计算机工程与应用 ›› 2008, Vol. 44 ›› Issue (7): 144-147.

• 网络、通信与安全 • 上一篇    下一篇

季节时间序列法在IP网异常流量检测的应用

曾国鉴1,鲁士文2,王卫东3   

  1. 1.中国科学院 研究生院,北京 100039
    2.中国科学院 计算技术研究所,北京 100080
    3.中联绿盟信息技术(北京)有限公司,北京 100089
  • 收稿日期:2007-11-15 修回日期:2008-01-31 出版日期:2008-03-01 发布日期:2008-03-01
  • 通讯作者: 曾国鉴

Anomaly detection application on IP backbone using time series forecast

ZENG Guo-jian1,LU Shi-wen2,WANG Wei-dong3   

  1. 1.Graduate School,Chinese Academy of Sciences,Beijing 100039,China
    2.Institute of Computing Technology,Chinese Academy of Sciences,Beijing 100080,China
    3.NSFOCUS Information Technology Co.,Ltd.,Beijing 100089,China
  • Received:2007-11-15 Revised:2008-01-31 Online:2008-03-01 Published:2008-03-01
  • Contact: ZENG Guo-jian

摘要: 针对TCP/IP骨干网,利用NetFlow技术,提出一种基于业务流量周期规律特性的建模与异常检测方法。该方法通过挖掘骨干网主要业务流量的规律性,结合时间序列分析方法,有效地预测流量的变化趋势,避免了对复杂的流量非线性趋势进行建模分析。

Abstract: An approach,using Netflow technology,to model service traffic and anomaly detection on IP backbone in the perspective of periodic traffic feature is proposed.This method avoids the difficulty to model the uncertainty and nonlinear facts in traffic data and provides more accurate information than the thresholds-based detection method.