Research On Deployment And Optimization Of Service Function Chains At The Network Edge

In recent years,edge computing,as a new computing model,has attracted a lot of attention in the current wave of information technology development.At the same time,the rise of Network Function VIrtualization(NFV)provides Network operators with the idea of flexible Network deployment and management.The core idea of NFV is to decouple the dedicated physical equipment and Network Function,and replace the dedicated physical equipment with Virtual Network Function(VNF)realized by flexible software technology.On this basis,the industry proposes to implement edge computing based on the NFV architecture.Different types of edge computing services are deployed on edge computing nodes in different locations in the form of VNF.Through the NFV orchestration and scheduling mechanism,the connection of different types of virtual network functions is formed to form service function chain(SFC)to meet the specific edge computing service requirements of end users.However,how to effectively deploy these VNFs in edge computing networks to provide users with high-quality services is an urgent problem to be solved.In this thesis,starting from the applicability of NFV in edge computing,the deployment of VNF on the edge of network is studied in different application scenarios.First,this thesis studies the task offloading network function allocation problem based on QoS guarantee.This thesis focuses on how to select the appropriate VNF instance for the service request from the user to complete the task unloading process.At the same time,the QoS requirements of the service request and the constraints of network resources must be strictly guaranteed.In this thesis,a delay model is established for the service request of task offloading,and an integer linear programming model is established to maximize the network throughput,then the problem is proved to be NP-hard.Due to the NP hardness of the problem,a heuristic algorithm based on genetic algorithm and Lagrange relaxation is designed,which can calculate the appropriate solution for each service request to satisfy QoS and resource constraints.The simulation results show that the proposed algorithm can maximize the network throughput and guarantee the QoS requirements of service requests.Then,the efficient deployment of dynamic SFC resources for edge networks is studied.In this thesis,a more complex VNF combination VNF forwarding graph is studied on the basis of simple chain SFC to satisfy the business complexity in edge network.Based on this,this thesis establishes an integer linear programming model with minimizing the total cost of network resources deployed by SFC as the objective function,and proves the NP hardness of the research problem.Therefore,this paper proposes a heuristic algorithm based on dynamic programming to solve the research problem.The dynamic programming model is used to solve the resource efficient deployment scheme for service requests.Due to the dynamic nature of SFC,network resource fragmentation occurs,and the algorithm further reduces resource cost by integrating the same type of VNF instance that runs inefficiently.The simulation results show that the proposed algorithm can effectively reduce the resource cost and has better acceptance ratio.