assisted Network Architectures

Abstract

The wireless communications landscape of beyond 5G and 6G systems in dynamic and dense smart city environments presents multiple interference challenges. UAV-mounted Reconfigurable Intelligent Surfaces (RIS) offer an effective solution to counter interference from unknown jammers. Our research begins with a study of security in digital and wireless networks and progresses through two phases, each contributing to a comprehensive anti-jamming solution. In the initial phase, we used UAV-borne RIS to address multi-jamming scenarios with a Reinforcement Learning-based technique, optimizing trajectory and phase shift beamforming. Using the lightweight Deep Deterministic Policy Gradient (DDPG) technique, we achieved near-optimal solutions, improving transmission rates and energy efficiency by 50-70\% compared to related works. This work is further extended with a multi-objective optimization approach to counter jamming threats in densely populated smart city environments, safeguarding essential services during public events. We employed Proximal Policy Optimization (PPO) to optimize UAV trajectory and RIS passive beamforming, demonstrating significant improvements in average sum rates and energy efficiency. In the second phase, we addressed dynamic multi-user clusters threatened by unknown jammers during dense crowded public events. We customized PPO for swarm UAV-borne RIS relays, optimizing device-to-UAV association, base station transmit power, multi-UAV trajectories, and RIS phase shifts. Our approach also provided adaptive swarm UAV formation and dynamic device clustering, ensuring coverage for dynamic and scalable environments. Continuous operations were ensured through UAV recharging and swapping facilities. Simulation results highlighted superior performance in jamming mitigation, data rate, energy conservation, connectivity, and scalability.

Date of Award	2025
Original language	American English
Awarding Institution	HBKU College of Science and Engineering