网络拓扑识别:基于traceroute的层析成像方法  

Network topology inference:tomography method based on traceroute

作　　者：廖海亮[1] 胡光岷[1] 钱峰[1] 杨志豪[1] 

机构地区：[1]电子科技大学宽带光纤传输与通信网技术教育部重点实验室,成都610054

出　　处：《计算机应用研究》2009年第1期276-279,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(60572092)

摘　　要：针对传统的网络拓扑识别方法(如traceroute)无法完成包含不协作节点的拓扑识别以及基于网络层析成像技术的拓扑识别方法的复杂性和不确定性问题,提出一种基于traceroute的层析成像技术的拓扑识别方法。该方法可通过提出的最小相似度聚类算法和匿名节点构造归并算法,将网络层析成像获得的拓扑信息与trace-route探测结果融合,构成最终的拓扑结构。NS2的仿真表明,该方法不仅可识别包含不协作节点的网络拓扑,且所使用的探测包的数量也大大减少。To the problem in the traditional network topology （for instance, traceroute ） that the network topology which contained uncooperative touters couldn＇t be inferred and the problem of complexity and indetermination in topology infe-rence based on tomography, this paper introduced a network topology inference method based on traceroute and tomography. This method combined the traceroute result with the topology information obtained by tomography to construct a final topology using minimum similarity cluster algorithm and merging and constructing anonymous routcrs algorithm. The simulation on NS2 shows that this method not only be able to infer the network topology which contains uncooperative routers, but also enormously reduce the number of probes.

关 键 词：跟踪路由 匿名路由 网络层析成像 三明治包 聚类分析 递归算法 

分 类 号：TP393.02[自动化与计算机技术—计算机应用技术]