【作者】 杨云； 高飞；

【Author】 Yun Yang, FeiGao ( Department of Computer Science, Yangzhou university, Yangzhou 225009,China) ( Department of Computer Science, Nanjing University of Science and Technology, Nanjing 210004, China)

【机构】 扬州大学信息工程学院计算机科学与技术系；

【摘要】 随着网络规模和复杂度的快速增长,对其相关技术的研究也越来越多。在网络技术研究中,拓扑建模是一项很重要的技术,其研究内容可归结为:如何生成更具代表性、与现实网络(如Internet)更接近的网络拓扑结构,供网络仿真实验使用。如今,在拓扑建模的研究中,Internet 作为一个大型网络的应用实例,已成为研究热点,受到了广泛的关注。而且,Internet在AS (Autonomous System)级上的拓扑结构对于外部网关协议的分析、设计与网络仿真具有重要的意义。从1995年开始的大规模Internet拓扑测量工作已经逐步展开,采集到了大量的拓扑数据(BGP)。这些数据对科研机构都是免费开放的,极大地推动了拓扑建模研究的发展。1996年,Doar提出了 Tiers模型,Zegura等人提出了Transit-stub模型,这些都属于层次型的网络拓扑模型。其中 Transit-stub模型在最近的网络仿真实验中得到了大量的应用。1999年,Faloutsos等人发现大型网络拓扑结构中存在着幂率(power-law)。幂率的发现将拓扑建模与生物学、社会学中的复杂网络联系起来,使其成为无尺度(scale-free)网络的一个实例。更多还原

【Abstract】 The tremendous growth of the Internet in both size and complexity has stimulated considerable research interest in the network. Network topology modeling is the foundation of network performance research. The aim of topology modeling is to generate the topology which is representative and esemble to real network. So we can use the topology in the correlative network simulation experiment. Now, Internet topology modeling has been widely concerned. The Internet topology on AS-level is significant to network protocol analysis and simulation experiment. The Internet topology measuring has begun since 1995. The large amount of measuring data which is free to researchers fiercely boosts the Internet topology modeling. In 1996, hierarchical topology model was devised (such as Tiers and Transit-stub). Transit-stub model is widely used in some network simulation experiment. After it was discovered that AS-level topology exhibits power law degree distribution, Internet and complex network in biology have been regarded as a instance of scale-free network. In this paper, by integrating the power-law degree distribution and the hierarchical structure, we present a hierarchical network topology generating algorithm for power-law HIPL. In HIPL, we try to produce power law degree distribution in each tier. There are two manners to generate topology which observes power law: (1) determine the model parameters, and set out-degree of every node as the form of power function; (2) By simulating the way in which real network grows, add nodes and links to the initial topology. When add new nodes, we follow the ’linear preferential’ rule. We use the second manner in HIPL. The HIPL is verified by simulation. With the parameters identified from real Internet data, HIPL capture the hierarchical structure and power-law degree distribution well. The experiments concern with the topological properties such as clustering coefficient, diameter, average degree and so on.更多还原