中国民航大学学报

• 工程技术 • 上一篇    下一篇

一种面向TCP 流的异常检测技术

焦文欢a,冯兴杰b   

  1. (中国民航大学a.信息网络中心;b.教务处,天津300300)
  • 收稿日期:2013-07-09 修回日期:2013-09-02 出版日期:2014-06-24 发布日期:2014-10-31
  • 作者简介:焦文欢(1987—),男,河南安阳人,助理工程师,硕士,研究方向为信息安全、智能信息处理理论与技术.
  • 基金资助:

    国家自然科学基金项目(60672174);中国民用航空局科技基金项目(MHRD200807)

Anomaly detection technique oriented to TCP flow

JIAO Wen-huana,FENG Xing-jieb   

  1. (a. Information and Internet Center;b. Department of Academic Affairs,CAUC,Tianjin 300300,China)
  • Received:2013-07-09 Revised:2013-09-02 Online:2014-06-24 Published:2014-10-31

摘要:

为了解决传统TCP流异常检测中无法区分突变正常流量和异常流量的问题,提出了一种面向TCP流的异常检测技术。模型采用概要数据结构,在基于小波求解Hurst 参数判断网络异常的基础上,根据TCP 握手和挥手协议的内容对TCP 报文的完整性进行验证,不仅实现了大规模数据的概要存储,提高了运算效率,而且实现了对突变的正常流量与异常流量进行区分的目的,实验结果表明,该方法具有较高的检测率和较低的误检率,可较好地应用于网络异常检测。

关键词: 突变正常流量, 异常流量, 概要数据结构, 小波系数方差, 报文完整性

Abstract:

In order to solve the problem that network-flow anomaly detection cannot distinguish between mutations in normal and abnormal traffic flow,an anomaly detection technique oriented to TCP flow is presented. Using method of summary data structure,this model calculates the Hurst parameter and TCP packet integrity to discover the network anomalies. Not only can it achieve a summary of large-scale data storage to improve the efficiency,but also realize the purpose to distinguish between the burst traffic with abnormal tratfic flow.Experimental results show that this method has high detection rate and low false detection rate,it can be better used in network anomaly detection.

Key words: burst traffic, abnormal traffic, synopsis data structure, wavelet variance, message integrity

中图分类号: