Drone base station positioning and power allocation using reinforcement learning

Large scale natural disasters can cause unpredictable losses of human lives and man-made infrastructure. This can hinder the ability of both survivors as well as search and rescue teams to communicate, decreasing the probability of finding survivors. In such cases, it is crucial that a provisional communication network is deployed as fast as possible in order to re-establish communication and prevent additional casualties. As such, one promising solution for mobile and adaptable emergency communication networks is the deployment of drones equipped with base stations to act as temporary small cells. In this paper, an intelligent solution based on reinforcement learning is proposed to determine the best transmit power allocation and 3D positioning of multiple drone small cells in an emergency scenario. The main goal is to maximize the number of users covered by the drones, while considering user mobility and radio access network constraints. Results show that the proposed algorithm can reduce the number of users in outage when compared to a fixed transmit power approach and that it is also capable of providing the same coverage, with lower average transmit power and using only half of the drones necessary in the case of fixed transmit power.