UAV Resource Cooperation Based on Reinforcement Learning

Internet of things (IoT) devices are generally incapable of transmitting data over a long distance due to their energy limitations. With the advantages of flexibility, mobility and line-of-sight links to target devices, UAV are becoming more and more widely used in data acquisition systems. Because of the limited airborne resources of UAV, we must replenish them in time. This paper focuses on the aerial replenishment strategy, which can minimizes replenishment consumption while guaranteeing the fastest completion of the mission. We employ reinforcement learning to optimize UAVs' paths and replenishment strategy. The results show that, compared to the greedy algorithm and genetic algorithm, the reinforcement learning algorithm not only has the smallest energy consumption, but also has faster convergence speed.