OD-Matrix Extraction based on Trajectory Reconstruction from Mobile Data

In this work, we propose a novel approach to reconstruct the individual's trajectories that has been extracted from Call Details Records (CDR) and Visitor Location Registry (VLR) data. Our methodology is based on the second-order Markov model for reconstructing the hidden history of the extracted sparse trajectories that reflects different routes visits by mobile users. After reconstructing the complete cell-to-cell trajectories of users, the OD matrix of the major cities of Estonia is derived and evaluated by the number of train passengers in the country. Our dataset contains more than 600 millions anonymized CDR and VLR events from approximately 3 hundred thousands users in Estonia.