Research On Kev Technology Of Encrvption Traffic Application Type Identification Based On Deep Learning

At present,the proportion of encrypted traffic in Internet traffic has exceeded 90 percent.However,the traffic encryption technology also brings many problems to protect the freedom,privacy and anonymity of network users.Attackers can use the method of traffic encryption to hide themselves,thus bypassing some security precautions to attack,and the network security is seriously threatened.The existing traffic classification technology is outdated.How to correctly and effectively identify the type of encrypted traffic without encryption and decryption traffic is the key to the research in the field of encrypted traffic classification.Researchers began to try to use deep learning to practice in the field of encrypted traffic classification,and found the great potential of deep learning.Therefore,in order to solve the problem of multiclass classification of encrypted traffic,this paper proposes a classification model of encrypted traffic based on deep learning.On this basis,an application system that can classify encrypted traffic is designed and implemented.The main work of this paper is as follows:(1)An encrypted traffic classification model based on SwinT-CNN is proposed.This model combines the improved CNN model with the core module of the Swin Transformer neural network framework to improve the accuracy of the model.The improved CNN model can extract the local spatial information features of encrypted traffic.The multi-head attention mechanism of Swin Transformer can obtain the global attention information of encrypted traffic.The encrypted traffic classification model based on SwinT-CNN is trained and tested on encrypted traffic data sets,and the accuracy rate reached 96.2%.The superiority of the model is proved by multi-model comparison experiment,and the necessity of the module is confirmed in the ablation experiment.(2)An encryption traffic classification system based on SwinT-CNN classification model is designed and implemented.The main work modules of the system include WEB module,data preprocessing module,task scheduling module and encrypted traffic classification module,which realizes the classification function of encrypted traffic application types,and displays the results visually.On this basis,the system is tested and can correctly classify the encrypted traffic,and the accuracy is 95.8%.