On The Cellular Signaling Based Mobile Trajectory Extraction,Cleaning And Prediction

With the popularity of mobile communication technology and mobile devices, the daily track record data become rich. Massive track data hides valuable knowledge about person and human society. Scholars from all sectors utilize data mining and machine learning methods to analyze the data from a variety of angles, aiming at transforming data into information and then to the value. Track prediction as one of the important sub-topics of Track Mining get more and more attention lately. If you can accurately predict the future position of each user. From the microscopic, life service App can push future projects to user, to enhance the user experience. From the macro, we can inference the future heat map from the view of the history heat maps, which can play an important role in the transport sector, urban management and other baking.Currently, the Institute for trajectory prediction data is commonly used GPS positioning data based on personal history,this form of the data set and this kind of method can be considered vertical method. And with respect to the GPS location data, base station location data is richer. Although this kind of the data contains more noise, we can not deny the potential of this kind of the data in predicting the trajectory.This paper proposes a more effective way in terms of the base station data cleaning, to make the track data becaming more pure and containing low levels of redundancy by using filtering algorithm. Next, integration and presents a longitudinal frame prediction method on the basis of previous studies. Based on a user stays daily stay region and stay time paper proposes a similarity index, and use condensed type hierarchical clustering algorithm to cluster users to groups, thus establishing a link between the transverse and longitudinal data classical methods. Paper also proposes a stay area extended algorithm to discriminate similarities and differences between stay points. Finally, paper conducts experiments on a real TDOA base station location data and implement the system,which shows the accuracy enhanced compared with tradictional one, proved horizontal base station location data on the trajectory prediction to some extent has potential.