MobilePulse: Dynamic profiling of land use pattern and OD matrix estimation from 10 million individual cell phone records in Shanghai

Abstract

In this paper, we present a new method for dynamically profiling land use pattern and generating Origin-Destination matrices from millions of individual mobile phone traces. It is based on analyzing individual anonymzed mobile phone call records. Starting from these records, we systematically analyze the different signals received from the cell phone towers and identify the key feature associated with land user patterns. Meanwhile, we derive individual traces and extract origin, destination and travel time of each estimated trip. We show that cell phone signals could indicate the land use patterns, and the estimated Origin-Destination matrices correlate well with the Shanghai Travel Survey 2004 data. Moreover, our methodologies have several advantages: 1) capture weekday and weekend patterns as well as seasonal variations; 2) capture work and non-work trips, 3) capture the traffic flows for all transportation modes. These features make this method for land use profiling and Origin-Destination estimation very useful.