无人机空中基站的路径规划研究

Research on the trajectory design of a UAV-Mounted base station

在无人机网络中,地面用户的移动可能降低用户与基站无人机间的无线通信速率,造成网络性能损失。为了避免这种损失,提出一种基于深度强化学习的基站无人机路径规划方法。该方法能够计算出在连续动作空间内无人机的飞行动作,帮助无人机实时追踪地面移动用户,提高用户与基站无人机间的无线通信速率,增强网络性能。将无人机提供通信服务的任务周期划分成多个时隙,每个时隙内地面用户移动的位置视为固定,每个时隙内的网络吞吐量为该时隙内所有用户的无线通信速率之和;以最大化任务周期内网络总吞吐量为目标,运用深度确定性策略梯度算法实时计算每个时隙内无人机的飞行动作,实现对无人机的路径规划。仿真实验结果表明:在考虑地面用户移动的无人机网络中,所提方法与3种常见的基准方法相比,在网络吞吐量上有更好的性能表现。

In unmanned aerial vehicle(UAV) networks,the movement of ground users may cause an attenuated transmission rate between the user and the UAV-mounted base station,resulting in a loss of network performance.In order to avoid this loss,this paper proposes a trajectory design method based on deep reinforcement learning(Deep Deterministic Policy Gradient-Trajectory Design,DDPG-TD).This method can calculate the flight direction and distance of the UAV in a continuous space according to the movement of mobile ground users,and increase the transmission rate between the users and the UAV,thereby improving the network performance.Firstly,a period of time is divided into multiple equal time slots,with a fixed position of the mobile ground users in each time slot and the network throughput in each time slot as the sum of the transmission rates of all users.Then,with the aim of maximizing the total network throughput within the task period,the flight direction and distance of each UAV in each time slot are calculated by DDPG-TD to consummate UAV trajectory design.The simulation results show that,in the UAV networks considering the movement of ground users,the proposed method has better performance in terms of network throughput than the three commonly used benchmarks.