Cross-Layer Adaptation in Multi-layer Autonomic Systems (Invited Talk)

This work presents a new reference architecture for multilayer autonomic systems called context-controlled autonomic controllers (ConAC). Usually, the principle of multiple system layers contradicts the principle of a global adaptation strategy, because system layers are considered to be black boxes. The presented architecture relies on an explicit context model, so a simple change of contexts can consistently vary the adaptation strategies for all layers. This reveals that explicit context modeling enables consistent meta-adaptation in multi-layer autonomic systems. The paper presents two application areas for the ConAC architecture, robotic co-working and energy-adaptive servers, but many other multi-layered system designs should benefit from it.