Analyzing GPRS mobile network traffic with map reduce

Abstract

Internet traffic monitoring and analysis have been a very essential way in network management and optimization for the ISP. GPRS networks are increasingly carrying Internet traffic for Web browsing, Email and online games etc. In this paper, we proposed a Hadoop based cloud computing platform for high-speed network traffic analysis. We experimentally prove that our platform is capable of analyzing Tera byte network traffic with low computational complexity. Furthermore, by implementing our high-speed analyzing platform, we study the GPRS network activity. The results from the analysis illustrate the business of mobile user log into the net. Our results also identify power-law properties in several mobile network traffic characteristics. In addition, we find that although the overall user mobility is low in the network, the user online time is increasing.