基于SDN的动态多播关键技术研究

发布时间：2018-06-17 14:52

本文选题：多播 + SDN　；参考：《上海大学》2017年博士论文

【摘要】：移动互联网、云计算及网络媒体的快速发展逐渐改变了互联网的形态和业务需求,这主要体现在以下四个方面:1)底层网络的拓扑结构可能会根据使用者的弹性需求而动态地改变;2)网络内移动终端规模爆发式增长;3)流媒体成为网络流量的主要贡献者;4)视频点播、在线会话等与流媒体相关的分布式应用成为一类重要的网络服务。这一新的互联网发展趋势可以概括为底层网络及终端设备的动态化,网络流量的媒体化,以及信息生产者的多元化。因此,在新的网络环境下需要有一种高效的通信方式为分布式应用提供有力地支撑。单播、广播、多播是互联网内三类主要的通信方式。与单播和广播相比,多播是用以支撑这类应用的理想手段。它不需要像单播一样在多方通信的参与者之间建立点对点连接,因此减少了带宽资源的占用;它也不需要像广播一样通过泛洪的形式向网络的所有分支发送大量数据包,因此有效减少了网络内无效数据包的数量,缓解了网络冲突。可见,多播通信在当前新的网络发展趋势下显得越来越重要。多播通信是在多播框架的支撑下实现的,多播框架起到了管理多播组和执行多播路由算法的作用。然而,在经典IP网络架构下,现存的多播框架在设计之初就具有通用性和独占性。其缺省的多播路由算法没有为底层网络和终端设备的动态性做出特别的优化,同时客制化的多播路由算法也无法被部署于该框架上。因此,现存的多播框架无法满足各种应用程序对多播通信的不同Qo S需求。不仅如此,目前所广泛采用的PIM、CBT等多播协议都基于汇集点建立多播树,这有可能引起数据包在汇集点附近产生拥塞,同时也引入了“最优汇集点选择”的新问题。为了解决上述问题,必须向网络内部署一个额外的多播框架,然后将面向动态多播的路由算法或面向其他Qo S需求的多播路由算法部署在这个多播框架上。然而,由于在多播通信设计之初并没有考虑其扩展性和灵活性,因此当网络内存在两个多播框架时,它们之间将会产生严重的冲突。为了向网络内部署客制化的路由算法,传统的方法是使用应用层多播框架,通过应用层协议管理多播组。而现存多播框架通过网络层的IGMP协议管理多播组,两种多播框架因此彼此隔离开来从而避免了冲突。然而,正是由于应用层多播框架没有采用标准的IGMP协议管理多播组,造成过去已经开发完成的海量经典多播应用程序无法兼容于该框架。多播应用程序的开发者不得不为每个多播路由算法开发专用的应用层协议和相应的应用层程序,这违反了软件工业的可重用原则。因此,如何向网络内部署适用于动态应用场景的多播路由算法而又保持与经典多播应用程序的兼容性成为多播研究领域内的一个待解的难题。为了解决这一问题,本文基于新兴SDN网络架构,对动态多播从物理框架、数学理论基础、路由算法、实际应用四个方面进行了研究。提出了一个基于SDN的插件化多播框架,该框架能够在不与现存多播框架产生干扰的情况下允许使用者向本框架内部署客制化的多播路由算法,同时又保持与经典多播应用程序的兼容性。在此框架的基础上,本文提出了面向动态应用场景的非重构动态多播路由算法和可重构多播路由算法。最后,我们基于本文所提出的多播框架和算法实现了一个分布式共享内存系统,该系统证明了多播框架和算法的可行性和高效性。全文的主要研究内容和创新点包括:针对现存多播框架的扩展性和灵活性不佳的问题提出了基于SDN的插件化多播框架。该多播框架能够在不与现存多播框架产生冲突的前提下使得客制化多播路由算法作用于网络。更重要的是,与应用层多播框架相比,它完全遵循标准IGMP协议实现,因此保持了与经典多播应用程序的兼容性。针对可重构多播路由算法对多播树的扰动而引起的丢包,提出了基于SDN的丢包保护机制。将链路的带宽因素引入到时变图理论内并将该理论从时间空间推广到时间空间和多播组空间,以更清晰、精确、全面地形式化描述SDN环境下的动态多播和可重构多播路由算法。针对新的网络环境下多播的高动态特性以及汇聚点对现有动态多播路由算法所带来的缺陷提出了基于任播模型的非重构多播路由算法和可重构多播路由算法。在上述四个研究成果的基础上提出了基于SDN的分布式共享内存。
[Abstract]:Mobile Internet, cloud computing and the rapid development of network media have gradually changed the form and business needs of the Internet. This is mainly reflected in the following four aspects: 1) the topology of the underlying network may be dynamically changed according to the user's elastic demand; 2) the scale of the mobile terminal in the network is increasing; 3) streaming media becomes a network flow. The main contributors of the quantity; 4) the distributed applications related to streaming media, such as video on demand, online session and so on, become an important class of network services. This new trend of Internet development can be summarized as the dynamics of the underlying network and terminal devices, the media of network traffic and the diversity of information producers. Therefore, in the new network environment, the new network environment is in the new network environment There is a need for an efficient way of communication to provide strong support for distributed applications. Unicast, broadcast, multicast are the three main types of communication in the Internet. Compared with unicast and broadcast, multicast is an ideal means to support such applications. It does not require a point to point connection between participants in a multiparty communication like a unicast. Therefore, it reduces the occupancy of bandwidth resources; it does not need to send a large number of packets to all branches of the network through flood like broadcast, so it effectively reduces the number of invalid packets within the network and alleviates the network conflict. It is clear that multicast communication is becoming more and more important in the current trend of new network development. Under the support of multicast framework, the multicast framework plays the role of managing multicast groups and executing multicast routing algorithms. However, under the classic IP network architecture, the existing multicast framework is universal and exclusive at the beginning of the design. Its default multicast routing algorithm does not do the dynamic performance of the underlying network and terminal devices. There is a special optimization, and the custom-made multicast routing algorithm can not be deployed on the framework. Therefore, existing multicast frameworks cannot meet the different Qo S requirements for various applications to multicast communications. Not only that, the currently widely used multicast protocols such as PIM, CBT and other multicast protocols are based on the pooling point to build multicast trees, which may cause data. In order to solve the problem, an additional multicast framework must be deployed to the network, then a dynamic multicast routing algorithm or a multicast route oriented to the other Qo S requirements are deployed on the multicast framework. It does not take into account its scalability and flexibility at the beginning of the multicast communication design, so there will be a serious conflict between them when the network is stored in two multicast frameworks. In order to deploy a customized routing algorithm to the network, the traditional method is to use the application layer multicast frame to manage the multicast group through the application layer protocol. Multicast frameworks manage multicast groups through the IGMP protocol of the network layer, and the two multicast frameworks are isolated from each other to avoid conflicts. However, it is precisely because the application layer multicast framework does not use standard IGMP protocols to manage multicast groups, resulting in the mass Scripture multicast applications that have been developed in the past can not be compatible with the framework. The developers of sowing applications have to develop special application layer protocols and corresponding application layer programs for each multicast routing algorithm, which violates the reusable principle of the software industry. So, how to deploy the multicast routing algorithm for dynamic application scenarios to the network and keep compatibility with the classic multicast application to the network. In order to solve this problem in the field of multicast, in order to solve this problem, this paper, based on the new SDN network architecture, studied four aspects of the dynamic multicast from the physical framework, the mathematical theory basis, the routing algorithm and the practical application. A SDN based plug-in multicast framework, which can not be used with existing multicast frames, is proposed. On the basis of this framework, a non reconfigurable dynamic multicast routing algorithm and a reconfigurable multicast routing algorithm for dynamic application scenarios are proposed. Finally, we are based on this framework. The multicast framework and algorithm proposed in this paper implements a distributed shared memory system, which proves the feasibility and efficiency of the multicast framework and algorithm. The main research content and innovation of the full text include: a plug-in multicast framework based on SDN is proposed for the problem of the scalability and flexibility of the existing multicast framework. The multicast framework can make the custom-made multicast routing algorithm act on the network without conflict with existing multicast frameworks. More importantly, it is fully consistent with the standard IGMP protocol implementation compared with the application layer multicast framework, so it keeps compatibility with the classic multicast application. Packet loss caused by multicast tree disturbance, a packet loss protection mechanism based on SDN is proposed. The bandwidth factor of the link is introduced into the time-varying graph theory and the theory is extended from time space to time space and multicast group space. The dynamic multicast and reconfigurable multicast routing algorithm in the SDN environment is described more clearly, accurately and fully topographic. In view of the high dynamic characteristics of multicast in the new network environment and the defects brought by the convergence point to the existing dynamic multicast routing algorithm, the non reconfigurable multicast routing algorithm and reconfigurable multicast routing algorithm based on the broadcast model are proposed. Based on the four research results, the distributed shared memory based on SDN is proposed.
【学位授予单位】：上海大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TP393.0

【相似文献】