부산대학교 도서관

Unmanned aerial vehicles (UAVs) are widely utilized to establish emergency communications, where the UAVs work as temporary base stations and the limited communication resource should be utilized efficiently. In this paper, we study how to maximize the utilization efficiency of spectrum and power based on the Multi-agent Soft Actor-Critic (MASAC) algorithm. In the proposed scheme, the centralized training with decentralized execution (CTDE) framework is used to achieve intelligent communication resource allocation for UAV emergency communications. The simulation results and theoretical analysis indicate that the CTDE framework can achieve better performance compared with the fully distributed (FD) framework and the MASAC algorithm outperforms the traditional Deep Q Network (DQN) algorithm.