Understanding usage behaviors of mobile apps by identifying app package in network traffic

Mobile applications have gained rising popularity due to the advantages in mobile technologies. Furthermore, the appearance of tablets and mobile devices with other forms has increased the diversity of apps. These phenomena have resulted in a growing need of app developers to understand usage behaviors of mobile apps. Previous studies in this area mostly rely on instrumented devices so that the scope is limited. To better understand the usage behaviors at a larger scale, we present an approach without assistance of any device which is designed to identify app installation packages instead of app traffic flows. Our study mainly focuses on measuring and analyzing the hottest apps and the distribution of app downloading based on both download number and volume. The result indicates that 10% apps account for 70% downloads both by number and volume. Moreover, it is observed that the download distribution of Android and iOS are different due to different system settings.