Farnaz Derakhshan - Assistant Professor, Faculty of Electrical and Computer Engineering University of Tabriz Tabriz, Iran
Fatemeh Khezerlou - Graduate student of Artificial Intelligence, Faculty of Electrical and Computer Engineering University of Tabriz Tabriz, Iran
Nasrin Shahpasandi - Graduate student of Artificial Intelligence, Faculty of Electrical and Computer Engineering University of Tabriz Tabriz, Iran

Using intelligent traffic control methods are the most practical approaches to solving problems caused by the lack of development of transport infrastructure. The purpose of this paper is to present an agent-oriented system to control urban traffic at intersections. In this paper, the information of mobile phones of car occupants in the intersections are applied and traffic load is estimated through collecting mobile phone signals. Then, using traffic load value and two intelligent methods, k-means clustering algorithm and genetic algorithm, good schedule is obtained for traffic signals at intersections.Our work has several advantages. First, using data of mobile phones of car occupants would be cost effective in regard with calculating traffic load with acceptance of a percent of error with no need to hardware construction. Second, intersections are clustered regardless of their number. Third, using genetic algorithm could estimate the best possible scheduling for the intersections, since genetic algorithm is among evolutionary processing algorithms and is not trapped in local optimization. Forth, existence of coordinator agent prevents emergence of probable problems among clustered zones.

intelligent traffic control; multi-agent systems; urban traffic control