Agent-based day-to-day traffic network model with information percolation
File(s)
Author(s)
Shang, W
Han, K
Ochieng, W
Angeloudis, P
Type
Journal Article
Abstract
This paper explores the impact of travel information sharing on road networks using a two-layer, agent-based, day-to-day traffic network model. The first layer (cyber layer) represents a conceptual communication network where travel information is shared among drivers. The second layer (physical layer) captures the day-to-day evolution in a traffic network where individual drivers seek to minimize their own travel costs by making route choices. A key hypothesis in this model is that instead of having perfect information, the drivers form individual groups, among which travel information is shared and utilized for routing decisions. The formation of groups occurs in the cyber layer according to the notion of percolation, which describes the formation of connected clusters (groups) in a random graph. We apply the novel notion of percolation to capture the disaggregated and distributed nature of travel information sharing. We present a numerical study on the convergence of the transport network, when a range of percolation rates are considered. The findings suggest a positive correlation between the percolation rate and the speed of convergence, which is validated through statistical analysis. A sensitivity analysis is also presented which shows a bifurcation phenomenon with regard to certain model parameters.
Date Issued
2016-08-10
Date Acceptance
2016-06-30
Citation
Transportmetrica A-Transport Science, 2016, 13 (1), pp.38-66
ISSN
2324-9935
Publisher
Taylor & Francis
Start Page
38
End Page
66
Journal / Book Title
Transportmetrica A-Transport Science
Volume
13
Issue
1
Copyright Statement
This is an Accepted Manuscript of an article published by Taylor & Francis Group in Transportmetrica A-Transport Science on 10 Aug 2016, available online at: http://www.tandfonline.com/10.1080/23249935.2016.1209254
Publication Status
Published