Throughput-Coverage Optimization in RSMA-Enabled UAV Networks Using Graph Diffusion Models

Abstract

Unmanned aerial vehicle (UAV)-enabled wireless networks provide flexible coverage for ground users in dynamic and infrastructure-limited scenarios. However, when multiple UAV base stations share the same spectrum, system performance is strongly affected by UAV placement, user association, and co-channel interference. To address this issue, this paper investigates throughput-coverage optimization in rate-splitting multiple access (RSMA)-enabled UAV networks. We formulate a joint optimization problem that considers UAV deployment, user association, beamforming, and common rate allocation, with the objective of maximizing a weighted utility of system throughput and coverage. Since the problem contains both discrete association decisions and continuous resource allocation variables, it is difficult to solve directly. We therefore propose a graph diffusionassisted optimization framework. Specifically, UAVs and ground users are modeled as graph nodes, while UAV-user associations are represented as graph edges. A graph diffusion model is used to generate high-quality deployment and association strategies through iterative denoising, and successive convex approximation is applied to optimize RSMA beamforming and rate allocation under the generated topology. Simulation results show that the proposed method achieves faster convergence and higher throughput-coverage utility than the considered baseline schemes.