【作者】 许毅；

【导师】 李腊元；

【作者基本信息】 武汉理工大学 ， 交通信息工程与控制， 2005， 博士

【摘要】 随着计算机网络技术的发展,视频会议、视频点播(VoD)和远程教学等实时多媒体应用已大量涌现,它们的共同特点是需要从一个源节点或多个源节点将信息传输到多个目的节点。由于这些实时业务对网络传输中的时延、时延抖动、带宽、包丢失率和代价等较为敏感,当突发性较高的FTP或含有图象文件的HTTP等业务在网络上传输时,实时业务的传输将会受到影响,这样就需要服务质量(QoS:Quality of Service)技术的支持,以确保实时业务的通信质量。传统的Internet路由协议,如开放最短路径优先(OSPF)和路由信息协议(RIP)是基于“最短路径”算法的路由技术,即路由是在单个特征值下的优化,只提供“尽力而为”服务,不能提供QoS保证。同时,随着移动网络和全光网络等高性能网络技术的不断发展及其应用领域的日益扩大,基于QoS约束的多播路由算法和协议的理论研究与实现,已成为网络领域中研究的重要课题。 QoS路由就是寻找满足特定QoS约束的一条可行路径(QoS单播路径)或一棵多播树(QoS多播路由),它与传统的尽力传输不同,QoS路由一般有多个约束条件,通常是一个NP-完全问题。而多播路由是网络层具备的功能,多播问题的关键在于多播路径的确定,实现多播的最常用的方法就是建立多播树,多播路由算法主要用来建立一棵性能好的多播树,并使它满足各种业务的服务质量(QoS)的需求。 目前提出的大多数QoS路由算法是启发式算法,有些算法由于计算的复杂性而无法满足实际应用的要求;有些算法因缺乏全局的观点,扩展性较差,难以适应有动态成员加入的大型群组。本文在总结这些算法的基础上,提出了解决QoS多播路由的相关方法,主要集中在多QoS约束的多播路由算法和协议、蚂蚁算法和移动代理在IP网络中路由的优化研究,具体包括以下几方面: 1)论文从研究QoS多播路由的现状出发,分析了IP网络中QoS路由的策略和QoS多播路由的问题;从IPQoS的定义和机制出发,研究了综合服务模型(IntServ)、资源预留协议(RSVP)、差分业务模型(DiffServ)、多协议标签交换(MPLS)、IPQoS的综合实现策略以及IPQoS在Linux中的实现;研究了QoS的多播路由机制,提出了在NS2下实现QoS约束的多播路由算法更多还原

【Abstract】 With the development of computer network technology, multimedia is enormously applied to real time video conference, VoD and long-distance teaching, etc. Their common feature is to transmit information from one or multiple source nodes to multiple destination nodes. As these real-time operations are sensitive to delay, delay-jitter, packet-loss and cost in the network transmission, they are liable to be affected by the operation portfolios of FTP and HTTP, and thus demand QoS to assure real time communications. Traditional internet routing protocols, such as OSPF and RIP, are routing technology based on "shortest path algorithm" in which the route is optimized with a single metric, and just offer "best-effort" service rather than QoS. Meanwhile, the theoretic research and realization of the multicast routing algorithms and protocols based on QoS have become a major research topic with the rapid development of such high-powered network technologies as mobile networks and AON All Optical Network and the progressive expansion of the scopes of their applications.QoS routing is how to select network routes with sufficient resources for the requested QoS parameters. In comparison with best-effort routing, QoS routing has some constraints and mostly is NP complete problem. Multicast routing is a network-layer function. A fundamental issue in multicast communication is how to determine an effcient message route (multicast routing). Tree construction is a commonly used approach in solving the multicast routing prolem. multicast routing algorithms are used to compute multicast trees that satisfy QoS requirements.Now, the propsed heuristic algorithms can’t meet the QoS requirements for computation complexity, and adapt to large groups with member join dynamically for their expansibility. A method for the QoS multicast routing is presented based on the summary of the algorithms and protocols proposed in the past. The research mainly concentrates on the algorithms and protocols of QoS multicast routing, ant algorithms and the mobile agent optimization routing in IP networks. In detail, the contributions of this paper include the following.1)Firstly, the analysis on the strategy of QoS routing and QoS multicast routing problem in IP networks are done by studying the status quo of QoS multicast routing. Secondly, based on the definition and mechanism of IP QoS , the research on the IntServ, RSVP, DiffSer, MPLS, the comprehensive implementation strategy of QoS in IP network and the implementation of IP QoS in Linux are conducted. Finally, an approach to QoS multicast routing algorithmsand protocols in NS2 is presented by studying QoS multicast routing mechanism.2)It discusses multicast routing problem with multiple QoS constraints of delay, delay jitter, bandwidth and cost metrics, describes a network model for researching the routing problem, and presents a dynamic multicast routing algorithm with multiple QoS constraints (DMRAQoS). In DMRAQoS, a multicast group member can dynamically join or leave a multicast session with no disruption of the multicast tree. Simulation results demonstrate that DMRAQoS provides an effective path for the research on multicast routing with multiple QoS constraints.3)The objective of multicast routing optimization is to make network resources utilization more effective by a connected mode to find an algorithm or strategy in the case of given network and multicast request. In this paper, firstly, genetic algorithms, simulated annealing and neural network are discussed and a globally optimal multicast routing algorithm (MROAQoS) is presented based on the ant colony ability of finding a shortcut between their nest and food sources in the process of hunting for food. Secondly the MROAQoS is modified according to application demand. Lastly, a mobile agent multicast routing algorithm with multiple QoS constraints (MAMRQoS) is presented, the mobile agent model in NS2 is designed and the validity and effectiveness of the MAMROoS is demonstrated.4)It presents a multicast routing protocol with multiple QoS constraints (MRPMQoS) based on the analysis on the protocols of YAM, QoSMIC and QMRP. The MRPMQoS does not maintain global network state but only requires the local state information of the link (or node). The MRPMQoS effectively decreases the overhead for constructing a multicast tree, diminishes the delay of the nodes join, and improves the success ratio of seeking links by combining single-branch search and multi-branch search. Simulation results show that the MRPMQoS provides an available means to implement multicast routing, and can adapt to all kinds of the topology networks and has good expansibility.5)It describes a network model for the hierarchical multicast routing with multiple QoS constraints, persents a QoS-based on hierarchical multicast routing protocol(HMRPQoS), which is based on the analysis on hierarchical routing and hierarchical QoS multicast routing algorithms. HMRPQoS solves the expansibility of the large scale networks.This paper is supported by Natioal Natural Science Foundation of China(90304018,60172035).更多还原