Multi-level holonification of multi-agent networks

Abstract

A networked multi-agent system is a group of intelligent agents structured in a social network induced by agent interactions. As the size and diversity of the multi-agent system increase, the interactions and mutual influence of the agents in the structure become more complicated, and this makes the coordination of the system hard to deal with. Holonic models for multi-agent systems have been considered as one of promising models to manage large scale problems. This paper tackles one of the key problems in the design of holonic multi-agent systems. Here we propose a method to construct a multi-level holonic structure for a multi-agent network based on the centrality of the agents. The proposed method is applied to an urban traffic problem and the evaluation results prove its effectiveness.