Distributed traffic signal intelligence control model based on non-hierarchical MAS

This paper presents a novel approach using non-hierarchical distributed MAS in urban intersection traffic signal control. It is called traffic signal control system based non-hierarchical MAS (TSCSNHM). In the model, each agent plans pattern of signals according to it local traffic status for the intersection it responds. But the patterns must be negotiated with neighboring agents before being implemented. By this means, bigger area effective control can be achieved. It uses a simulator to validate the idea. This simulator is programmed by Java, and uses KQML to implement communication and negotiation between agents. To contrast with the approach, this paper uses other two typical control methods, fixed-phase control model and centralized control model in simulation. The results of simulation show that TSCSNHM has excellent performance that is closed to centralized control model, and is more simple, flexible and robust than the latter.