Mobile application traffic identification method based on multi-layer classifier

Abstract

The invention belongs to the field of network traffic analysis, and aims at the problem that the existing mobile application traffic identification method cannot detect and process background traffic, and provides a mobile application traffic identification method based on a multi-layer classifier. The technical scheme is as follows: the first step is to extract traffic The features of the training set are used to obtain the feature representation of the traffic samples; the second step is to train the first layer classifier to initially detect the samples to be detected as target traffic or background traffic; the third step is to train the second layer classifier to classify the target traffic Perform fine-grained identification; the fourth step is to train the third-layer classifier; the fifth step is to use the trained multi-layer classifier to perform mobile application traffic identification on the samples to be detected. The present invention fully considers the traffic distribution in the real network. In the absence of a complete background traffic data set, the characteristics of the target traffic sample are learned layer by layer, so that the classifier can also exclude the background traffic while identifying the target traffic. , to reduce the false positive number of the classifier.

Description

technical field [0001] The invention belongs to the field of network traffic analysis, and relates to a machine learning-based network traffic identification method, in particular to a multi-layer classifier-based mobile application traffic identification method. Background technique [0002] With the popularity of mobile devices and the prosperity and development of mobile applications, mobile applications have become the most commonly used way of surfing the Internet. As of the first quarter of 2018, there were 3.8 million applications available for users to download in Google's application market, and an average of 6,140 new applications were added every day. As of 2017, 57% of all web traffic comes from mobile devices. As a result, mobile network traffic has surpassed traditional workstation traffic to become a major component of network traffic. The focus of research has also shifted from traditional workstation traffic identification to mobile network traffic identif...

AI Technical Summary

Problems solved by technology

[0013] Aiming at the problem that the existing mobile application traffic identification method based on machine learning cannot detect and process background traffic, the present invention provides a mobile application traffic identification method based on a multi-layer classifier, which learns the characteristics of target traffic samples layer by layer, so that the classifier While identifying the target traffic, it can also exclude the background traffic and reduce the false positive number of the classifier

Each flow sample is denoted as flow

Images