GNSS-based road use metering systems need performance metrics, as well as ways to test and reliably compare them. Bern Grush and Joaquín Cosmen write about the function of the GNSS Metering Association for Road-use charging (GMAR), recently set up to address this issue
After a period of uncertainty, the implementation of Global Navigation Satellite System (GNSS)-based Road User Charging (RUC) approaches reality in Europe. The Netherlands' Government is assuming the risk of being the first authority to implement a system for all roads and all vehicles. In spite of this progress, it is important to recognise that standards are not yet mature and different initiatives could lead to solutions whose performance characteristics are not consistent across roadway environments. Of special concern is the lack of standardised performance metrics that are necessary for the procurement, validation, certification and ongoing operation of these systems via Service Level Agreements (SLAs).The future - a perspective
It's 2013 or 2017. Vehicles are equipped with OnBoard Units (OBUs) which meter road use and generate a bill for each trip made. The road networks used have had distance-based tariffs set according to demand - so-called Time/Distance/Place (TDP) charging. Each time a driver makes a particular journey, over the same route, at the same time of day and with the same degree of congestion, he or she can expect to be charged the same amount - certainly with only a small variation, perhaps under a tenth of a percent. If two drivers made the identical journey in vehicles of the same emission class they too would be charged the same amount. If the same manufacturer made both vehicles' meter, one would expect this consistency in the same way one would expect two Blackberries or two iPods to be similar. This turns out to be difficult to achieve with GPS - especially in built-up areas or other difficult signal environments. There are several approaches to mitigating this problem, including interconnection to the odometer, multiple sensors, map matching and strategic geo-fencing to represent virtual gantries.Obviously, multiple companies are solving this problem in a variety of creative ways. In any open market the preference is for a selection of road-use meters from which to choose, since all these are telematics devices that could be bundled with other services such as navigation or emergency services. Already the Netherlands is calling for multiple providers.
Once we get past the early and relatively easier stage of truck tolling, we will start tolling every vehicle on every road. It is conceivable that there may end up being as many manufacturers and models of road-user meters as there are for mobile phones now. And each driver will still want to be sure that he or she is not paying more than his or her neighbour, while the road owner, whether private operator or government, wants to be sure the right amount is charged to all users, regardless of which meter they use.
Performance levels
The GNSS road-use metering industry is expecting to see an explosion in demand for TDP charging in the next two years. There will likely be three or four more truck-tolling Requests for Proposals (RfPs) for GPS-based systems in 2009. One is expected shortly from the Netherlands for large trials of several tens of thousands of units.How will performance be described in these RfPs? In the responses? How should tests be set up? How can two meters be compared? How will SLAs be written? These will be critical issues to those authorities seeking to acquire such systems and for the industry that wants to adapt technology to these needs.
Already, the remaining gap between the variable performance we experience and the consistent performance we require has led several countries to allocate considerable resources to the procurement process and trials. They are possibly duplicating unnecessarily the test activities already being performed in other countries.
The implementation of a road-use metering system requires operational performance levels that protect the interests of all stakeholders. Road operators are anxious not to lose revenue; motorists want guarantees they will not be over-charged. Operators want no tampering; motorists want privacy. If measurable performance criteria for charging reliability, tamper protection and privacy were framed in such a way as to permit consistent and comprehensive system tests, as well as cross-manufacturer comparison (even when not tested in the same geography), the burden of system design and testing, writing RfPs, proposals and SLAs would be eased.
Creating a vicious circle, some authorities have investigated the capabilities of existing technology by means of comparison trials which are then used to establish the range of numerical values appropriate to the development of performance requirements so as to make those requirements appropriate to the capabilities of existing technology. It is a case of "Let me see what you have so I know what to settle for" versus "How much longer must I wait for the technology to be ready?" Performance requirements for a secure meter to determine and demand payment should be based on error tolerances for payment systems and not on the results of some trials. A road operator should be able to point to a desired set of tolerances and a standard test framework and ask who can meet that need. A manufacturer should be able to read such a demand, know how to carry out those tests and decide whether to respond. That is not the current state of affairs.
Unfortunately, which performance requirements are critical and how they should be defined and measured is not well established. How can an industry say it is ready with a supply if demand is not clearly established in standard, measurable terms? How can a government procure a system of such complexity and compare different offers? How can the system be validated for acceptance? How can an SLA be established and a certification process defined? How will the performance of the system be monitored?
GMAR
For this reason the newly formed GNSS Metering Association for Road-use charging (While the need for such a guidance framework may be evident, the way requirements need to be established is less so. This requires consideration of several relevant topics: Road charging schemes vary from simple distance-based to zone/time-based or other combinations. Distance-based metering may require some expression related to the maximum allowed distance measurement error; other schemes may require a more sophisticated recognition of charging accuracy that includes weighted false alarms, misses and correct rejections. The GMAR framework will include definitions valid for all charging schema types.
The framework will handle the need of the authorities and service providers to guarantee that users are properly charged while addressing the concerns of the user not to be over-charged or charged for use which was not made of infrastructure.
The requirements framework will be developed to be independent of the technology and system designs so that industry has the freedom to implement any solution which satisfies established performance requirements; anything less stifles innovation.
While positioning performance clearly plays a critical role, the performance metrics suitable to a secure payment service are paramount. Requirements will be written at a high level describing charging performances (for example, overcharging probability) rather than positioning performances (for example, position accuracy). As positioning technology continues to mature, payment service reliability will remain a key performance matter, while our current absorption in complex distance and positioning measurements will recede.
Finally, performance metrics will be described for ongoing certification and operation as well as definition and acquisition.
An example can help to underline the importance of a test guideline. Distance-based charging requirements are usually established in terms of 'accuracy'. Many trials report that the accuracy of some OBUs is greater than 1 per cent but what is the meaning of that and how were the trials organised? For instance: did all cars and all trajectories have errors smaller than 1 per cent? Was the average error for a car in one month less than 1 per cent? Was the probability that error is less than 1 per cent (say, 99 per cent) for all trajectories? Are all trajectories long enough to consider that time to first fix was negligible? Are trajectories in open roads, urban environments or tunnels?
What to expect
The implementation of RUC requires as a first and key step the definition of a set of performance metrics and tests to be used for the procurement, validation, certification and operational assessment of the system.GMAR is preparing its first draft framework for performance measurement of GNSS road use metering systems. This framework, expected by mid-2009, will describe qualitative and testable performance requirements, usable for testing and comparing GNSS metering products and services. These include charging accuracy, integrity and availability, as well as tamper resistance and privacy protection. We will assert that these performance requirements are particularly critical for defending evidentiary weight and, potentially, legal admissibility.
GMAR is an independent industry association, proposing a metric framework to enable comparisons between vendors and test locations and for a meter operator during operations. GMAR will not associate numeric figures with the proposed quantitative performance standards. That will remain the role of the tolling authority.
