An Effective Scheme for Delay Minimization in a Multi-UAV-Enabled NOMA-MEC System

Integrating unmanned aerial vehicles (UAVs) with mobile edge computing (MEC) presents a viable approach to addressing challenges posed by computation-intensive tasks. Faced with scarce communication resources, this letter adopts non-orthogonal multiple access (NOMA) to enhance spectrum efficiency, aiming to minimize the sum maximum delay of processing tasks from wandering user equipment (UEs) in a multi-UAV-enabled MEC system by jointly optimizing user association, offloading strategy, resource allocation and UAV three-dimensional (3D) trajectories. We propose a K-means-based user association and multi-agent soft actor-critic (MSAC)-based algorithm, which utilizes entropy maximization for exploration and stable learning to address the non-convex problem. Simulation results show that our proposed algorithm outperforms other schemes.