Distributed Trajectory Design for Cooperative Internet of UAVs Using Deep Reinforcement Learning
Jingzhi Hu, Hongliang Zhang, Kaigui Bian, Lingyang Song, Zhu Han · 2019
In this paper, we consider a cellular Internet of UAVs, where UAVs execute multiple sensing tasks continuously and cooperatively through sensing and transmission with the objective to minimize the age of information (AoI). However, the cooperative sensing and transmission is coupled with the trajectories of the UAVs, which makes the trajectory design a challenging problem. To tackle this challenge, we first propose a distributed sense-and-send protocol to coordinate the UAVs. Based on this protocol, we formulate the trajectory design problem for AoI minimization and propose a deep reinforcement learning algorithm to solve it, which we refer to as the compound-action actor-critic (CA2C) algorithm. Simulation results show that the CA2C algorithm outperforms two baseline algorithms for AoI minimization.