Resource Allocation for SFC Networks: A Deep Reinforcement Learning Approach

The rapid evolution of mobile communications technology has led to the widespread adoption of cutting-edge technologies such as Network Function Virtualization (NFV) and Software-Defined Networking (SDN). Network flexibility and scalability have been greatly improved by virtualizing traditional dedicated hardware functions on standard hardware and servers. However, the effective deployment of Service Function Chains (SFC) and the allocation of computing resources in this virtualized network environment remain to be solved. This study presents a resource allocation approach using the Proximal Policy optimization (PPO) algorithm. It features a novel SFC network model, utilizing reinforcement learning to intelligently train and make decisions, focusing on the cost-performance relationship in node deployment. Simulation results show that this algorithm effectively balances the relationship between node deployment cost and end-to-end service latency while ensuring the quality of service requirements.