The impact of global on-line information provision on transport networks and how random early detection can help

Research output: Contribution to journalResearch articleContributedpeer-review

Contributors

  • H. Grzybowska - , University of New South Wales (Author)
  • S. Willmott - , Red Hat (Author)
  • S. T. Waller - , University of New South Wales (Author)

Abstract

Undoubtedly real-time traffic information strongly influences individual and group driver behaviour. To a lay person, the perceived possible effect of additional (accurate) information is likely to appear unambiguously beneficial. However, just as Braess and Down-Thomson, paradoxes for the addition of road infrastructure, more is not always better. It seems likely that similar negative effects can also occur with the addition of information to route planning. In this paper we show three related results. First, it is indeed the case that highly accurate real-time traffic information can have negative effects on drivers individually and altogether. Secondly, we show that these negative effects specifically arise due to the creation of oscillations in traffic pattern which exacerbate small occurrences of otherwise benign traffic congestion. Lastly, we show that novel adoptions of Random Early Detection strategies dampen such oscillations making real-time information a benefit to outcomes rather than a potential threat.

Details

Original languageEnglish
Pages (from-to)1356-1377
Number of pages22
JournalTransportmetrica B
Volume7
Issue number1
Publication statusPublished - 23 Dec 2019
Peer-reviewedYes
Externally publishedYes

External IDs

ORCID /0000-0002-2939-2090/work/141543759

Keywords

Keywords

  • random early detection, Real-time traffic management, traffic management under information provision, transport traffic assignment random early detection