QUIC traffic classification method based on multi-modal deep learning

Abstract

The invention discloses a QUIC traffic classification method based on multi-modal deep learning, which comprises the following steps: S1, preprocessing QUIC traffic, carrying out shunting processing on the QUIC traffic to be classified to obtain bidirectional flow samples, and extracting flow statistical characteristics and flow payloads of the bidirectional flow samples, S2, converting the flow payloads into images, extracting fields of the flow payloads, converting the fields into grey-scale map samples, and S3, classifying the grey-scale map samples; the invention relates to the technical field of network communication and machine learning. According to the QUIC flow classification method based on multi-modal deep learning, the flow statistics characteristics and the time sequence characteristics of the network flow can express the time structure relation of the whole flow message, and the effective load of the flow can express the content characteristics of each message; the two different emphasis dimensions of the whole and details are embodied in the traffic features. The isomerism of the two different traffic information modes can be utilized, and the traffic features can be better restored.

Description

technical field [0001] The invention relates to the technical field of network communication and machine learning, in particular to a QUIC traffic classification method based on multimodal deep learning. Background technique [0002] Currently, encrypted traffic classification methods generally include: traditional methods, machine learning methods, and deep learning methods. The traditional method is to rely on the port number in the header of the data packet, the behavior pattern of the traffic, etc. to classify. [0003] The traditional traffic classification method has very high recognition efficiency and accuracy in the traditional network environment, but as more and more encrypted traffic protocols appear in the network, and a large number of network applications begin to use dynamic port technology, the traditional traffic classification method gradually loses its advantages . Machine learning methods and deep learning methods are to classify traffic by learning de...

AI Technical Summary

Problems solved by technology

Most of the existing machine learning and deep learning traffic classification methods only consider the traffic information of the modality, and cannot make full use of the heterogeneity between different modal information of the traffic.

Images