Scholarlink

We carry out an effective clustering of distributed time series graphs. The work uses singular value decomposition (SVD) of spectral gaps of graphs derived from BGP data found at the RouteViews Project at the University of Oregon. We detect extremal events and examine anomalous events that cause disruptions in Internet routing. We use spectral gaps to examine how redundant and sparse the graph is. We base the anomaly detection method in BGP networks on the fact that a drop in the value would signify a lack of this redundancy, which in the distributed redundant Internet would be an anomalous event.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Novel Graph Spectra Approach to Anomaly Detection in Border Gateway Protocol

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Leigh Metcalf,
  • Will Casey,
  • Timur Snoke,
  • Heeralal Janwa,
  • Shirshendu Chatterjee,
  • Ernest Battifarano

摘要

We carry out an effective clustering of distributed time series graphs. The work uses singular value decomposition (SVD) of spectral gaps of graphs derived from BGP data found at the RouteViews Project at the University of Oregon. We detect extremal events and examine anomalous events that cause disruptions in Internet routing. We use spectral gaps to examine how redundant and sparse the graph is. We base the anomaly detection method in BGP networks on the fact that a drop in the value would signify a lack of this redundancy, which in the distributed redundant Internet would be an anomalous event.