Towards Discovering Delayed Mutual Influences in Organic Computing Systems

Abstract:
Current developments in systems engineering are characterised by a trend towards substantially more complex technical solutions, which especially manifests itself in an increasing number of (heterogeneous) subsystems, direct and indirect interactions between them, and mutual influences on each other. Often, even today’s systems are no longer manageable by humans and this trend is assumed to intensify rapidly. However, to master the resulting complexity, we have to identify such mutual influences in a first place. In this paper, we present a novel technique to estimate mutual influences in self-adaptive and self-organising systems, with a certain focus on Organic Computing systems. Compared to previous work, this technique is able to detect and quantify hidden mutual influences even if the impact of that effect is time-delayed. We demonstrate the potential benefit of the approach in terms of a case-study based in the field of adaptive industry automation.