Navigating congested cities : understanding urban mobility using new data sources

Abstract

Rapid urbanization and technological innovations sparked the generation of massive volumes of data that is continually improving in resolution. In particular, mobile phones, having reached penetration levels above 97% in Europe and Americas according to the World Bank, transformed into passive sensors of urban mobility by signaling movement at the individual level. The data generated by these devices has a wide range of applications concerning how people and cities interact through the infrastructure. This thesis presents new analysis tools that utilize large geolocated datasets to provide new insights towards human mobility, road networks, congestion, and energy. In the first part of this work, we analyze the emergence of vehicular congestion in an urban road network through the use of a simple traffic flow model. We show that spatial constraints and the topology of the road network are determinant factors that shape the nature of the city's phase transition to a congested state. In the second part, we outline a methodology that processes raw geolocated data to extract aggregate mobility information that is comparable to local surveys and existing origin-destination matrices for five different metropolitan areas. Next, we analyze how the unique congestion fingerprint of a city is produced through the combination of travel demand, population density, road supply and route choice. We evaluate the potential of implementing socially aware routing solutions for congestion alleviation, and assess the implications of such solutions. Finally, we couple urban travel demand with energy demand of electric vehicles, and present their relationship while exploring the potential benefits of optimized arrival hour and charging timeshifts.