Regional Group Behavior Analysis And Prediction Based On Mobile Communication Data

As the current major means of communication, smart phones have become indispensable communication device in people live and work, and then mobile communication network has become the largest media communications network. With the popularity of smart phones, tablet computers and other portable mobile terminal, and mobile devices such as smart terminals support the rise of mobile social networking, social activities between people is facilitated. People as the main body of the mobile social networks, the position is changing all the time, So we can record the location information of the people in mobile communication to establish the people’s movement trajectory model, and then analysis and prediction group behavior, which can solve some mass incidents caused by the crowd, such as traffic congestion, and other emergencies.To study the behavior of the group in mobile communication environment, the most important task is to obtain the user’s location information. There are three main ways to get the user’s location: GPS positioning, Base station positioning and WIFI positioning. GPS positioning accuracy is the highest, but GPS data generally provided by volunteers, it is difficult to obtain large amounts of GPS data. The WIFI hotspot coverage is small, it’s only cover a campus or a community, so it’s a great limitation in the study of the group behavior. Although the Base station positioning has the lowest accuracy, it has the characteristics of data acquisition and wide coverage, so Base station positioning is the most suitable for research the group behavior. By studying the characteristics of mobile communication data, we propose a model of group behavior analysis, it can be used to real-time identify whether there are group aggregation behavior. Through the establishment of weighted Markov prediction model to forecast the crowd behavior, we can waring to group aggregation behavior earlier.The main work of this thesis:Introduced main function of 3G core network, get billing data in SGSN device, analysis of billing data format, extract useful data for the analysis of group aggregation behavior.Study of base station networking mode, through characteristics of the networking mode, I propose a user fan-shaped distribution model in the base station, the model improve the location accuracy of the user in the base station. Establish user group behavior analysis model, real-time monitoring group aggregation behavior. Achieve a heat map, so that users beyond the normal distribution of the number of regional intuitively displayed.Prediction of single user trajectory by weighted Markov prediction model, statistical prediction of the trajectory of all users to predict the group aggregation behavior of the population. The Markov model were established at different time periods, that can improve forecast accuracy.