A Measurement of Mobile Traffic Offloading

A promising way to use limited 3G mobile resources efficiently is 3G mobile traffic offloading through WiFi by the user side. However, we currently do not know enough about how effective the mobile traffic offloading is in the wild. In this paper, we report the results of a two-day-long user-based measurement of mobile traffic offloading by over 400 android smartphone users in Japan. We first explain that the variation of aggregated traffic volume via WiFi is much greater than that via 3G in our dataset. Next, we show that the traffic volume offloading through WiFi is common over whole weekend and weekday night, though weekday rush hours have less chance of traffic offloading. Our results emphasize that a small fraction of users contribute to a large fraction of offload traffic volume. In fact, our per-user level analysis reveals that the top 30% of users downloaded over 90% of their total traffic volume via WiFi. However, bottom 20% of users stuck to 3G only and over 50% of users turned off the WiFi interface in business hours. Also, 17.4% of the total traffic volume was generated by users whose WiFi traffic volume was less than 1MB. We observed that some hybrid users downloaded most of their traffic volume via WiFi in shorter durations. In this sense, there is more room to improve the current traffic offloading by promoting users to use WiFi more effectively. Furthermore, we demonstrate that WiFi offloading is mainly performed by access points (APs) in homes while the use of public WiFi APs is still uncommon in our dataset.