Reinforcement Learning for Service Function Chain Allocation in Fog Computing
José Santos, Tim Wauters, Bruno Volckaert, Filip De Turck · 2021
Recently, distributed cloud infrastructures have become a potential business opportunity for most service providers due to the exponential growth of connected devices. The advent of the Internet of Things (IoT) and softwarized networks made centralized cloud systems impractical. In response, Fog Computing (FC) emerged, enabling the deployment of services on computational resources from the cloud up to the edge. However, the adoption of FC concepts is still in its early stages and challenges persist to fully benefit from fog–cloud infrastructures. One of them is known as Service Function Chaining (SFC) where providers benefit from network softwarization to create virtual chains of connected services. Recent research has tackled SFC allocation through theoretical modeling and heuristic algorithms, which often cannot cope with the dynamic behavior of the network. Thus, in this chapter, we explore a subset of machine learning (ML) called Reinforcement Learning (RL) to provide an efficient solution for SFC allocation in FC. The proposed approach learns about the best resource allocation decisions, focused on energy efficiency from a previously presented mixed-integer linear programming (MILP) formulation. Results showed that RL algorithms perform comparably to state-of-the-art ILP-based implementations while offering more scalable solutions. Future research directions and open challenges are discussed.