大数据的分布式网络入侵实时检测仿真

Simulation of Real Time Intrusion Detection for Distributed Network Based on Big Data

对大数据的分布式网络入侵的检测,能够有效遏制网络犯罪现象。对网络入侵的实时检测,需要对入侵的数据进行分类,进而对样本子集进行随机的重复采样,完成大数据下分布式网络入侵实时检测。传统方法设计三维多层空间可视化模型,开发检测分析原型,但忽略了对入侵数据进行随机采样,导致检测精度偏低。提出基于K-NN分类的分布式网络入侵实时检测方法。首先对大数据分布式网络的数据进行特征提取,采用自适应的搜索方法,利用神经网络的神经元种群间的进化差异度的变小进行特征提取,对分布式网络入侵的数据进行分类,在大数据的数据训练集中对训练样本的子集进行随机的重复采样,依据K个邻域的数据在数据类别中的数目对构建的置信指派进行加权,完成对大数据的分布式网络入侵实时检测。实验结果表明,所提方法能有效地遏制犯罪现象的发生。

In this paper, we propose a method for real-time detection of distributed network intrusion based on K -NN classification. Firstly, feature of data in big data distributed network is extracted. Secondly, the adaptive search method and diminution of evolution diversity between neural network and the neuron population are used for feature extraction and intrusion data in distributed network are classified. In training setof big data, the subset of training sample is sampled repeatedly and randomly. According to the number of data of K neighborhoods in the data categories, constructed fiducial assignment is weighted. Finally, the real-time detection of distributed network intrusion for big data is completed. Simulation results show that the proposed method can effectively restrain the crime.