IRIS: Propulsion Aware Secrecy Energy Optimization in Dynamic RIS Assisted UAV IoT Networks

Abstract

  The use of reconfigurable intelligent surfaces (RIS) and Unmanned Aerial Vehicles (UAVs) has started to become a viable solution for secure and energy efficient communication in IoT networks. However, most of the current approaches concentrate on the aspect of secrecy performance only and neglect the aspect of propulsion energy and the evolvable wireless environment, which have a significant impact on practical UAV system. In this work, a learning driven framework called IRIS (Intelligent Reconfigurable Integrated Security) is introduced to handle this issue through propulsion-aware secrecy-energy optimization. This research piece tackles the problem of jointly optimizing the trajectory of the UAV, the transmit power and the RIS phase configuration, with the help of reinforcement learning and enables the system to adapt continuously as the network conditions evolve. To estimate its performance IRIS is compared with PPO, TD3, SAC, DDPG, and DQN algorithms. The simulation outcomes confirm significant improvements with nearly 3.32 bits/s/Hz secrecy rate, 3.51 bits/J energy efficiency and quick convergence, without any communication delay in dynamic IoT scenarios.