Unmanned Aerial Vehicles Meet Multi-Agent Reinforcement Learning

Unmanned aerial vehicles (UAVs) can be served as aerial base stations (BSs) to provide cost-effective and on-demand wireless communications. UAVs are rapidly deployable for complementing the terrestrial communication based on a 3GPP LTE-A. Machine learning as a promising tool provides an autonomous and effective solution in an intelligent manner to enhance the UAVs enabled communication networks. However, most of the proposed machine learning algorithms focus on single UAV scenarios or multi-UAV scenarios by assuming the availability of complete network information for each UAV. In practice, it is difficult to have perfect knowledge of dynamic environments due to the high movement speed of UAVs, which imposes formidable challenges on the design of reliable UAV enabled wireless communications. In this talk, a novel framework based on stochastic game theory will be provided to model the dynamic resource allocation problem of multi-UAV networks and a multi-agent reinforcement learning (MARL) based resource allocation approach will be presented for solving the formulated stochastic game of multi-UAV networks. With the help of stochastic modelling, reinforcement learning based automated trajectory optimization approach will also be presented.