Combining Multi-Agent-System Methodologies for Organic Computing Systems

As the complexity of computing systems steadily increases, self-managing systems - as autonomic computing systems (ACS) proposed by IBM - are an adequate approach to minimize human effort spent on system administration. While ACS primarily are limited to servers and networks, organic computing systems (OCS) are intended for widespread applications in various domains. In addition to the self-x properties of ACS, OCS are adaptive and context-aware. Thus agent technology is particularly suitable for an implementation of OCS. Nevertheless a key prerequisite for a successful, industrial application remains in a systematical engineering of OCS according to accepted standards. However no single agent methodology is applicable to OCS as self-x properties are not supported directly. Therefore we have combined different proved agent concepts into a system architecture for OCS and developed an adequate, model-driven software engineering methodology based on the Unified Modeling Language (UML) and the model driven architecture (MDA)