Service Function Chain Deployment and Optimization at the Network Edge: An Application of A3C Algorithm

NFV technology deploies network functions on a universal server, and constructs management and orchestration modules to create and delete VNF and its resources. This provides a way to realize the virtual resource orchestration function of SFC orchestration system. The Actor-Critic algorithm combines two types of reinforcement learning algorithms, PolicyGradient and value-based Q-learning. A3C algorithm has been proved to have better convergence, and it performs better in high-dimensional and continuous space. Deep reinforcement learning has strong perception ability and decision-making ability at the same time, and the effect is stable and good in the face of various complex mobile edge computing problems. Dynamic service function chaining can adaptively set and adjust SFCs to ensure service continuity when users move. This paper presents the novel methods of service unction chain deployment and optimization at the network edge: an application of A3C algorithm. Finally, a comparison is carried out to verify the effectiveness of the proposed new algorithm. The experimental results show that $\mathrm{A3C}$ algorithm is better than DDPG algorithm in SFC placement efficiency.