改进的网络拓扑推断算法

摘要

Due to the larger scale of network,network management is more and more difficult,and network security is also suffered a certain threat.In order to obtain the network topology structure that closed to the real network,and have a better management to network,combining traditional network topology inference with NT technology to infer the network tolology structure.The level information of nodes will be got by the traditional method,and then the delay variation will be gained through sending 4-tuple packets of NT technology.According to analyze the character of Depth first search sequence and computing the delay variation covariance to confirm the relationship of terminal nodes, and then network topology will be infered accroding to the nodes relationship.Finally it proves the improved algorithm has the higher accuracy through NS2 simulation experiment.%网络规模的日益壮大使网络管理成为困难，网络安全也会受到一定的威胁，为了推断出更接近真实的网络拓扑结构，更好的对网络进行管理，将传统的网络拓扑推断算法与网络断层扫描技术相结合，准确的推断网络拓扑结构。根据传统网络测量方法获得节点的层次信息，通过NT技术中的四元分组列车方法获得目标节点的时延抖动，分析深度优先搜索序列的特点并计算节点间时延抖动协方差来判断节点间的相关性，进而推断出网络的拓扑结构。经过NS2仿真模拟验证表明，该算法具有更高的准确性。