面向分布式网络入侵检测的实验测试仿真

Experimental Simulation Testing for Distributed Network Intrusion Detection

针对目前方法对分布式网络进行入侵检测时,由于未对网络数据进行去噪处理,导致方法存在攻击正确识别比例低、样本正确分类比例低以及检测性能差的问题,提出面向分布式网络入侵检测的实验测试方法。所提方法首先依据小波变换方法对网络数据进行去噪处理,依据去噪结果提取网络数据特征;通过获取的数据特征建立网络入侵数据的马尔可夫检测模型,依据上述模型确定数据的攻击行为;最后使用上述模型完成网络数据的入侵检测。实验结果表明,运用上述方法进行网络入侵检测时,攻击正确识别比例高、样本正确分类比例高、检测性能好。

Due to low proportion of correct identification for attacks and correct classification for samples in current methods,this paper proposed an experimental method for distributed network intrusion detection.Firstly,the noise was removed from network data according to the wavelet transform method.Based on the denoising results,the network data features were extracted.On this basis,Markov detection model of network intrusion data was built.Secondly,the model was used to determine attack behaviors of data.Finally,the model was adopted to complete the intrusion detection for network data.Experiment results show that the proportion of correct identification for attacks is high after using the proposed method,and the proportion of correct classification for samples is also high.In addition,the detection performance of the method is good.