Abstract: As 3G network and the number of 3G users rapidly grow, the research on data analysis of 3G networks has become a hot topic. This paper sets up a bigdata storage and computing platform and conducts an analysis of large-scaled mobile user data. Based on the short term user traffic features, mobility and device type, this paper predicts long-term traffic of new users, using naive Bayesian classifier and further validats the feasibility of this method by experiments with an average accuracy more than 80%.

Key words: 3G network, data/control plane analysis, mobile device traffic analysis, naive Bayesian classifier, traffic prediction