A dynamic planning model for deploying service functions chain in fog-cloud computing

Fog computing allows services to be deployed on computing resources at the edge of the network to address the limitations of centralized cloud systems. However, the adoption of fog computing concepts is in the early stages, and there are still many challenges to benefiting from infrastructure in Fog -Cloud Computing-based Networks (FCCN). One of them is known as Service Function Chain (SFC), which can use network software instances instead of costly dedicated hardware to share resources. Network Function Virtualization (NFV) technology separates hardware middleboxes such as gateways, firewalls, and set-top boxes from hardware and treats them as Virtual Network Functions (VNFs), where they can execute as software instances on decentralized nodes in the FCCN. VNFs are chained together in specific sequences that form SFCs. Meanwhile, deploying VNFs on nodes in the FCCN to accomplish SFC is an NP-Hard problem that can lead to efficient utilization of resources and reduce latency and cost. Recent research has performed SFC placement through heuristic algorithms that often cannot cope with the dynamic behavior of the network. In addition, existing works explicitly ignores SFC placement with the reuse of VNF instances. Hence, in this paper, we address the SFC placement problem by reusing VNFs through Deep Reinforcement Learning (DRL) based approaches. The proposed algorithm as a dynamic planning model can reconcile service costs and Quality of Service (QoS) by considering resource con-straints and dynamic distribution analysis of VNFs required in the FCCN. Here, the Asynchronous Advantage Actor-Critic (A3C) algorithm is used as a DRL approach with the aim of maximizing long-term cumulative reward. The simulation results through real-world data traces show that the proposed algorithm effectively improves the system performance and outperform the best result of benchmark methods ranging from 14% to 28% by considering the resource cost.(c) 2022 Published by Elsevier B.V. on behalf of King Saud University. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).