Traffic signal control with iterated prisoner's dilemma model

This paper shows a traffic signal control model using a concept of iterated prisoner's dilemma (IPD). Traffic signal control is a dynamic problem influenced by traffic, geographic and social conditions. To manage the system, we introduce a decentralized artificial intelligence method by multiagent. We approach to represent the traffic condition as prisoner's dilemma of traffic signal agents. The agents play a game with neighbor agents for getting a payoff. Two states of traffic signal Blue and Red are defined as an agent's states. The payoff is decided from states of neighboring agents and stress points of the car in that place. Then we show that keeping the dilemma condition of agents is traffic signal control by multi agent system.