基于协议分析与PNN组合的网络异常检测  

Network Anomaly Detection Combined with Protocol Based Analysis and PNN

作　　者：彭海深[1,2] 

机构地区：[1]中国科学技术大学,安徽合肥230026 [2]重庆电子工程职业学院,重庆401331

出　　处：《计算机与网络》2010年第20期51-53,共3页Computer & Network

基　　金：国家863计划(编号2006AA01Z452)

摘　　要：提出了一种基于协议分析与概率神经网络结合的网络异常状况检测方法。该方法首先基于网络协议分析,对网络运行中的敏感数据进行捕获及状态扫描;然后结合Bayes最小风险准则和基于Parzen窗的概率神经网络(PNN),对网络特征数据与网络基线数据进行比较判断,从而及时、准确地检测出网络发生的异常状况。This paper proposes a network anomalies monitoring method based on protocol analysis and combined with probabilistic neural network. This method is used to perform capture and status scanning for sensitive data during the network running based on network protocol analysis, and compare and judge the network feature data and network baseline data combined with the minimum Bayes risk rule and probabilistic neural network （PNN） which based on Parzen window, thereby detect the network anomalies timely and accurately.

关 键 词：网络异常检测 协议分析 概率神经网络 

分 类 号：TP393.06[自动化与计算机技术—计算机应用技术]