Traffic State Classification in Condition-Responsive Traffic Control

The performance of a condition-responsive traffic control system primarily depends on the appropriateness of its selection mechanism for picking the right traffic control scheme for any traffic condition observed in the field. A traffic control plan is typically selected by comparing actual traffic conditions with certain thresholds pre-defined for a few distinctive traffic states. This paper presents a traffic state classification method using modified linear discriminant analysis, and illustrates its use for coordinated traffic signals operated in traffic responsive plan selection (TRPS) mode. A step-by-step description for determining TRPS thresholds is given, and application of the method is briefly discussed. Although the method was primarily developed for TRPS mode of signal operation, with minor modifications it can be applied to other traffic state classification problems with similar constraints.