Research On SDN Traffic Scheduling Based On Ant Colony Optimization Algorithm

In recent years,with the rapid rise of mobile Internet and big data business,the size of the network is going to expand,and the number of devices to access the network are constantly increasing.The traditional network management is limited in scalability and other aspects are slowly come out.Software Defined Network(SDN)is a new type of network architecture,control layer and data layer separation,which can carry out centralized management and dynamic maintenance of distributed network.It also can effectively solve the IP network management and scalability problems.Traffic engineering as a typical application of SDN,is one method of managers optimize network performance and traffic transmission.It includes dynamic analysis,forecasting and management of data flow.The traffic scheduling strategy focuses on traffic engineering,which is mainly divided into traffic scheduling of data layer,control layer scheduling,and traffic scheduling virtualization.This thesis studies the data layer flow scheduling for different network environments: one is in the traffic volume of similar size in the network environment,we could put forward the traffic scheduling strategy for the load balancing problem of network links according to the characteristics of global visual and centralized control.The second is in large volume size difference the flow of the network environment,and there are elephant and mice flows in the network,they have different transmission demands,So it could produce forwarding conflict.We should classify the flows before calculating routing,then according to different types of traffic flow to develop the corresponding scheduling scheme.The main work and contribution are as follows:(1)The thesis proposed a link load balancing algorithm based on Ant Colony Optimization(LLBACO),and defined a new path determination rule,then found the optimal transmission path of the current traffic network in the multi-constraint conditions,Finally,it realizes the load balancing of each link,and ensures the quality of network service.(2)In order to identify more elephant flows,this thesis sets new packets identification threshold,limits the failure rate of identifying elephant flow,and then takes the second detection of the data flows with the combination of sFlow and OpenFlow.By the experiment,the strategy could reduce the risk of network congestion because of elephant flows.(3)Considering the data-flows splitting will affect their transmission quality,this thesis proposed a k-ACO algorithm(k is the routing number).On the premise of meeting the bandwidth of elephant flows transmission,minimizing the value of K according to the current state of the network,reduced the elephant flows segmentation numbers,realize multi-link forwarding of elephant flows.