Research Of User Traffic Analysis Techniques And Application In3G Network

As3G network and the number of3G users rapidly grow, more andmore researchers are getting involved in the research of data analysis of3Gnetworks. With the support of storage and computing tool-hadoop and hive,this paper first conducted an overall analysis of real3G bigdata,implemented Deep Packet Inspection to identify application with arecognition rate more than90%, identified the terminal types by matchingwith the data, analyzed user behaviors from both the view of signalingplane and data plane. This paper, for instance, analyzed traffic distributionof application or terminal, visiting frequency to base stations and so on.Furthermore, this paper predicted long-term traffic of new user based onthe short-term user features, using Naive Bayesian Classifier with about1TB data and further validated the feasibility of this method byexperiments with an average accuracy more than80%. This paperovercomes the difficulty of bigdata storage and computation, and predictsuser traffic with multiple user features. This paper can provide advices tonot only for researchers but also the relevant industry.