In recent years, a new approach for estimating people’s movement in cities has emerged through mobile phone positioning. As opposed to the more traditional methods of traffic surveys, automated counts, or individual counters on streets, the use of aggregated and anonymous cellular network log files has shown promise for large-scale surveys with notably smaller efforts and costs. In addition, a frequent data feed from the cellular network has also been argued to demonstrate fine grain over-time variation in urban movements, which are lacking from the traditional prediction methods. However, despite the positivist approach to the new methodology, additional evidence is needed to show how cellular network signals correlate with the actual presence of vehicles and pedestrians in the city.
The purpose of this paper is to address this shortcoming by presenting the results of a survey effectuated in Rome, Italy in January 2007. Using the results of the two-day experiment, we will employ statistical models to investigate the relationship between empirical pedestrian and traffic counts on the streets of rome with the simultaneous Telecom Italia Mobile (TIM) network signal and traffic prediction. Secondly, we will explore whether the mobile network data demonstrates the significant time-dependent variation that is missing from traditional fixed predictors like space syntax choice and integration analysis and could thus describe cities dynamically over time. Finally, we will also outline some general issues of accuracy in using aggregate mobile network data for estimating people’s movement in cities.