User Mobility Analysis And Location Prediction Based On Mobile Communication Data

Mobile phone has become an important communication tool in modern society,and also plays an irreplaceable role in our life.The accumulation of a large amount of mobile communication data makes it possible to exploit the user's behavior by using the relevant information in those data.Therefore,mining user's mobile behavior based on location information has become an important research direction.In the past,most of the data used in the study of mobility behavior were GPS data,however,GPS data were difficult in collecting,which makes lager research limitations.Mobile data cover a wide range of people and easy to be collected,besides,it can reflect people's movement in urban district.Therefore,it is very important to analyze the user's mobile behavior through the mobile communication data.In addition,the user's location can also be predicted by using mobile communication data.Experiments show that the user's position is predictable,and the accuracy of prediction is also related to the rule of user movement.By predicting the user's future position in advance,we can monitor the areas where the crowd gathers,and make evacuation plans in time to prevent accidents such as stampedes from happening.In this paper,the location information of mobile communication data is used to complete the analysis of user's mobility,user's mobile behavior modeling and predict the location of mobile user respectively.The main work of this paper is as follows:First of all,this paper introduces the commonly used mobility modeling methods.Then,completes a comprehensive analysis of the user mobility,and depicts the user's mobile behavior by movement frequency,movement distance and movement direction.Secondly,we divide the base station area into several functional areas by combining POI information.Meanwhile,we present a density-based outlier detection method to deal with the trajectory.Then,we model the user's trajectory by using semantically methods.We use an association rules algorithm to explore the user's frequent trajectory patterns,and analyze the user's movement behaviors at different time periods.In this part,we further explore the characteristics of the user's mobile behaviors.Finally,we use a modified Markov model to predict the location of users.This paper aggregates the base station area into a state,and studies the user's movement between functional areas.We propose a weighted Markov model to predict the user's location by comparing the advantages and disadvantages of the first-order and high-order Markov model.The result shows that the prediction accuracy is effectively improved by using weighted Markov model.Through the research and analysis in this paper,it reveals that the user's mobile behavior has the spatial-temporal regularity,and the mobile behaviors among the user groups are similar.Therefore,the work in this paper is meaningful as well as valuable.It can be used as a reference for urban planning,traffic management,epidemic spread and information dissemination.