Rethinking traffic count methodologies to derive OD matrices from aggregated mobile phone data

Abstract

Among the recently identified Big Data sources identified for mobility analysis, Mobile Phone Data (MPD) has been is considered as a promising passive source of information to complement traditional travel surveys thanks to large samples that are not limited to one mode of transport. Unlike individual MPD, aggregated MPD avoids privacy concerns and can be collected over long periods of time, but it only provides the number of people in a given area during a given time interval and requires further processing for mobility management. This paper therefore proposes a theoretical framework for generating OD matrices from AMPD, using methods derived from traffic counts. To ensure proper data transformation, several algorithms are tested, first in simulation and then using real data. The results demonstrate the potential of AMPD to generate high quality OD matrices on a continuous basis.