一种面向军事物联网的网络流量异常检测模型

A Network Traffic Anomaly Detection Model for Military Internet of Things(MIOT)

针对军事物联网中网络流量数据日趋复杂,数据特征维度高等特点,将卷积神经网络算法应用到网络流量分析领域。根据数据特点,构建出一种基于无池化层改进型卷积神经网络(NPCNN,No Pooling CNN)的网络流量异常检测模型。采用Modbus、NSL-KDD和KDDCup99数据集对NPCNN网络流量异常检测模型进行验证,同时将NPCNN网络结构同传统的卷积神经网络对比,通过对实验结果的分析发现,该模型在军事物联网网络流量异常检测中具有可行性和可扩展性。同时NPCNN网络在准确率性能方面优于传统的卷积神经网络,为军事物联网网络流量分析提供一种可靠方法。

In view of the increasing complexity of network traffic data and the high dimension of data characteristics,convolutional neural network algorithm is applied to network traffic analysis in the military internet of things.According to the characteristics of the data,a network traffic anomaly detection model based on no-pooling improved convolutional neural network(NPCNN,No Pooling CNN)is constructed.Modbus,NSL-KDD and KDDCup99 data sets are used to verify the network traffic anomaly detection model based on NPCNN,then the NPCNN network structure is compared with the traditional convolutional neural network.Through the analysis of the experimental results,it is found that the model proposed is feasible and expandable in the network anomaly detection for military internet of things.Meanwhile,NPCNN network,which is superior to traditional convolutional neural network in accuracy rate,provides a reliable method of traffic analysis for military internet of things.