Coordinated UAV Deployment and Task Offloading in UAV-Assisted LEO Satellite Edge Computing Systems via Proximal Policy Optimization

Low earth orbit (LEO) satellites and unmanned aerial vehicles (UAVs) are emerging as key enablers of next-generation edge computing, providing connectivity and computational support in remote and infrastructure-limited regions. While LEO satellites offer global coverage, their constrained computational resources and intermittent connectivity pose significant challenges for handling dynamic workloads. UAVs, with their flexible mobility, can serve as intermediate edge nodes, augmenting satellite-based processing and reducing latency. However, efficiently coordinating UAV positioning and task offloading in such networks remains a complex problem due to their dynamic topologies, limited energy reserves, and heterogeneous service demands. This paper introduces a proximal policy optimization (PPO)-driven reinforcement learning framework that intelligently adapts UAV placement and computation offloading strategies to minimize latency and optimize resource utilization. Unlike traditional approaches, which rely on static heuristics or predefined rules, our method continuously learns and refines decision-making based on real-time network conditions. By dynamically balancing computational loads between UAVs and satellites, our approach enhances task execution efficiency while respecting coverage constraints and energy limitations. Simulation results demonstrate that the proposed framework significantly outperforms baseline methods in latency reduction, resource efficiency, and energy conservation.