Mobile Data Usage Prediction System and Method

The popularization of Long Term Evolution (LTE) leads to the issues of mobile data usage monitoring and prediction. The analyses of these issues can be referred for the deployment of mobile network and the recommendation of customer solutions. However, a great deal of statistical computing power and time is needed for several previous mobile data usage prediction methods. Because large variances may exist among the quantities of mobile data usages from different client devices, the error of mobile data usage prediction may be large. Therefore, this study proposes a mobile data usage prediction method to analyze the historical mobile data usages of target client devices for the prediction of future mobile data usages of target client devices. Then the marketing information can be obtained to client devices in accordance with the predicted mobile data usage, and the income of operator can be increasing. In experimental results, the mobile data records from 565,606 client devices during July of 2014 were collected and analyzed for the evaluation of the proposed method. The accuracy of the proposed method was 87.97% which was higher than previous methods.

I. Introduction

With the promotion and universality of cellular networks (e.g., Universal Mobile Telecommunications System (UMTS) and Long Term Evolution (LTE)) the issues of mobile data usage prediction are more and more important for the strategies of mobile network deployment and mobile data plan recommendation. The investigation indicated that the output of industrial production of global communication devices has been increased by 0.46 million US dollars, and the annual growth rate of mobile communication industries in Taiwan is measured as 3.7% [1]. Therefore, the volumes of mobile data usages will be generated and produced, and the techniques of mobile data usage prediction are required for mobile users and telecom companies.