Research Of Identification Methods Of Network Terminal Devices Based On Decision Trees

With the rapid development of Internet,Internet of Things and Big Data,the network environment has a large number of terminal equipments access,and the number and variety are continuously growing rapidly.The access of a large number of devices makes the network traffic more complicated and enormous.The identification of the network terminal devices is of great significance to the network management,network operation,and improving the user experience.At present,the commonly used method of detection network devices is collecting fingerprint information.As a result,a large number of detections are needed in order to acquire distribution information of equipments in the whole network environment.In addition,it is hard to obtain the fingerprint information of the equipments with barriers of the routers and firewalls.Different from identifying devices by fingerprint which can be gotten by detecting messages,the main research of this paper is identifying devices based on network traffic.The research gathers the information of network devices and related properties by the collected network traffic and then identifies devices by decision trees model constructed on the needed features of network traffic.The main research in this paper includes three aspects.First,putting forward a method of the identification of NAT devices based on network traffic.The NAT technology effectively concealed the network structure to the upper network,which makes illegal terminals access more convenient by forming an independent network environment.In this paper,a NAT device identification model based on decision tree is proposed and validated by experiments.It can effectively identify NAT devices.Secondly,the identification of mobile terminals can roughly classify the network terminal equipments through the separation of handheld devices and non-handheld devices.In this paper,a method of handheld devices identification based on C4.5 decision trees is presented.The results of experiment show that the method can achieve 95%accuracy rate,but less than 65%that of the method only use UA.Finally,this paper puts forward a method of network terminal equipments identification.This method combining the identification of NAT and the mobile terminal devices,and constructs a model of the network terminal equipments identification.Through this model,the general steps of network devices identification using decision tree are given,and the advantages and disadvantages of the method are analyzed.