深入了解Internet拓扑的结构性质有利于更好地设计和发展Internet.由于Internet规模巨大,以及获得完整的路由器级Internet拓扑方面的困难,目前无法研究整个路由器级Internet拓扑.因此,分别研究每个国家级或跨国因特网服务供应商(Interne...深入了解Internet拓扑的结构性质有利于更好地设计和发展Internet.由于Internet规模巨大,以及获得完整的路由器级Internet拓扑方面的困难,目前无法研究整个路由器级Internet拓扑.因此,分别研究每个国家级或跨国因特网服务供应商(Internet service provider,简称ISP)网络拓扑结构成为了解Internet拓扑特征的一种可选方法.以中国教育科研网为例,简要描述了多点测量其路由器级拓扑结构的测量结果.分析了该实例拓扑图的节点度分布特征、较大特征值的有关性质以及谱密度分布特征.分析了该实例拓扑图的无符号拉普拉斯谱(SLS)、规格化拉普拉斯谱(NLS)以及群集系数等度量特征.分析结果表明,大型ISP拓扑确实具有某些幂律特征;不同于自治系统级拓扑的情形,对ISP拓扑的节点度补累积分布来说,幂律分布未必拟合得最好;ISP拓扑是一种无标度图,但不符合Barabási-Albert(BA)生长模型;SLS和NLS具有区分不同的路由器级拓扑结构的能力;Internet路由器级拓扑的发展可能遵循一种不同于BA模型的生长过程.展开更多
Fitness of node can denote its competing power and clustering denotes the transitivity of network. Because the fitness of node is uncertain or fuzzy in some social networks, an explicit form of the degree distribution...Fitness of node can denote its competing power and clustering denotes the transitivity of network. Because the fitness of node is uncertain or fuzzy in some social networks, an explicit form of the degree distribution on fuzzy fitness is derived within a mean field approach. It is a weighted sum of different fuzzy fitness. It can be found that the fuzzy fitness of nodes may lead to multiscaling. Moreover, the clustering coefficient of node decays as power law and clustering coefficient of network behavior not-decrease-but-increase’ phenomenon after some time. Some computer simulation results of these models illustrate these analytical results.展开更多
文摘深入了解Internet拓扑的结构性质有利于更好地设计和发展Internet.由于Internet规模巨大,以及获得完整的路由器级Internet拓扑方面的困难,目前无法研究整个路由器级Internet拓扑.因此,分别研究每个国家级或跨国因特网服务供应商(Internet service provider,简称ISP)网络拓扑结构成为了解Internet拓扑特征的一种可选方法.以中国教育科研网为例,简要描述了多点测量其路由器级拓扑结构的测量结果.分析了该实例拓扑图的节点度分布特征、较大特征值的有关性质以及谱密度分布特征.分析了该实例拓扑图的无符号拉普拉斯谱(SLS)、规格化拉普拉斯谱(NLS)以及群集系数等度量特征.分析结果表明,大型ISP拓扑确实具有某些幂律特征;不同于自治系统级拓扑的情形,对ISP拓扑的节点度补累积分布来说,幂律分布未必拟合得最好;ISP拓扑是一种无标度图,但不符合Barabási-Albert(BA)生长模型;SLS和NLS具有区分不同的路由器级拓扑结构的能力;Internet路由器级拓扑的发展可能遵循一种不同于BA模型的生长过程.
基金This project was supported by"System management",the i mportant subject of shanghai (T0502)
文摘Fitness of node can denote its competing power and clustering denotes the transitivity of network. Because the fitness of node is uncertain or fuzzy in some social networks, an explicit form of the degree distribution on fuzzy fitness is derived within a mean field approach. It is a weighted sum of different fuzzy fitness. It can be found that the fuzzy fitness of nodes may lead to multiscaling. Moreover, the clustering coefficient of node decays as power law and clustering coefficient of network behavior not-decrease-but-increase’ phenomenon after some time. Some computer simulation results of these models illustrate these analytical results.
