SDN架构下全局动态反馈拥塞控制算法的研究

发布时间：2018-02-25 20:23

本文关键词： 异构网络 SDN 网络融合拥塞控制　出处：《北京交通大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着移动互联网的兴起,网络服务日益增多,数据量呈爆炸式增长。传统网络架构经过二十多年的发展已处于瓶颈期,难以适应复杂多变的网络。SDN技术打破了传统网络架构的局限性,将控制平面与数据转发平面分离以及网络可编程的特性为解决互联网中出现的问题提供了新的探索及应用方向。论文的研究内容如下:(1)本文以有线网与无线网融合的异构网络为研究背景,基于SDN的分层思想完成了 OpenFlow无线交换机的设计与实现,并提出了一种使有线网络与无线网络高效融合的方案。由SDN控制器同时控制管理有线交换机以及无线交换机,从而能对全局网络的资源实现共同管理及统一调配。该方案解决了传统网络中有线网络设备与无线网络设备之间相对封闭独立的问题。(2)本文提出了一种基于全局网络链路状态监控的拥塞控制算法。利用SDN控制器监测网络状态动态调度数据流,算法在网络链路拥塞之前,能快速根据多径为数据流重新规划路径,并且能够根据链路状态判定最优路径进行流传输。鉴于SDN网络中的短数据流的生存时间较短,而且对实时性要求较高,过度调度可能会导致其时延增大,造成与预期相反的结果,算法会对网络中长短数据流进行区分,并且只对长数据流进行调度。如果有多条长数据流等待调度,算法会根据它们占用链路带宽容量的大小以及覆盖拥塞链路的条数计算调度值,并挑选出调度值最大的长数据流进行优先调度。(3)本文使用Mininet平台对有线网与无线网融合的异构网络环境进行仿真,并由RYU控制器实现对网络的集中控制,在RYU控制器上部署相关的功能模块后进行实验验证本文算法的有效性。通过与基于SDN的CAAS算法以及SDT算法进行对比,本文提出的算法能有效缓解网络拥塞,并能提高网络的链路利用率以及吞吐量,降低时延。通过RYU控制器提供的接口,本文完成了网络控制管理平台的设计与实现,该平台可以直观的显示网络状态,并能实现对网络的控制。
[Abstract]:With the rise of mobile Internet, network services are increasing day by day, and the amount of data is increasing explosively. After more than 20 years of development, the traditional network architecture has been in the bottleneck period. It is difficult to adapt to the complex and changeable network. SDN technology breaks the limitations of the traditional network architecture. The separation of the control plane from the data forwarding plane and the programmable characteristics of the network provide a new exploration and application direction for solving the problems in the Internet. The research contents of this paper are as follows: 1) this paper combines wired network with wireless network. Heterogeneous network is the background of the research, The design and implementation of OpenFlow wireless switch based on the layered idea of SDN is completed, and a scheme of efficient integration of wired network and wireless network is proposed. The SDN controller controls and manages both wired switch and wireless switch simultaneously. This scheme solves the problem of relatively closed independence between wired network equipment and wireless network equipment in traditional network. A congestion control algorithm for network link state monitoring. SDN controller is used to monitor the network state dynamic scheduling data flow. Before the network link congestion, the algorithm can quickly replan the path for the data stream according to the multipath, and can determine the optimal path according to the link state. In view of the short lifetime of the short data stream in the SDN network, Moreover, the real-time requirement is high, and overscheduling may lead to the increase of delay, which will result in the opposite result. The algorithm will distinguish the long and short data streams in the network. If there are more than one long data stream waiting to be scheduled, the algorithm will calculate the scheduling value according to the amount of bandwidth they occupy and the number of bars covering the congested link. In this paper, the Mininet platform is used to simulate the heterogeneous network environment that the wired network and the wireless network merge, and the centralized control of the network is realized by the RYU controller. After deploying the related functional modules on the RYU controller, the effectiveness of the proposed algorithm is verified by experiments. Compared with the CAAS algorithm based on SDN and the SDT algorithm, the algorithm proposed in this paper can effectively alleviate the network congestion. Through the interface provided by the RYU controller, the design and implementation of the network control management platform is completed. The platform can display the network state intuitively. And can realize the control of the network.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.06

【相似文献】