Meta dynamic states for self healing autonomic computing systems

Abstract

Studying the fundamental behavior of complex systems for emergent systemic properties forms the foundation for designing robust and intelligent systems. Computational systems like Autonomic Computing Systems (ACS) are no different. Researchers working on designing ACS are faced with a challenge trying to identify what is it that bestows autonomic behavior to such systems. Complex systems exhibit complex behavior like homeostasis, robustness etc, but the way they achieve is probably very simple. This paper is an attempt to view ACS as complex dynamical systems exhibiting certain lower order behaviors, which can be used as scaffolding to construct higher order behaviors like homeostasis, robustness and self-healing. The technique discussed here exploits the simple behavior of all dynamical systems organizing their state space into attractors and basins of attractors and uses it to impart homeostatic properties to such systems. This is a classic case of higher level orders emerging out of the lower level order for free. ©2005 IEEE.