Transport for London, Congestion Charging Technology Trials ? Stage 1 Results
D Firth, Transport for London, UK
Transport for London (TfL) successfully implemented a congestion charging scheme in central London, enforced using cameras and automatic number plate recognition (ANPR) technology, in February 2003. TfL believes that there could be significant benefits to charging or enforcement using a more flexible technology and has embarked on a series of trials to test the suitability of such technologies in the London environment.
This paper covers the results of Stage 1 of the trials, being a proof of concept of a range of solutions employing cameras and automatic number plate readers (ANPR), satellite positioning, digital mobile and tag & beacon technology.
Objectives and scope of the trials
The Stage 1 trials were intended to provide a proof of concept of the technologies trialled, rather than a detailed assessment of their viability, whether from an operational or cost perspective. They have demonstrated which technologies are worth further investigation in future stages.
The trials also addressed the issues of the compatibility of the technologies trialled with other national and European road charging initiatives and of the maturity of the technologies.
The trials addressed four different groups of technologies:
?h Cameras and ANPR technology;
?h ¡§Tag & beacon¡¨ detection systems, of which Dedicated Short Range Communications (DSRC) is the leading solution for road tolling;
?h Positioning using Global Navigation Satellite System (GNSS) technology;
?h Digital mobile phone technologies (GSM ¡V Global System for Mobile).
The technologies trialled are all very different, and thus the assessment approach adopted was case specific. However, two key factors assessed in each case were:
?h accuracy ¡V is the location of the vehicle accurate?; is the timing accurate?; is the number plate correctly identified?; is the vehicle correctly matched to the on-board tag? etc.
?h evidential integrity ¡V does the system provide information which is sufficiently robust to be used for enforcement?
Results and conclusions ¡V GNSS
The accuracy of the GNSS solutions trialled is such that, on average, a buffer zone of 60m around the boundary of the present central London congestion charging zone would be required to be 99% confident that a position reported as being within the zone was actually so. Around some parts of the boundary this increases to 250m or more. While the performance of different GNSS on-board units (OBUs) varied, none gave a significant improvement on this average result. This included GNSS with additional support, such as dead reckoning or differential signals. Some units gave a significantly worse result.
Results and conclusions ¡V GSM
Using the Location Based Services of GSM mobile telephone networks to establish the position of a vehicle gave results with an average error, as quoted by the operators, of 800m and, as observed, of 2,400m. Clearly this level of accuracy is inadequate to support congestion charging.
The key opportunity for GSM identified during the trials so far, relies on deploying micro- or pico-cell GSM. This could allow cell sizes of only 25m to be used, potentially covering a junction or charging point. This would effectively use GSM as a tag and beacon system, with the phone becoming the on-board tag. GSM can also be used as a delivery channel for information or value-added services to the customer about the zone, or a reminder to pay.
Results and conclusions - Tag and beacon ¡V DSRC
Where a 1.5m overhang of the carriageway is permitted, a DSRC beacon mounted at a height of 6 m can achieve vehicle tag detection rates over a 2-lane 8m highway in excess of 99% - typically 99.5% in the tests. In order to cover wider carriageway widths a wider overhang is required. As a general rule, coverage extends at least 4.5m beyond the end of the overhang.
Infra-red (IR) based DSRC allows detection of tags at a greater range than microwave DSRC (up to 20 m). This, combined with a wider beam aperture, gives a wider communication zone and therefore better serves the carriageway widths of London. The trials indicated similar detection performance to microwave. However, there are a number of areas where IR DRSC is less mature than microwave DSRC; this does not preclude a solution being developed. In addition, the use of IR DSRC poses potential compatibility issues with other schemes.
Results and conclusions - ANPR / Cameras / Broadband
A number of variables around the existing ANPR/camera technology could improve performance and deliver further benefits. It would, for example, be feasible to use number plate readers with multiple video inputs or to deploy roadside ANPR systems reducing communications needs. Digital Subscriber Line (DSL) broadband technology is suitable for carrying images from roadside systems to a central point without loss of quality and brings significant cost reductions. The use of off-line ANPR processing to reprocess images can significantly improve results.
Future trial stages
Stage 2 of the trials is currently ongoing and will continue until 2006, during which a number of practical scenarios will be constructed, including a tag and beacon ¡¥mini-zone¡¦ with approximately 20 trial charging points. Stage 2 will also continue with the resolution of design issues for roadside ANPR solutions as well as research into GNSS-based on-board units and the potential for GSM solutions for improved customer experience and the use of pico cells.
Association for European Transport