Network traffic classification method based on K_means and KNN fusion algorithm

Abstract

The invention puts forward a network traffic classification method based on a K_means and KNN (K-Nearest Neighbor) fusion algorithm. The network traffic classification method is characterized in that:on the aspect of frame, a binary classifier is built for each application protocol, outputs of all classifiers are integrated into a final output by a decision rule; on the aspect of algorithm, a unsupervised K_means algorithm and a supervised KNN algorithm are fused, in addition, the network traffic classification method also puts forward an iterative feature selection algorithm based on K_meansin order to select features with high separation degrees, so as to save time and space and improve a classification effect. An experiment result shows that the network traffic classification method of the invention enables both an accuracy rate and a recalling rate for traffic recognization on real traffic data to reach more than 90%, thus the network traffic classification method has a better effect in comparison with a conventional typical traffic classification method; in addition, the network traffic classification method also can recognize non-logged traffic, thus the network traffic classification method extends functions in comparison with the typical traffic classification method.

Description

Technical field [0001] The present invention relates to the fields of traffic engineering, network security and the like, in particular to the identification of network traffic classification, network management, Qos and unregistered traffic. Background technique [0002] As the foundation and supporting technology of many network research topics, network traffic classification technology is receiving more and more attention from network researchers and ISPs. In terms of network security, the emergence of a large number of applications and network services implies a variety of malicious traffic and illegal behaviors, such as network viruses, spam, and network attacks. The correct classification and identification of network traffic can not only filter out such bad information, but also have a deep understanding of the health of the current network, optimize and manage designated traffic, and protect the Internet to a certain extent, ensuring the quality of network bandwidth and a...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to solve the problem that the current network traffic classification method has a single form and low accuracy rate, and cannot identify unregistered traffic, and provides a network traffic classification method based on the fusion algorithm of K_means and KNN

Images