Self-organizing Service Discovery: Applying Organic Computing Techniques to Service Oriented Architectures

Diplomarbeit, Oktober 2008

Present service discovery approaches are based on registries, whose design and matching shortcoming still curb the advantages of Service-oriented Architecture applications. Thus in this thesis, a decentralized and self-organizing service discovery approach by applying nature inspired Organic Computing techniques is presented that aims to overcome these shortcoming. Discovery agents of the Self-organizing Service Discovery model automatically locate service requesters and transport their service descriptions for matching purpose to specific service providers in the network in a total decentralized manner. Thereby discovery agents are coordinated in a stigmergic self-organizing way to those service providers, that apparently offer the requested kind of service. To enhance the efficiency of the service discovery and to improve the discovery precision and recall, service discovery model extensions are presented. In order to demonstrate the efficiency and performance of the Self-organizing Service Discovery and the improvements of its extensions, the model variations are evaluated based on a simulation, whose results are promising.