| The traditional network based on the distributed architecture uses a "best effort" method to transmit network data,which cannot meet the high-quality transmission needs of a large number of users,and network resources cannot be fully utilized and are likely to cause waste of resources.As a centralized architecture,software-defined network(SDN)can well solve the above problems.Its biggest feature is the separation of control logic and forwarding plane.It can schedule network resources through centralized control and improve the utilization rate of network resources.In addition,under the software-defined network architecture,users can develop new functions according to their own needs,which has strong flexibility.However,it is difficult to achieve large-scale deployment of software-defined networks in today’s networks.The main reason is that the deployment of software-defined networks requires the replacement of current network equipment.Overlay network is a virtual network technology,which is a logical network built on top of the network.It improves the properties of the bottom layer of the network by adding a virtual network,shields the dynamic changes of the physical network,and provides high-quality services to the upper network.Although the overlay network has flexible deployment capabilities,it cannot solve the problem of centralized scheduling of network traffic.Firstly,this thesis analyzes the advantages and disadvantages of software-defined network and overlay network technology in detail,and proposes a cloud-edge data transmission framework based on software-defined overlay network with cloud-edge data transmission as the application background.Software-defined overlay network is an overlay network that adopts the idea of software-defined network.It makes full use of the centralized control of software-defined network and the flexible deployment of overlay network,and deploys software-defined network functions without changing the current physical network.This method can effectively solve the problem that the software-defined network is difficult to deploy in the current Internet,and can realize the centralized optimization of network resources.Secondly,this thesis studies the construction of overlay network for service quality optimization of cloud-edge data transmission.In a network topology with limited link bandwidth and definite transmission delay,an overlay network is constructed by establishing data transmission paths for different cloud edge nodes to minimize construction costs while avoiding network congestion.Considering the high efficiency of genetic algorithm(GA)in solving complex optimization problems,the coding strategy,fitness function and genetic operator of classical genetic algorithm are improved so that it can be better applied to the above optimization problems.By comparing with the existing algorithms,the results show that the algorithm proposed in this thesis has good performance.Finally,this thesis studies the cloud-edge traffic scheduling problem based on software-defined overlay network.In the cloud-edge data transmission process,there are large data flows that occupy a lot of bandwidth resources and exist for a long time,and there are small data flows that occupy few bandwidth resources and exist for a short time.This thesis studies traffic scheduling algorithms for large and small data flows.In this thesis,a network status detection method is developed,and the data flow in the network is divided into big data flow and small data flow.Traffic scheduling is performed based on network status and traffic information,and different traffic scheduling strategies are designed for different data flows.Compared with the existing algorithms through experiments,the results show that the algorithm proposed in this thesis has certain advantages in solving traffic scheduling problems. |
