【摘要】 随着Internet规模的不断扩大,Internet拓扑模型已成为网络领域的研究热点。从最初的Waxman随机模型到基于幂律[1]特征提出的BA模型、GLP模型、PFP模型,都试图去发现和解析Internet拓扑演化的规律。节点度、聚簇系数、介数等拓扑度量分别被提出,用于刻画拓扑特征。引入统计物理学上非广延熵[2]作为一个新的拓扑度量,从概率分布的均匀程度上刻画拓扑模型的节点度分布的特征。为了定量地分析非广延熵在不同模型上的差异,通过从CAIDA获得最新的真实拓扑数据,比较BA模型和Waxman模型与真实拓扑数据在非广延熵上是否具有一致性,来验证非广延熵作为拓扑度量的合理性。最后指出BA模型的非广延熵接近真实的Internet拓扑,从而也说明非广延熵作为一种新的拓扑度量的可行性。更多还原

【Abstract】 With the size of Internet growing fast,the topology model of Internet is becoming an appealing point in research community.People make try to put deep insight into the nature of Internet topology,by constructing more models from initial Waxman random model to BA,GLP,and PFP.A number of topology metrics were proposed,including node degree,cluster,and betweeness,to describe the topology characters.Non-extensive entropy,which is employed extensively in statistical physics,was introduced for a new Internet topology metric in this paper.It is used to depict the purity of node degree distribution.In order to evaluate the non-extensive entropy as topology metric,we analyed the consistency of BA,Waxman and real topology generated from CAIDA,by their non-extensive entropies.Finally,we argued that BA is more in accordance with real Internet topology by comparing their non-extensive entropy,which confirms that non-extensive is fit for topology metric.更多还原