Design and implementation of very large agent-based systems

Abstract

One of the primary reasons to use an agent paradigm to develop a system is that it allows developing systems that can be distributed and scaled to include a large number of participants without the common pitfalls of traditional systems. In this sense, certain key design considerations must be identified and addressed while developing a very large agent system. In this chapter, we will explore these design considerations as applied to the study of an energy distribution use case. We will also explore the nuances of the implementation of this use case in an agent infrastructure such as CybelePro™, which has been used to deploy systems of this magnitude. We finalize by stressing the importance of verifying the system properties of very large agent-based systems with a small demonstration of how to do it.