Recent developments in computer technology have made substantial computational capability, in the form of mini- and micro-computers, available at modest prices. This, coupled with the available traffic sensors, controllable traffic signals and communications, makes the computer control of traffic feasible in many urban areas today. Many such systems are either operating or are being planned.
Software for today's hardware environment can be oriented toward several, or many, small machines each of which controls a small local area. Alternatively,the software can do more pre-processing of data in a small machine located near the detector thereby reducing communication costs.
This paper describes a set of filtering, prediction and control algorithms that are primarily intended for traffic responsive control of networks. A way to partition the network algorithms among a collection of microprocessors is proposed. Because of the way in which the control computations can be partitioned, the algorithms will also work well for isolated intersections.