A New Method of Improving the Traditional Traffic Identification and Accuracy

Abstract

Abstract As the traffic generated by the increasing number of applications on the Internet is becoming more and more complex, how to improve the quality of service and security of the network is also increasingly important. This paper studies the application of Support Vector Machine (SVM) in traffic identification to classify network traffic. Through data collection and feature generation methods and network traffic feature screening methods, SVM is used as a classifier by using the generalization capability, and the parameters and the kernel functions of SVM are adjusted and selected based on cross comparison ideas and methods. Using the cross-validation method to make the most reasonable statistics for the classification and recognition accuracy of the adjusted support vector machine avoids the situation that the classification accuracy of the support vector machine is unstable or the statistics are inaccurate. Finally, a traffic classification and identification system based on SVM is implemented. The final recognition rate of encrypted traffic is up to 99.31%, which overcomes the disadvantages of traditional traffic identification and achieves a fairly reliable accuracy.