基于强化学习的智能超表面辅助无人机通信系统物理层安全算法

该文从物理层安全的角度出发研究了智能超表面(RIS)辅助的无人机(UAV) 3D轨迹优化。具体地说，当RIS辅助的UAV向地面用户进行无线传输时，通过联合优化RIS相移和UAV的3D轨迹来最大化物理层安全速率。然而，由于目标函数是非凸的，传统的优化技术很难直接求解。深度强化学习能够处理无线通信中动态复杂的优化问题，该文基于强化学习双深度Q网络(DDQN)设计一种联合优化RIS相移和无人机3D轨迹算法，最大化可实现的平均安全速率。仿真结果表明，所设计的RIS辅助UAV通信优化算法可以获得比固定飞行高度的连续凸逼近算法(SCA)、随机相移下的RIS算法和没有RIS的算法有更高的安全速率。

Physical Layer Security Algorithm of Reconfigurable Intelligent Surface-assisted Unmanned Aerial Vehicle Communication System Based on Reinforcement Learning

In this paper, the optimization problem of the 3D trajectory for Unmanned Aerial Vehicle (UAV) assisted by Reconfigurable Intelligent Surface (RIS) in physical layer security is studied. Specifically, when the RIS assisted UAV transmits wirelessly information to the ground user, the physical layer security rate is maximized by jointly optimizing the RIS phase shift and the UAV's 3D trajectory. However, because the objective function is non convex, the traditional optimization technology is difficult to solve it directly. The dynamic and complex optimization problems in wireless communication can be solved by deep reinforcement learning. Based on reinforcement learning Double Deep Q Network (DDQN), a joint optimization algorithm of RIS phase shift and UAV 3D trajectory is designed in this paper to maximize the achievable average safety rate. The simulation results show that the designed RIS assisted UAV communication optimization algorithm can obtain higher safety rate than the Successive Convex Approximation (SCA) algorithm with fixed flight altitude, RIS algorithm with random phase shift and algorithm without RIS.