SFC embedding in Space-Air-Ground Integrated Network Based on DRL

With the potential to provide ubiquitous access and various services, Space-air-ground Integrated Network (SAGIN) has attracted widespread attention as an emerging network architecture in recent years. Traditional terrestrial communication network can hardly provide users or equipment located in rural areas and disaster sites with high quality connection due to limited coverage and fixed structure. Therefore, SAGIN is envisioned to be an indispensable component of the next generation mobile communication technology However, the resource allocation and service provision in SAGIN face many challenges caused by its heterogeneous and time-varying characteristics. The development of Software-defined network (SDN) and network function virtualization (NFV) bring potential for better management of the heterogeneous networks. Thus, the resource allocation and service provision in SDV/NFV enabled SAGIN can be deemed as a Service function chain (SFC) orchestration problem. In this paper, we study the SFC embedding based on deep reinforcement learning (DRL) in SAGIN. The process of embedding of the SFC is formulated as a markov decision process (MDP) and the algorithm for SFC embedding based on deep reinforcement learning is proposed to achieve the goal of improving long term revenue The simulation results show that the proposed algorithm has better performance than the traditional algorithm in terms of service acceptance ratio and revenue to cost ratio.