Posing Some Design Questions for Crafting Agent Simulat

Abstract

Today there is a proliferation of research into systems based on the autonomous agent, multi-agent and holonic paradigms. These systems are being applied to environments including manufacturing, logistics, defense, and supply chain management to increase the flexibility, realism, openness and mass-customization of real-world decentralized operations. It is now well recognized that prior to deploying agents into any physical environment where machines or any form of hardware is controlled by the agents, a key step in verifying the functionality and reliability of the agent-based control system must take place, namely simulation of the agents' actions and their interactions. However, there is very little formal basis or methodological practice for creating a robust simulation of the distributed agent-based control system that would be used in a physical setting. This paper takes an initial step in trying to identify a set of design questions that must be answered by the agent system developer as part of the instantiation of the agents in the simulation