基于分形特征的网络异常检测方法研究  

A Novel Approach Based on Fractal Characteristics for Anomaly De tection

作　　者：刘峰[1] 胡昌振[1] 帅艳民[2] 

机构地区：[1]北京理工大学机电工程学院,北京100081 [2]北京师范大学地理与遥感科学学院,北京100875

出　　处：《计算机工程与应用》2004年第22期34-36,共3页Computer Engineering and Applications

基　　金：国家863高技术研究发展计划支持项目(编号:2002AA4Z3420)

摘　　要：网络连接同许多其它现象一样,呈现出一种统计自相似性。论文根据此特征提出了一种基于分形特征的网络异常检测方法,该方法能够根据网络连接分形维数的变化,有效地检测网络拒绝服务攻击和端口扫描等异常情况,并通过模拟实验证明了该方法的可行性。Just as many other phenomena,network connection takes on a kind of characteristic of statistical self-simi-larity.Based on this intrinsic factor,a new anomaly detection approach using fractal theory is proposed in this paper.The method can efficiently detect port scan and network DoS attacks according to the change of network connection's fractal dimension.Simulated tests given in the paper have proved its feasibility and applicability.

关 键 词：网络安全 分形 自相似 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]