Design And Implementation Of Fog Computing Network Based On SDN

Fog computing is a highly virtualized platform that provides limited distributed computing,storage,and networking services between end users and cloud servers.It can decrease the request delay for end users,relieve the heavy burden of storage and computing for cloud servers as well as solve the problem of spectrum shortage by achieving some functions for end users and cloud servers.Fog computing is so a novel research area that there is no unified platform and standard in academe yet.But the existing studies have shown that fog computing is a promising technology in the construction of smart cities.However,fog computing devices are located in the edge of the network,forming a complex and heterogeneous edge network.This will result in complicated management problems for fog network.To solve this problems,a fog computing network architecture model which can manage the computing,storage and network resources uniformly with the help of software defined network(SDN)controller is designed in this paper.This model consists of some end users,a fog network layer and a cloud computing center.Meanwhile,fog network layer includes access devices that provide network access and protocol conversion,fog nodes that supply the function of computing and storage,and a control node that can manage the fog network layer resources uniformly.Next,the process of completing the terminal data processing request based on this model is described in detail.As to the limited resources of fog nodes,this paper virtualizes the fog nodes by using a lightweight virtualization technology,i.e.,container technology,which can provide flexible computing and storage resources for end users.Based on the fog computing network model and SDN technology,this paper builds a fog network platform,and deploys the video edge caching application on the platform.Test results in terms of response time,transmission bandwidth and transmission delay have shown that the fog computing can reduce the request delay of users,provide low delay application services and improve the quality of experience of the end users.In the earlier designed fog computing network,it will trigger a wireless access point handoff when the user leaves the original location.At the same time,in order to allow end users in different locations to get low latency services,it is necessary to remove the application service of the end user to the nearest fog node.To solve this problem,this paper designs a method based on SDN to realize fast handoff among the access points.Next,by the container checkpoint and restore mechanism,we successfully achieve live migration of docker containers among fog nodes,and analyze the relationship between the consumed time of each stage in migration and the memory occupation during the process of migration.Finally,the superiority of container technology in migration time and the rationality of applying container technology as fog node virtualization technology are verified by comparing the migration time of docker container with that of KVM virtual machine.