UAV Trajectory Planning and Resource Optimization for NOMA-VLC Systems via MAS-MARL

The advent of 6G communication technologies drives the development of innovative solutions, such as unmanned aerial vehicle (UAV)-assisted visible light communication (VLC) systems. However, these systems face challenges, including limited bandwidth and scalability issues. To address these, we introduce non-orthogonal multiple access (NOMA) into the VLC framework and propose a mixed action space multi-agent reinforcement learning (MAS-MARL) algorithm. This algorithm optimizes both continuous and discrete variables, facilitating cooperative communication among multiple UAVs. Extensive simulations demonstrate its effectiveness in improving the overall performance of multi-UAV NOMA-VLC systems.