Research On Efficient VNF Placement Method In The Mobile Edge Network

5G has been emerged as a promising communication technology to provide users in the mobile network with high-bandwidth and low-latency network services.Mobile Edge Computing(MEC)can provide such high-quality network services by migrating computing resource from data centers(DC)to the proximity of mobile users.Virtualization technologies,such as Software Defined Network(SDN)and Network Function Virtualization(NFV),reduces the dependency of network functions on dedicated network hardware,enhances the flexibility of network device deployment,by running network functions in the form of VNFs in Virtual Machines(VMs).However,there is limited computing resource in the cloudlet,which cannot meet the rapidly increasing computing resource demand from mobile users.How to utilize the computing resource in the cloudlet to provide more network services is a challenging problem.This paper proposes a multi-layer network architecture to deal with the problem of computing resource limitation in Cloudlets.Considering that the computing resources in DCs are more abundant than Cloudlets,this article deploys VNF instances of a network service with low service quality requirements in DCs under the premise of ensuring the Quality of Services(QoS)requirements of each network request,which releases more computing resource of Cloudlets.In terms of VNF instance deployment and routing problem,this paper designed a heuristic algorithm based on unsplittable flow algorithm.At the same time,there is user data security problem in the mobile edge network.For example,when a Cloudlet suffered malicious attack,user data security will be threatened.To cope with this problem,this paper designed a trust model for mobile users to evaluate the security level of Cloudlets,which enables mobile users to select a secure cloudlet to apply for computing resources.Finally,the performance of the two proposed methods is verified through a large number of simulations.Results show that the performance of the proposed algorithm achieves a better performance than their counterparts.