A comprehensive distributed architecture for railway traffic control using multi-agent systems

Abstract

In this paper, we propose a Distributed Architecture for Railway Traffic Control using Multi-Agent Systems called DARTCMAS. It has a hierarchical organizational model that consists of three main layers of diverse agents: “Train Agents” at the bottom, “Station Agents” in the middle and “Center of Traffic Control Agent” in the top layer. Any train in the railway traffic system is considered as an intelligent agent that has reactive, proactive and interactive properties, simultaneously. DARTCMAS considers local traffic control centers that are embedded in some selected stations as “Station Agents”. Lastly, “Center of Traffic Control Agent” acts as a comprehensive monitoring and supporting reference. We define an internal architecture for each of the agents. Furthermore, two types of interactions among agents are developed: first, inter-layer interactions that represent communications between two adjacent layers in the organizational model of DARTCMAS; and second, intra-layer interactions that represent relations between the same agents type in a specific layer. Simulations have been performed using MATLAB and the “Java Agent DEvelopment framework” whose results show the effectiveness of the model.