Statistical Physics and Modeling of Human Mobility

Gallotti, Riccardo (2013) Statistical Physics and Modeling of Human Mobility, [Dissertation thesis], Alma Mater Studiorum Università di Bologna. Dottorato di ricerca in Fisica, 25 Ciclo. DOI 10.6092/unibo/amsdottorato/5198.
Documenti full-text disponibili:
Documento PDF (English) - Richiede un lettore di PDF come Xpdf o Adobe Acrobat Reader
Download (4MB) | Anteprima


In this thesis, we extend some ideas of statistical physics to describe the properties of human mobility. By using a database containing GPS measures of individual paths (position, velocity and covered space at a spatial scale of 2 Km or a time scale of 30 sec), which includes the 2% of the private vehicles in Italy, we succeed in determining some statistical empirical laws pointing out "universal" characteristics of human mobility. Developing simple stochastic models suggesting possible explanations of the empirical observations, we are able to indicate what are the key quantities and cognitive features that are ruling individuals' mobility. To understand the features of individual dynamics, we have studied different aspects of urban mobility from a physical point of view. We discuss the implications of the Benford's law emerging from the distribution of times elapsed between successive trips. We observe how the daily travel-time budget is related with many aspects of the urban environment, and describe how the daily mobility budget is then spent. We link the scaling properties of individual mobility networks to the inhomogeneous average durations of the activities that are performed, and those of the networks describing people's common use of space with the fractional dimension of the urban territory. We study entropy measures of individual mobility patterns, showing that they carry almost the same information of the related mobility networks, but are also influenced by a hierarchy among the activities performed. We discover that Wardrop's principles are violated as drivers have only incomplete information on traffic state and therefore rely on knowledge on the average travel-times. We propose an assimilation model to solve the intrinsic scattering of GPS data on the street network, permitting the real-time reconstruction of traffic state at a urban scale.

Tipologia del documento
Tesi di dottorato
Gallotti, Riccardo
Dottorato di ricerca
Scuola di dottorato
Scienze matematiche, fisiche ed astronomiche
Settore disciplinare
Settore concorsuale
Parole chiave
Complex Systems, Statistical Physics, Human Mobility, Traffic, GPS Data, Markov Process, Travel Time Budget, Mobility Patterns, Mobility Networks, Data Assimilation, Statistical Forecast, Time Perception
Data di discussione
22 Febbraio 2013

Altri metadati

Statistica sui download

Gestione del documento: Visualizza la tesi