Methods and models for the design and study of dynamic agent organizations

Agents are so called intelligent computer programs that are able to function autonomously and act pro-actively to perform tasks. A group of agents that is cooperating to perform a large task is called a Multi-Agent System (MAS). Similar to human organizations, a MAS also requires a form of organization to facilitate decision making, task distribution, resource allocation and other organizational activities to ensure effective and efficient cooperation. The aim of this thesis is to identify and describe methods and models for the design and study of dynamic Multi-Agent System (MAS) organizations.
In the first part of this thesis, we focus on the design of static agent organizations. For this purpose we define a number of organizational building blocks and we show how these can be used to design different types of organizations. In a case study in the smart-grid domain we show how such an organization is implemented. In the second part of this thesis, we focus on dynamic agent organizations. We present AgentCoRe, a decision making framework for agent coordination and reorganization in MAS organizations. In a number of studies we show how the different aspects of AgentCoRe can be operationalized. The third and final part of this thesis describes methods and models for the study of MAS performance. We introduce the Extended Organization Design model which is a domain independent model for describing the task-environment of agent organizations and its relation to organization performance. The model is then used to design a simulation environment for studying MAS performance.