A DRL Approach to Multi-UAV Deployment for Cellular Systems With Non-Ideal Backhaul

In this paper, we investigate the user association and uncrewed aerial vehicle (UAV) deployment problem in a cellular system with non-ideal backhaul. We formulate a throughput maximization problem with backhaul capacity constraints. On this basis, we propose a multi-agent proximal policy optimizationbased algorithm to optimize UAV deployment. To satisfy backhaul capacity constraints, we design a user association strategy combining the maximum instantaneous received signal strength initial association with two user offloading mechanisms. The results show that the proposed algorithm performs better than the existing ones in terms of the throughput under non-ideal backhaul.