论文部分内容阅读
针对现有方法无法对大规模协同性域间路由事件进行有效检测的问题,提出一个基于时空张量的分布式域间路由分析框架.首先从BGP路由表中提取域间路由的时空特征,并使用MapReduce框架生成时空张量,用于表征域间路由的动态变化,然后使用张量分解的方法从时空张量中提取二维事件,最后运用矩阵聚类的方法生成三维域间路由事件.将该分析框架运用于不同时间跨度和时间粒度的路由表数据集,实验结果表明,该方法不仅能够准确检测局部域间路由事件,而且能够从时间和空间尺度上对全球域间路由事件进行有效剖析,揭示了域间路由事件具有影响范围广、持续时间短等一般规律.
Aiming at the problem that existing methods can not effectively detect large-scale collaborative inter-domain routing events, a distributed inter-domain routing analysis framework based on space-time tensor is proposed. First, the spatio-temporal characteristics of inter-domain routing are extracted from the BGP routing table Using the MapReduce framework to generate the space-time tensor, which is used to characterize the dynamic change of the inter-domain routing, then extracting the two-dimensional events from the space-time tensor using the method of tensor decomposition, and finally using the method of matrix clustering to generate the inter-domain routing events. The results show that this method not only can detect the inter-domain routing events accurately, but also can effectively dissect the inter-domain routing events from time to space and reveal Inter-domain routing events have a wide range of impact, short duration and other general laws.