A Mobility Analytical Framework for Big Mobile Data in Densely Populated Area

Due to the pervasiveness of mobile devices, a vast amount of geolocated data is generated, which allows us to gain deep insight into human behavior. Among other data sources, the analysis of data traffic from mobile Internet enables the study of mobile subscribers' movements over long time periods at large scales, which is paramount to research over a wide range of disciplines, e.g., sociology, transportation, epidemiology, networking, etc. However, to efficiently analyze the massive data traffic from the view of user mobility, several technical challenges have to be tackled before releasing the full potential of such data sources, including data collection, trajectory construction, data noise removing, data storage, and methods for analyzing user mobility. This paper introduces a mobility analytical framework for big mobile data, based on real data traffic collected from second-, third-and fourth-generation networks, which covered nearly 7 million people. To construct a user's history trajectories, we apply different rules to extract users' locations from different data sources and reduce oscillations between the cell towers. The comparison of mobility characteristics between our mobile data and other existing data sources shows the large potential of mobile Internet data traffic to study human mobility. In addition, our experiments discover the changing of city hotspots, the movement patterns during peak hours, and people with similar history trajectories, which uncover the common rules that exist among huge populations in a city.