have been an important element in evaluating the operations and performance of traffic surveillance and control systems. For example, a ``floating car'' survey was used in evaluating the performance of SCOOT [Hunt et al.(1981), Clowes(1983)], an areawide adaptive signal control system developed in the U.K. and implemented in many cities worldwide. Similar surveys have also been conducted by the California PATH program in its Freeway Service Patrol project [Petty et al.(1996)]. In these evaluations, vehicle trip time and distance traveled are recorded before and after the implementation of the system, and compared. Another example of operational tests is the evaluation of INFORM, a traffic management system consisting of integrated signal control and driver information systems, in a 40-mile stretch of highway corridor in Long Island, New York [Smith and Perez(1992)]. The INFORM control elements include traffic monitoring in the control center, ramp metering, variable message signs, and traffic signals at intersections. Motorist response to and the effectiveness of ramp metering and variable message sign strategies was evaluated, using extensive field data and motorist perception surveys.
An evaluation based on field test was also designed for the EUROCOR project. Corridor Peripherique in Paris and A10 West in Amsterdam were selected as the test sites. The strategies evaluated were composed of a combination of ramp metering (with or without ALINEA ramp metering, see [Papageorgiou et al.(1997)]) and variable message signs (display queue or travel time information). Data such as volume, occupancy, speed, travel time, and queues in the testing networks were collected and used as indicators for the performance of the system. The evaluation for each strategy was planed to be conducted during the evening peak hours (4 hours) for at least eight days.
The ADVANCE project is a recent example of using field test to evaluate and demonstrate the benefit of ATIS [Bowcott(1993), Saricks et al.(1997)]. Vehicles equipped with a Mobile Navigation Assistant (MNA) act as probes, sending real time travel information about arterial and local streets to a traffic management center, which in term transmits the information to other equipped vehicles to aid drivers' dynamic route planning. A field test for evaluating the ADVANCE dynamic route guidance system was conducted on a suburban area in Chicago, Illinois [Schofer et al.(1996)]. Three ADVANCE-equipped vehicles were driven from five predefined origin to destination pairs. Two vehicles follow MNA-provided routes, generated based on real-time traffic information collected by a fleet of 18 equipped vehicles driven on designed routes. The third follows routes based on historical information. Experienced travel times were collected and compared.
Although operational tests are necessary, they are expensive. Furthermore, the system has to be fully operational in order to conduct such tests. On the other hand, computer simulation models allow for testing alternative system designs under a controlled environment, before conducting operational tests and refining original designs, thus resulting in more effective implementation.