EdgeSFG: A matching game mechanism for service function graph deployment in industrial edge computing environment

Abstract

Network function virtualization-enabled industrial edge computing is a promising service paradigm. In industrial edge computing (IEC), virtual service functions (VSFs) are used to implement basic functionalities and deliver concrete services. To improve the service response speed, VSFs are organized into service function graphs (SFGs) to process service data parallelly. The main challenge is to select proper edge servers for SFG deployment and minimize latency. However, finding an optimal solution to the SFG deployment problem is an NP-hard problem. This paper proposes a matching-game-based algorithm to deploy a large number of concurrent SFGs in an IEC environment. In the proposed algorithm, the deployment of concurrent SFGs is formulated as a college admission problem where SFGs propose to edge servers expecting a low service latency, and then the edge servers decide whether to accept the SFG proposals according to the possible service gain. A distributed matching procedure is introduced to solve the matching game problem between the SFGs and edge servers. The simulations are conducted to verify the SFGs and evaluate the performance of the proposed algorithm in concurrent SFG deployment. Simulation results show that proposed algorithm performs well in reducing average service latency and improving service acceptance rate.