A Classification Method for Encrypted Traffic Based on Dual-Channel Convolutional Neural Network

Abstract

The invention discloses a method for classifying encrypted traffic based on a dual-channel convolutional neural network. Network streams are represented in two dimensions based on network data packet header attributes and time-segmented low-order statistical features, and then through dual-channel convolution The neural network learns encrypted traffic features from two kinds of network flow representations at the same time, and realizes the classification of application types carried on encrypted traffic. The present invention makes full use of the local and overall advantages of data packet header attributes and traffic statistical features without the need for expert knowledge, automatically learns traffic characteristics from two dimensions, and improves the classification accuracy of encrypted network traffic; and only uses data packet headers The field information in the section has strong generalization ability and will not violate the data confidentiality and privacy protection policies.

Description

technical field [0001] The invention belongs to the field of network security and information technology, and in particular relates to an encrypted traffic classification method based on a dual-channel convolutional neural network. Background technique [0002] Network security and privacy protection are increasingly becoming the focus of enterprises and network users. More and more enterprises choose VPN to protect the security of enterprise data transmission. Network users use various encryption and anonymous communication technologies to protect network terminals and personal privacy. . And these technologies are also used by criminals to engage in illegal network activities to evade network security monitoring. Traffic masquerading and obfuscation technologies are widely used. Methods such as protocol encapsulation and traffic proxy encapsulate one encrypted traffic in another encrypted traffic, changing the original characteristics of encrypted traffic. Bearer traffic...

AI Technical Summary

Problems solved by technology

However, different network flow representations are usually formed by network flow preprocessing methods based on different dimensions or granularities, and there is a problem of inconsistency in size.

Images