Adaptive signal control at individual intersections relies on detectors to provide advance vehicle arrival information for real-time optimization of the signal operations. As much as 25 sec of advance information may be needed to achieve near optimal operations if flow rates reach about 700 vehicles per hour per lane (vphpl). However, it is often impossible or impractical to place detectors far enough from the intersection to provide the desired amount of information. The use of predicted data becomes a tempting alternative under the circumstance. In this paper, computer simulation is used to assess the desirability of using predicted data in combination with the data provided by the detectors for signal optimization. Three predictors are compared and one is chosen to assess the impact of using the predicted data. It is found that reliance on limited advance arrival information provided solely by the detectors is more desirable than using predicted data to increase the amount of advance information.
Abstract