Abstract
Decentralized systems have emerged as an alternative to centralized approaches for
dealing with dynamic requirements in new business models. These systems should provide mechanisms that contribute to flexibility and facilitate adaptation to changes in the environment. In this paper, we present two self-organization mechanisms for a decentralized service discovery system in order to improve its performance. These mechanisms are based on local actions of agents that only consider local information about queries they forward during the discovery process. The self-organization actions are chosen by each agent individually when the agent considers them to be appropriate. The actions are: remaining in the system, leaving the system, cloning, and changing structural relations with other agents. We have evaluated each self-organization mechanism separately but also the combination of the two as the environmental conditions in the service demand change.
The results show that the proposed self-organization mechanisms considerably improve the performance of the service discovery system
