Design Of Terminal Application Data Flow Recognition System Based On Decision Tree

With the rapid development of network technology and the rapid popularization of smart phones,mobile applications are becoming more and more abundant.Data flow identification technology has always played a very important role.Operators need to use this technology to provide Qos services,and security departments need to use this technology to identify traffic.However,with the abundance of mobile applications and the application of various new technologies,the development of this technology has encountered new difficulties.Data flow recognition technology has mainly gone through three stages of development,from the initial use of application ports for classification,to the later DPI recognition technology based on feature string matching and recognition,and then to the popular use of machine learning algorithm for traffic classification in recent years.Although the use of machine learning for traffic recognition can achieve high recognition accuracy and solve the problem of the failure of DPI recognition technology caused by data stream encryption,how to apply it in the context of smart phones is a problem worth thinking about.This paper mainly focuses on two parts of this problem.The first part is mainly based on the application scenarios of smart phones.Four machine learning algorithms are used to carry out experiments from the aspects of data acquisition,data processing,model training and parameter tuning.The number of packets,the number of features,model parameters and other parameters are tested,so that when the number of packets is 60 and the number of features is 9,the model recognition effect is optimized.In addition,the model is evaluated from the aspects of accuracy,precision and computational consumption,so that the recognition accuracy of data flow based on decision tree can reach more than99.9% and the computational consumption is low,which makes it the best algorithm among the test algorithms in this paper.The second part is mainly based on DPI technology and decision tree algorithm,design and implementation of terminal application data flow recognition system.The system can collect the mobile phone communication data in real time and output the classified results.The system combines DPI technology with decision tree algorithm,which makes the system not only high recognition accuracy,but also very low calculation cost.Finally,the system was tested in real time.For most apps,the recognition accuracy of the system was more than 99%.In general,this paper,based on the application scenarios of smart phones,discusses the best algorithm for data flow recognition,and strives to improve the recognition accuracy.At the same time,combining with DPI technology,it designs and implements the terminal application data flow recognition system,and achieves a good effect.