Reinforcement Q-learning enabled energy-efficient service function chain provisioning in multi-domain networks

Network function virtualization (NFV) technology is an efficient way to address the increasing difficulty of provisioning and managing network services. However, NFV-related service function chaining (SFC) deployment in multi-domain networks remains challenging, and there is still room for performance improvement. This paper investigates many heuristic algorithms in the same field and proposes a new method for dynamic SFC deployment in a multi-domain network. In our study, we combine a heuristic algorithm with reinforcement learning and divide the complex problem into several parts. This algorithm efficiently gives the SFC deployment scheme in the multi-domain network with subdomain privacy protection requirements and considers the energy savings of the multi-domain networks. Compared with the existing approach, the proposed algorithm has superiorities in terms of deployment success ratio, deployment profit, time efficiency, and energy consumption.