Real-time passenger location estimation using CDRs and train objects generated from crowdsourced timetables

Abstract

This paper shows that cell phone Call Detail Records (CDRs) and train objects in GIS, generated from crowdsourced timetable information, can be used to estimate the train on which a specific passenger is riding. Passenger train ride likelihood calculation is first conducted using CDRs and targeted train objects. Then, the results obtained are compared with the trains specified in GPS logs. Empirical results obtained contain both good cases such as trains being estimated from CDRs and train objects corresponding to the train specified in GPS logs and bad cases such as the estimated train not corresponding to the actual train on which the passenger is riding. Therefore, analysis of each case is also carried out. The application of train objects and cell phone CDRs can facilitate analysis of how passengers ride trains, identification of origin-destination rail routes, and congestion/delay reduction.