The assumption of independent vehicle arrivals at traffic signals, such as that in the Poisson distribution, has been widely used for modelling delay at urban intersections. The degree of correlation among vehicles determines whether this convenient assumption of independence is realistic. Collection of vehicle headways seems to be the only well-known method by which an estimate of their autocorrelation can be found. This, however, is a tedious and time-consuming task. To cope with this problem, an effective methodology using a discrete model for estimating the degree of interaction among vehicles under given traffic and geometric conditions is proposed in the present study. As an example of this technique's usefulness, a model is proposed and estimated, using information collected from 40 locations. Preliminary results appear to confirm the strength and applicability of the proposed method even though the developed model is constrained by the limited available data.
Abstract