【作者】 吴茜；

【导师】 吴建平；

【作者基本信息】 清华大学 ， 计算机科学与技术， 2006， 博士

【摘要】 互联网已成为现代社会最重要的信息基础设施和人们工作、生活的重要组成部分。移动互联网将满足人们在移动情况下访问网络的需求,具有广阔的发展前景。组播技术具有提高网络利用率,节省发送者资源等优点。在移动互联网中实现组播,能更好地发挥组播高效传送的优势。由于移动互联网结构复杂、节点位置变换而导致频繁加入/退出组播组等原因,移动互联网组播研究面临巨大挑战,缺乏相应的体系结构、算法和协议,处于起步阶段。论文研究了移动互联网的组播问题,主要研究内容和结论如下:1、分析总结了在移动互联网络中实现组播的难点和相关问题,综述了该领域的研究现状和发展趋势。2、提出了一种基于区域代理的层次化移动互联网组播结构。通过建立层次化模型,提出层间组播转换、分层组播组和组播树管理等机制,确保该结构在网络规模、异构环境和异类组播路由协议三个方面具有良好的可扩展性。提出了一种移动互联网组播算法的综合性能评价方法,从用户和网络两个角度综合进行评价,可广泛用于算法的评价、选择和改进。3、以层次化结构为基础,提出了一种快速切换的移动互联网组播算法FHMM。模拟实验表明该算法具有较高的组播用户性能和网络性能,是一个综合、高效的移动互联网组播算法。4、研究并发现了在移动互联网中组播树的稳定性问题比一般互联网更加严重。从移动互联网组播算法、组成员数量和网络规模、移动模型、移动速度和子网覆盖范围等五个方面研究了对组播树稳定性因子的影响,找出了主要影响因素,为移动互联网组播算法的稳定性设计提供了参考。5、提出了带移动预测的快速组播切换算法P-FHMM,进一步提高了FHMM算法的组播树稳定性。模拟实验结果显示,该算法在显著提高组播树稳定性的同时,仍然保证了较好的组播转发效率和较低的分组丢失率。6、设计实现了移动互联网组播的模拟分析系统,它具有可扩展性好、易配置、易管理等优点。更多还原

【Abstract】 Internet has become the most important information infrastructure in modern society, and an important component in our lives. Mobile Internet can extend the network services even when people are moving and it has wide development prospects. Meanwhile, IP multicast can efficiently deal with multi-destinations delivery and release the burden of replications from the source, so it has attracted enough attentions of researchers. Providing multicast in mobile Internet can make better use of efficiency, and it will be fruitful to bring these two technologies together. But the structure of mobile Internet is more complex, and the handoff of mobile nodes will cause the joining and leaving behaviors occur much more frequently, so it is challenging when providing multicast for mobile nodes. It is a new research area and still lacks of architecture, algorithm and protocol. In this thesis, the multicast in mobile Internet (mobile multicast for short) is studied in details. The main research works and conclusions are as follows:1. The difficulties of providing multicast in mobile Internet are analyzed, and then a comprehensive overview of existing researches on mobile multicast is provided.2. A hierarchical mobile multicast framework based on domain agent is proposed. Some key techniques are proposed, such as hierarchical multicast model, cross-domain multicast group transform mechanism, hierarchical multicast group management, and hierarchical multicast tree maintenance mechanism. These techniques make sure that the framework achieves high scalability on network scale, heterogeneous network environments and heterogeneous multicast routing protocols. Then an integrated performance evaluating theory is proposed. It evaluates the mobile multicast algorithms from the aspects of both multicast users and the whole Internet. This integrated performance evaluating criteria can be applied to design, select and improve the mobile multicast algorithms.3. Based on the hierarchical mobile multicast framework, the author proposes the Fast Multicast Handoff Based Hierarchical Mobile Multicast Algorithm (FHMM). The simulation results show that FHMM achieves better performance on both sides of multicast users and whole network. It is a comprehensive and efficient solution for mobile multicast.4. The author finds out and proves that mobility makes the stability problem of multicast tree more serious. In order to find out how the various elements impact on the stability factor, five elements are investigated. They are mobile multicast algorithm, the ratio of the number of mobile nodes and network size, the mobility model, the move speed and the power range. The main correlative elements are found which gives references when designing mobile multicast algorithms.5. In order to improve the stability of FHMM algorithm, the idea of motion prediction is introduced, and a new algorithm called P-FHMM is proposed. The simulation results show that P-FHMM can improve the stability of multicast tree remarkably while maintaining high efficiency and low packet loss percentage.6. A simulation system to investigate mobile multicast algorithms has been developed. It can be easily configured and managed, and has the merit of high scalability.更多还原