Chinese Multimedia Messaging Service (MMS) analysis using hadoop

Abstract

Multimedia Messaging Service (MMS), is one of the most popular mobile data services in China. The current MMS user base is still undergoing steady growth with the popularization of 3G in China. Hence, a comprehensive understanding of the characteristics of MMS traffic is necessary to Internet Service Providers (ISPs) in network planning. In this paper, based on massive data captured from a backbone node for half a year, we provide a detailed analysis of MMS traffic, including the MMS successful receipt rate, the distribution of MMS content length, the distribution of MMS receipt duration and the distribution of average MMs transmission rate. Furthermore, we used K-means clustering method to evaluate the performances of the cell sites in the MMS network. All the analyses in this paper were based on our cloudy-computing platform, and the results also showed that this platform is very useful in network traffic analysis.