Research On Real-time Network Traffic Identification Algorithm Based On Cloud Computing

With the rapid expansion of Internet technology and the more popularization of the mobile network usage,a variety of real-time network applications(such as streaming media,P2P,etc.)continue to emerge constantly.How to solve the problem of real-time network traffic identification has become the key of Network security and network monitoring.This paper take network application data as the research object,such as web browsing,file transfer,e-mail,etc,optimize attribute selection algorithm to reduce the dimension of network data by analyzing the behavior of traffic behavior,the parallel processing mechanism of Map/Reduce in cloud computing technology is widely used in the operation of large data,so the simulation and analysis are carried out by using Map/Reduce framework.Asymmetric routing in the network will affect the traffic identification,so an adaptive measure is proposed to reduce this effect.Finally,by constructing a real-time network traffic identification model,to identify the data of the network application.The details are as follows:1.the attribute selection is the data preprocessing stage of the traffic identification method,this paper proposes a Wrapper attribute selection algorithm based on ReliefF,which can maintain highly accurate data characteristics while reducing the dimension of data.2.this paper study the asymmetric routing problem in the real-time network,analyze the occurrence of this phenomenon and the negative effects on flow identification,and propose an adaptive measure to reduce the negative effects.3.this paper propose a real-time traffic identification model based on semi supervised learning method,the model can not only fast and accurately identify the traffic,but also has the new features,which the previous traffic identification model does not have,which have a greater advantage in identifying real-time network traffic.