Dynamic Computing Migration in Space-Ground Integrated Networks

The proliferation of edge computing has significantly reduced latency and improved network responsiveness, but its full potential remains unrealized due to challenges in managing the dynamic migration of mobile station (MSs). Existing research predominantly focuses on terrestrial cellular networks, overlooking the complexities of space-ground integrated systems. This paper addresses this gap by introducing a combined network architecture that enables seamless MS migration. We propose a distributed migration strategy assisted by the core network, leveraging a Distributed-Deployed Deep Q-Network (DP-DQN) algorithm. The proposed framework minimizes the computation delay and system overhead while optimizing resource allocation. Simulation results demonstrate that our approach outperforms traditional migration methods, achieving lower average delay and reduced network overhead.