Research And Implementation Of High-Speed Traffic Identification System

In recent years,with the popularization of mobile intelligent terminals and the prosperity of mobile application market,the proportion of mobile traffic in the total network traffic has increased rapidly.The traffic identification technology of mobile applications has become a hot research topic.The existing mobile application traffic identification work has many problems,such as low identification rate,slow identification speed and so on.This dissertation mainly focuses on the standardized collection and high-speed analysis of flow data.Firstly,the method of downloading mobile applications and quickly obtaining pcap traffic files is realized;Then,a traffic feature sequence extraction method based on joint profile is proposed,and more than 3000 mobile application traffic features are extracted;Finally,a high-speed application traffic identification engine is implemented,and the performance test and high-speed calculation of application identification rate under different protocol traffic are realized.Specific contents include:First,rely on rxjava to integrate retrofit2,and flexibly realize the user authority management function and network request sending function.The above methods are designed from the point of quickly obtaining multiple APK files,and realize the automatic acquisition of multiple applications.Second,in the aspect of traffic feature sequence extraction,Wireshark encapsulates the interactive information of network applications,and realizes the extraction of traffic feature sequence based on joint profile.At present,the profile has been used to extract more than 3000 network application traffic characteristics.At the same time,the traffic high-speed identification engine is applied to load the configuration file to realize the high-speed calculation of identification rate.Third,according to the extracted feature sequence,in order to solve the problem of high-speed identification of network applications,this dissertation implements a high-speed identification engine of application traffic.For different protocol traffic,the engine uses the normalized traffic characteristic sequence in the joint configuration file to realize high-speed identification of unknown network applications and accurately calculate the application identification rate.After the actual test in Hubei Telecom campus network platform,the byte recognition rate of the system in this dissertation reaches more than 90%,which is better than the original manufacturer’s system.It has been highly praised by users and achieved good social and economic benefits.