Network Traffic Classification Based On Deep Learning And Research On Intrusion Detection

At present,with the new round of scientific and technological revolution and accelerated industrial evolution,new technologies and new applications such as artificial intelligence,big data,and the Internet of Things have developed rapidly.The Internet has ushered in stronger development momentum and broader development space.The field of network security is also facing more newer and more complex challenges.Therefore,it is of great significance to research on network security technologies.As an important branch of network security research,intrusion detection can effectively discover intrusion behaviors by analyzing computer systems and key information on the network,so as to intercept and respond to attack behaviors.There are many methods for intrusion detection.With the rise of deep learning,more efficient and accurate identification methods have been provided for intrusion detection.Intrusion detection technology is essentially the identification and differentiation of normal behavior and intrusion attack behavior,so it can turn intrusion detection problems into classification problems.Based on the spatial characteristics of network traffic data and the characteristics of convolutional neural networks to extract image features,this paper first designs a network traffic classification model based on CNN-IDS.Then according to the time characteristics of network traffic data,combined with the characteristics of cyclic neural network suitable for processing sequence data,the network model based on LSTM-IDS is designed.It is found through experiments that both classification models can produce high recognition accuracy,but learning the spatial characteristics of the network traffic data or the temporal characteristics of the network traffic data alone is not sufficient for the characteristics of the data.Therefore,in view of the spatiotemporal dual characteristics of network traffic data,an intrusion detection model-CNN-LSTM-IDS is proposed,which combines convolutional neural network and recurrent neural network.The data is first preprocessed and input into a convolutional neural network to learn the spatial characteristics of the traffic data.Then,the learned features are input into a long-term and short-term memory network for learning the timeseries features of the traffic data.The Experiments show that the performance of the hybrid structure model is improved compared to using the convolutional neural network and long-term short-term memory neural network alone.In addition,the CNN-LSTMIDS intrusion detection model is compared with other algorithms.Experimental results show that the algorithm has better performance indicators than other algorithms,and provides a new idea for intrusion detection research.