Algorithms for Dynamic Road-Traffic Control

Abstract

One of the main objectives of traffic control is to determine the control parameters of a traffic system in an optimal manner for the users of the system, namely to minimize delay times and number of stoppages. The signal-program parameters which affect road-traffic flow are: cycle time, greentime splits and offsets. The first two are determined within the “minor control loop” which sets these parameters at individual intersections. The “major control loop” links together all traffic-lights into a coordinated system by proper selection of the offsets throughout the whole traffic network.This paper presents algorithms for optimal and sub-optimal synchronization programs in a traffic responsive manner. The algorithms produce rapid results even for very large signal networks. The calculation is based on minimization of an objective function composed of updated individual link cost functions, subject to the constraints imposed on the offset across-variables by the structure of the network. The minimization is performed as a multistage serial decision process based on the Principle of Optimality.Faster suboptimal algorithms are devised for networks in which updating of the program is not feasible within the “major loop” time requirements. The guiding principle underlying these algorithms is that the gain in time attained by compiling a sub-optimal program should outweigh the loss in performance incurred by applying an outdated program. The net result is minimization of total cost to traffic in the network.