Research Active Queue Management For Software Defined Networks

In recent years,the development of Internet,the explosive growth of business volume and user demands that cause the exponential growth of the network flow.As a result,there is a shortage in actual network bandwidth resources which means network congestion would occurs frequently.Due to the traditional congestion control that is based on distributed work,lack of global information network,its performance has been limited.The software-defined network which is a new network architect,separates the data layer and control layer.Data layer obtain the status information of the whole network and control layer use the whole network view of the network to control manage network information altogether.This could get rid of the limitation of traditional network and improve the use ratio of network resources.However,there still is time delay between controller and switch that could affect the order of controller accept events and the time delay of give indication to switch.If switch could not the speed of unexpected the data flow,network crowded would happened.The situation would be reduced if the data layer can autonomic response to these.Meanwhile,controller have enough configuration time while queue management algorithm and SDN are well-matched.As a light of this,control management of network crowded would be efficient and the performance of the genetic algorithm is improved.This thesis will focuses on SDN that based on queue management algorithm.Mainly aiming at the limitation of network resource distribute that only rely SDN controller.Only relying SDN controller result in data transmission delay and delay jitter and other issues.In order to alleviate the congestion,the queue management algorithm use for the data plane,waiting queue length real-time control management of the SDN network data layer nodes,network load balancing.The main work of this paper is as follows:Firstly,introduces the analysis process of SDN network architecture,OpenFlow technology and data forwarding plane.Analysis the exist problem of network and solve the congestion problem by control the midsize node of network flow.Secondly,analyze the typical algorithms in Active Queue Managemen,the random early detection algorithm is recommended by IETE based on the congestion control mechanism of the router.After analyzed the advantages and disadvantages of RED and the improved algorithm,putting forward PTRED algorithm,put PTRED algorithm in SDN network data layer.Use global views of SDN controller adjust parameters dynamically.Simulation experiments proved that this algorithm have feasibility and Validity of network congestion.Finally,the work done in this paper is summarized and the future work is prospected.