Design And Implementation Of Low-latency Orchestration Method For Edge Computing Service Function Chain

Compared with the traditional way that processing data in remote cloud centers,Mobile Edge Computing(MEC)provides a high-bandwidth and low-latency network environment by sinking computing and storage capabilities to the edge of the network.This can improve the service quality of delay-sensitive services.As one of the key enabling technologies of MEC,network function virtualization(NFV)supports the decoupling of network functions from the underlying hardware resources,and the configuration of virtual network functions(VNF)on a unified physical infrastructure.So that network deployment and management can be carried out in a flexible,efficient and flexible way.A group of VNFs are arranged in a certain order and connected together by virtual links to form a service function chain(SFC).The data flow of the service needs to traverse the VNFs in order to complete the end-to-end service delivery.The process of placing VNFs and virtual links on a physical network and assigning physical network resources to them is called SFC orchestration.A reasonable orchestration scheme helps improve network performance and further reduces the end-to-end delay of the MEC network.At the same time,in order to improve resource utilization and reduce deployment costs,multiple VNFs can be deployed on a single MEC server,allowing them to share hardware resources in an efficient manner.Therefore,it is of great significance to study the SFC orchestration technology in the MEC network.However,the SFC orchestration in the MEC network still has the following problems:1)In previous studies,only the transmission delay on the link is considered,and the processing and queuing delay on the node are not considered,and there is a lack of accurate quantitative analysis of the end-to-end delay;2)In the MEC network with complex and changeable environment and business requirements,the method of establishing a solvable mathematical models is difficult to adapt to the dynamic environment.Based on the above background,a low-latency SFC orchestration method is proposed in this paper.Specifically,the coordinated allocation of resources between VNFs is mainly considered when multiple VNFs are deployed on a single server.Aiming at the problem that the existing delay calculation models only consider the transmission delay on the link,a series queuing model is proposed to accurately evaluate the end-to-end delay.The model determines the processing rate according to the amount of computing resources allocated to the VNF,and uses queuing theory to analyze the queuing,processing and leaving process of data packets when passing through each VNF on the SFC,so as to determine the average time when the business data flow traverses the SFC.According to the end-to-end delay of the tandem queuing model,a model for the collaborative allocation of resources with minimal delay is established.Aiming at the problem of low efficiency and poor adaptability of mathematical solving methods,a deep reinforcement learning method is proposed to learn the optimal resource allocation strategy through interactive experience with the environment.A low-latency SFC orchestration algorithm based on asynchronous advantage actor-critic(A3C)is proposed.Based on the above research content,the NFV resource allocation simulation system is designed and implemented.According to the standard NFV architecture design,the SFC orchestration component is implemented,allowing users to directly run the SFC orchestration algorithm on the system and observe the simulation results.This article has carried out detailed requirements analysis,outline design and detailed module design,and implemented the system based on the Open Stack platform.The functional test shows that the system can meet the requirements of users for the simulation experiment of SFC arrangement algorithm.In summary,in response to the delay requirements of services in the MEC network in the NFV environment,this article establishes a series queuing model based on queuing theory to accurately evaluate network end-to-end delay,and proposes a low-latency orchestration algorithm based on deep reinforcement learning,which can improve the delay performance of MEC network.Finally,a simulation system for SFC orchestration was designed and implemented,which laid the foundation for continuous improvement of SFC orchestration method.