A cooperative multi-agent system in traffic control

Multi-agent system (MAS) has been proven to be a useful domain for solving complex coordination and control problems, involving large numbers of autonomous entities interacting in a dynamic environment. Traffic control is one such problem. In this paper, a kind of architecture based on MAS is presented to implement simplified traffic control in a pro-active way by trying to predict and avoid road congestion. Then model is given to optimize ExplorationAnts in TCS. Also algorithms of agents are described in detail. At last experimental results are presented to validate the approach. The major conclusion is that pro-active congestion control is better than no congestion control not only in system throughput & system stability but also in predicting & avoiding road congestion.