The detection of conditions of congestion will be an important aspect of the performance of a number of intelligent transport systems (ITS) applications, including traffic management, incident management, and public transport information systems. One indicator of congestion is mean traffic speed. However, measurements of mean traffic speed will not necessarily detect congestion that is due to unpredictable traffic flow, eg 'stop-start' driving conditions. Drane and Scott show that two ergodic theoretic based measures may provide a useful measure of 'stop-start' congestion. The two measures are the Kolmogorov-Sinai Entropy and the correlation dimension. Both these measures provide estimates of the unpredictability (in the sense of chaos theory) of the flow. The Kolmogorov-Sinai Entropy is a global measure and so will give an indication of overall traffic conditions. The correlation dimension is a local measure and so provides an estimate of the congestion at a particular place. The aim of this paper is to provide a qualitative explanation of these measures in a manner which is accessible to practitioners. As well, some new results are presented concerning the performance of these estimators. (a) For the covering entry of this conference, see IRRD abstract 868345.


  • English

Media Info

  • Features: References;
  • Pagination: p. 151-65

Subject/Index Terms

Filing Info

  • Accession Number: 00722231
  • Record Type: Publication
  • Source Agency: ARRB Group Ltd.
  • ISBN: 0-86910-663-5
  • Files: ITRD, ATRI
  • Created Date: Jun 28 1996 12:00AM