Research On Network Slice Orchestration And Optimization Algorithms Based On SDN

With the development of mobile Internet,more and more novel services appear in the network,such as virtual reality,automatic driving and so on.These services have different quality of service requirements for network.Traditional ensuring schemes are difficult to meet the challenges of a large number of novel services in 5G scenarios,so the concept of network slice has been introduced.Network slice can transform the network from "one size fits all" to "one size per service".It divides the underlying physical network into a series of independent logical network,each of which serves a business scenario to meet the specific needs of such scenarios.This makes 5G meet the challenges of different business,and improves the reliability of the network.At the same time,because each logical network is independent,the cost of network deployment and maintenance is also reduced.Network slice orchestration and optimization algorithm is one of the hotspots of network slice research.There are some aspects that can be improved.The existing network slice orchestration algorithm uses CPU and bandwidth attributes of nodes to evaluate the resource amount of nodes.In some cases,it can not accurately reflect the real resource amount of nodes,which affects the nodes orchestration of network slice,this may cause the failure of network slice orchestration,and reduces the acceptance of network slice to a certain extent.The existing network slice optimization algorithm uses a single adjustment strategy to adjust the network slicing resources with high load,the adjustment time and success rate of the algorithm can be optimized in complex scenarios.The main work of this paper is as follows:(1)A network slice orchestration algorithm based on resource weight is proposed.In this paper,the current network slice algorithm is analyzed,and the topological properties of network are introduced into the resource evaluation scheme of nodes.The concept of resource weight is proposed combining the topological properties with the resources of adjacent nodes to measure the importance of a node in the network.The resource weight is used to select the location of node orchestration,which enhances the correlation between node orchestration and link orchestration.To improve the success rate of link orchestration,a link orchestration algorithm supporting multi-path scheduling is proposed.When link resources are insufficient,virtual links are arranged into multiple links.The experimental results show that the network slice orchestration algorithm proposed in this paper has higher acceptance rate and resource utilization rate,and can achieve greater revenue-cost ratio in the same scenario.(2)A network slice optimization algorithm based on hybrid decision-making is proposed.Links in network slices are divided into different states according to different load rates.A trigger mechanism for resource adjustment is designed based on link states.In order to calculate the amount of resources that need to be adjusted,the revenue and cost of network slice are modeled,and the resource optimization problem is transformed into the problem of maximizing the ratio of revenue to cost,and the particle swarm optimization algorithm is used to solve the problem.In the stage of resource adjustment,a resource adjustment strategy combining resource redistribution and link re-orchestration is proposed.Different resource adjustment strategies are selected according to link status and underlying resource status,which reduces effectively the time of resource adjustment.At the same time,it improves the success rate of network slice optimization.