集成模型在网络入侵检测中的仿真研究

Integrated Network Intrusion Detection Model in Simulation

研究保证网络安全问题,针对网络入侵具有多样性和复杂性,信息冗余十分严重,传统检测方法不能很好消除冗余信息,导致检测时间长和检测正确率低的难题。为了提高检测准确性,将主成分分析和RBF神经网络相结合起来,组成一个集成的网络入侵检测模型。模型首先通过主成分析分析法对网络原始数据进行预处理,降低特征维数、消除冗余信息,将处理后特征作为神经网络的输入,网络入侵类型作为神经网络的输出,建立RBF神经网络入侵检测模型对网络数据进行检测。在Matlab平台上,采用权威网络入侵数据DARPA数据集对集成模型进行预试,仿真结果表明,集成模型的网络入侵检测正确率高于传统入侵检测模型,加快了网络入侵检测速度,为网络入侵提供了一种实时检测方法。

Study the problem of network security.Network intrusions are of diversity and complexity,and have redundant information,the traditional neural network intrusion detection methods have the disadvantages of complicated network structure,long training time and low accuracy.In order to improve the network intrusion detection rate and the network security,the principal component analysis and RBF neural network are combined and formed an integrated network intrusion detection model.The network intrusion data are pretreated by principal component analysis to reduce the characteristic dimension,eliminate redundant information and reduce the RBF neural network input.Then pretreatment features are used as neural network＇s inputs,network intrusion types are used as neural network＇s outputs,and the RBF neural network intrusion detection model is established.Finally,network data were detected by using the network intrusion detection model.In Matlab,the integrated model is tested by the DARPA network intrusion datasets.Simulation results show that the integration model accuracy is higher than the traditional network intrusion detection method,the mistake examining rate is reduced,and the network intrusion detection speed is speeded up,It is a real-time detection tool for the network intrusion detection.