Multi-Agent Learning Approach for UAVs Enabled Wireless Networks

The unmanned aerial vehicle (UAV) technology provides a potential solution to scalable wireless edge networks. This paper uses two UAVs, with accelerated motions and fixed altitudes, to realize a wireless edge network, where one UAV forwards the downlink signal to user terminals (UTs) distributed over an area where another UAV collects uplink data. Both downlink and uplink transmissions consider the active user probability and the queue structure as well as the hovering times of UAVs. Specifically, we develop a novel joint Q-Learning multi-agent (JQ-LMA) algorithm to maximize the overall energy efficiency of the edge networks, through optimizing the UAVs trajectories, transmit powers, and the resistant distance between UAVs. The simulation results demonstrate that the proposed algorithm achieves much higher energy efficiency than other benchmark schemes.