Use this URL to cite or link to this record in EThOS:
Title: Information and dynamics in urban traffic networks
Author: Petri, Giovanni
ISNI:       0000 0004 2732 1285
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2012
Availability of Full Text:
Access from EThOS:
Access from Institution:
The study of complex systems has intensified in recent years. Researchers from many different disciplines have realised that the study of systems possessing a large number of degrees of freedom interacting in a non-linear way can offer insights into problems in engineering, biology, economics and many other fields besides. Among the themes in complexity, we focus here the issues of congestion and congestion emergence in the context of urban networks, with particular reference to the effects of dissemination of information about the system’s status. This topic is of great relevance today, due to the increasing availability of real-time information about traffic conditions and the large diffusion of personal devices that allow travellers to access such information. Through the analysis of a few simple models of information propagation in urban environment, we uncover that, contrarily to the naïve expectation, complete information is often detrimental to the global performance of the urban traffic network. Indeed, global or long-range dissemination induces correlations in the systems that become a source for spatial disorder, making the system more prone to the emergence of congested states and pushing it away from its Wardrop equilibrium. The models we study range from simple flow models on network to complete agent-based simulations on real-world networks with interacting agents and dynamical information. We then analyse real data, coming from London’s network of traffic detectors. We confirm that the heterogeneity in the distribution of traffic flow and occupancies across the network reduces its performances, consistently with the results obtained for the information propagation models. In addition, we find a rich phenomenology strikingly similar to the one found in critical self-organised systems. Indeed, we measure power-law correlation functions and 1/f power spectra, hinting to long spatial and temporal effects in the traffic flow, and confirm this result through the community detection analysis of the detectors’ correlation network, which showing that the whole urban area behaves as a single large chunk. We conclude discussing the origin of these features and how they can be used to improve the network performances.
Supervisor: Jensen, Henrik ; Polak, John Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral