A Pre-migration Strategy of VNF in Satellite-Ground Integrated Networks

This paper investigates the issue of Virtual Network Function (VNF) migration in Satellite-Ground Integrated Networks (SGIN) by transforming the time-varying network topology into a software-defined Time Evolving Graph (SDTEG). A Markov prediction model is developed to address the instability of Inter-Satellite Links (ISLs) and the unpredictability of user requests, which is then converted to a Markov Decision Process (MDP). A joint strategy of VNF migration and Service Function Chain (SFC) routing rescheduling is proposed to determine if traffic can reach the destination node before ISL disconnections. Additionally, a VNF pre-migration algorithm based on the Softmax Deep Double Deterministic (SD3) algorithm and the Actor-Critic (AC) framework is proposed. Simulation results demonstrate that the proposed algorithm can effectively reduce the delay of SFC traffic due to changes in satellite topology and optimize load balancing performance to some extent.