Self-adaptive Worker-Helper Systems with Self-Organized Task Allocation

In this chapter a self-organized worker helper system is described, which is part of an abstract Organic Computing system (OC system). It consists of normal worker components and helper components, and the workers need some service from time to time in order to continue with their normal work. The service is done by the helpers, which have reconfigurable hardware to perform the different service tasks. The speed of service for a certain task depends on the amount of resources configured for this task. Strategies are presented that can be used by the helpers to decide whether to accept a service task and how to reconfigure themselves. It is also described how the worker helper system can be organized without global knowledge about the type of service requests and the set of available helper components. In order to obtain a decentralized mechanism and to make it suitable for the paradigm of OC a fully decentralized and dynamic clustering algorithm has been combined with a self-organized task allocation system. Empirical results show that the described worker helper system can adapt to dynamic situations with changing probabilities for service, and that decentralized clustering is able to reduce the reconfiguration cost significantly.