Multi-UAV Offloading and Trajectory Optimization in SAGIN with Dynamic Role Switching

The space-air-ground integrated network (SAGIN) provides wide-area coverage, flexible deployment, and efficient data transmission, and its integration with mobile edge computing (MEC) technology enhances support for smart devices in remote areas. However, the limited onboard energy of unmanned aerial vehicles (UAVs), coupled with imbalanced energy consumption across UAVs with different roles, significantly impacts the overall system service time. To address this challenge, we propose a MEC-enabled SAGIN architecture that incorporates dynamic UAV role switching (DURS), allowing each UAV to switch its role as a UAV base station (UAV-BS) or a UAV edge computing server (UAV-ECS). Based on this architecture, we propose an optimization problem that aims to maximize system service time by simultaneously optimizing device offloading decisions, UAV trajectories, and UAV role switching strategies, while ensuring real-time communication coverage and meeting the computational demands of devices. Given the complexity of the optimization problem, which involves multiple interconnected decision variables, the twin delayed deep deterministic policy gradient (TD3) algorithm is used to solve this problem. Simulation results show that the proposed DURS scheme significantly improves the system service time compared to other schemes by effectively balancing the energy consumption among UAVs.