Abstract: This paper presents an identification method of network topology based on end-to-end unicast measurements. For data measurement, it obtains delay difference by improving the sandwich group to infer from the relevant node. For the topology inference, the reference to the tree depth-first node of the relevant sequence comparison, it uses a combination of iterative and recursive methods to build the network topology to reduce the amount of probes sent. Simulation results show that the topological identification method is effective.

Key words: network tomography, end-to-end measurement, sandwich grouping, queuing delay, topology inference

摘要： 提出一种基于端到端单播测量的网络拓扑识别方法。在数据测量方面，采用改进的三明治分组探测机制获取往返时延差，进行节点相关性推断。在拓扑推断方面，以树的深度优先序列作为参考，通过比较节点对的相似性，采用迭代与递归相结合的方法反向构建拓扑结构，减少探针发送量。仿真实验结果证明了该拓扑识别方法的有效性。

关键词: 网络层析成像, 端到端测量, 三明治分组, 排队时延, 拓扑推断

CLC Number: 

• TP393