Applied Agent-Based Modeling: Design and Simulation
of Self-Organizing Systems
This article will focus on the use of agent-based modeling
(ABM) in the simulation and design of self-organizing systems. There is a
growing commercial interest in self-organizing systems for addressing complex
business problems. Attractive features of self-organizing systems include
robustness to perturbations, adaptability to changing environmental conditions
and the scalability of systems due to the lack of need for centralized control
and management. Sample business areas include logistics, supply web coordination
and
distributed peer2peer computing systems.
Self-organization can be defined as the spontaneous emergence
of a macro behavior from the interactions of micro agents and includes such
concepts as symmetry breaking, order through fluctuations, far-from-equilibrium
dynamics, phase transitions, constraint construction, meta-stable states
and entropy production. ABM is a convenient computational simulation method
for characterizing the global dynamics of interacting heterogeneous agents
of such systems. Additionally, ABM provides an exploratory framework for
the design of simple rules of micro agent behavior and interactions that
result in desirable
global behaviors.
Theoretical approaches to the design of self-organizing
systems with ABM will be outlined and supported with representative projects.
Stephen Guerin
Redfish Group
http://www.redfish.com
stephen.guerin@redfish.com