Towards Verifying and Optimizing Self-Organizing Systems through an Autonomic Convergence Method

Bruno de C. B. A. SoaresMaíra A. de C. GattiCarlos J. P. de Lucena

Self-organizing emergent systems will only be acceptable in an industrial application if one can give guarantees they will accomplish exactly what they were designed for. The main issue is how to know if a certain self-organizing emergent system will exhibit the required set of macroscopic properties considering that it is unfeasible and sometimes even impossible to model all the behaviors resulting from the interactions between local agents. Therefore, we need a practical and efficient method to anticipate and verify this emergent behavior. That said, we propose our method through an autonomic computing approach. We complement and exploit the technique proposed by Kevrekidis and applied and reviewed by DeWolf through the removal of mathematical programming and the incorporation of self-configuration and online planning. We argue that online planners are excellent tools for converging global properties of self organizing multi-agent systems and pointing up how local decisions perturb the system changing the global behavior. Finally, we present a first specification of the framework that implements the method proposed.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: