【作者】 王珩；

【导师】 孙亚民；

【作者基本信息】 南京理工大学 ， 计算机应用技术， 2004， 博士

【摘要】 随着网络技术的飞速发展，当前通信网络带宽和处理能力的提高使网络能够提供更多的多媒体业务，也使得支持“点到多点”或“多点到多点”的组播通信方式成为网络支持多媒体业务的必要形式。组播路由是网络层具备的功能，组播问题的关键在于组播路由的确定，寻找简单、高效、健壮的组播路由算法一直是网络界致力研究但未完全解决的问题。另一方面，许多分布式的多媒体应用对时延、时延抖动、带宽以及包丢失率有不同的要求，这需要当前网络能够传送具有这些QoS要求的实时多媒体信息。因此，作为QoS为中心的网络体系结构中不可缺少的组成部分，基于QoS约束的组播路由算法的研究成为网络研究领域的重要内容和热点问题。 本文主要研究基于QoS约束的组播路由算法，针对一些典型的具有NP难度的QoS组播路由问题，提出几种简单、有效、实用的QoS组播路由启发式算法。主要研究工作和取得的成果如下： (1)实现了一个通用、简单、开放性强的QoS路由仿真器QRSIM，为后续章节提出的几种QoS路由算法的性能测试构建出真实、准确的仿真平台。而且，只要使用者按照规范的接口编写新的QoS路由算法程序，该路由仿真器就能动态加载这些路由算法进行仿真实验； (2)针对时延约束的最小代价组播路由问题，提出四种时延约束组播源路由算法。其中，LRDLMA算法是基于拉格朗日松弛方法解决该问题，对构建的封闭图采用Prim最小生成树算法，进行拉格朗日松弛，求得高质量的可行解。仿真实验表明LRDLMA算法代价性能较好，接近性能很好的BSMA算法，并具有时延稳定、运行速度快的特点；TSESMA和TSPSMA算法是两种基于禁忌搜索方法的路由算法，但是在邻域解集的构造上有所不同，前者采用基于链路交换的思想，而后者使用路径交换策略。仿真结果表明两种算法的稳定，可靠性高，收敛速度快，有效地降低路由计算时间。TSPSMA代价较低，优于BSMA算法，对提高网络效率，优化网络资源起到很好的作用。LODMA算法是从最小时延树开始，利用链路优化策略来寻求满足条件的组播树，具有快速，时延低的特点，适合于对时延要求比较高的实时多媒体业务； (3)将模拟退火思想引入组播路由计算中，提出一种基于模拟退火方法的时延及时延抖动约束的最小代价组播路由算法SADVMA。算法采用路径交换策略在可行解范围内构造邻域集，避免了搜索区域的扩大和计算时间的增加。仿真实验表明算法的可行性、有效性和稳定性，具有代价低、摘要博士论文 收敛快的特点;(4)研究多点到多点组播路由问题，根据解决问题的策略不同，提出两种解 决时延约束的多共享组播树问题的算法:SCA算法和分布式算法DISA。 仿真实验表明，SCA算法在保证中心数不增加的条件下，有效地减少运 行时间;与同类算法相比，DJSA算法所获得的中心数较少，显著降低 了共享树的管理开销。更多还原

【Abstract】 With fast development of network technologies, increase of network bandwidth and processing power makes the network provide more multimedia applications, and also makes the multicast communication that supports "one-to-many" or "many-to-many" become a necessary mode of multimedia services. A fundamental issue in multicast communication is how to determine an efficient multicast routing, and finding simple, effective and robust multicast routing algorithms is unsolved problem in network fields. In addition, many distributed multimedia applications have various demands on delay, delay variation, bandwidth and packet loss, which requires current network to transmit real-time multimedia information with these quality-of-service (QoS) constraints. So, as an indispensable component in a QoS-centric network architecture, research on multicast routing algorithms based on QoS constraint becomes an important part and hotspot issue of network research fields.This dissertation concentrates on researching multicast routing algorithms with QoS constraints, proposes several simple, effective and practical QoS multicast routing heuristic algorithms to solve some classical multicast routing problems with NP difficulty. Generally, the main achievements in this paper are as follows:(1) Develop a versatile, simple, and open QoS routing simulator (QRSIM), which provides a real and exact simulator platform for evaluating performance of proposed QoS routing algorithms. Also, as long as user presents new QoS routing algorithms programming according to standard interfaces, QRSIM can dynamically load these routing algorithms to simulate.(2) Propose four delay-constrained multicast source routing algorithms to solve delay-constrained least-cost multicast routing problem. LRDLMA makes use of the characteristic of Lagrange relaxation method, and finds multicast tree satisfying constraint by constructing closure graph and using Prim algorithm to make relaxation to this graph. A large number of simulations demonstrate that cost performance of the algorithm is close to BSMA algorithm whose performance is best, and it has characteristics of stable delay and quickness. TSESMA and TSPSMA algorithm are two routing algorithms based on Tabu search, and they take different approach to construct neighbors. TSESMA isbased on "edges-switching", and TSPSMA uses "paths-switching" strategy. Simulations show that two algorithms performs stable performance, high reliability, rapid convergence and reduce computing time. TSPSMA performs excellent performance of cost, and its cost is lower than BSMA algorithm’s, which can improve network efficiency and optimize network resource. LODMA algorithm begins from a least-delay tree, then gets final multicast tree satisfying delay constraint by using "link optimizing" strategy. This algorithm has faster execution time and lower delay than other heuristics, and it is fit for those real-time multimedia applications with higher delay demands.(3) Apply the simulated annealing to multicast routing, and propose a multicast routing algorithm (SABDMA) based on simulated annealing to solve delay and delay variation constrained least-cost multicast routing problem. To avoid enlargement of search area and increase of computing time, SABDMA uses "paths-switching" strategy, which constructs neighbor set in the range of feasible solutions according to the relationship between delay and delay variation. Simulations demonstrates that the algorithm has characteristics of feasibility, stability and rapid convergence, and it can effectively construct multicast tree with lower cost according to QoS request, and has better real-time property.(4) Study many-to-many multicast routing problem and propose two heuristic algorithms to solve delay-constrained multiple-shared multicast tree problem: SCA algorithm and distributed heuristic algorithm DISA. The simulations show SCA algorithm consumed smaller run-time than others, without increasing number of centers. DISA algorithm has the best performance o更多还原