Autonomic monitoring in large-scale digital ecosystems via self-organisation

As the complexity, heterogeneity and dynamism of modern digital business ecosystems increases the long-standing signal-grounding problem remains prominent: It is required that systems are enabled to attach intrinsic meaning to their observable events in context, rather than simply reacting to perceived stimuli most often specified and encoded at design-time. Such a method would align system and process to autonomously characterise, reason and develop reaction models for new (or unforeseen) events. Addressing such a concern will be vital for the design, analysis and engineering of modern ecosystems. Whilst, autonomic computing models via policy-based management of context-aware systems are becoming common-place, further development of the general model, using a random reinforcement learning approach (collectivist memory-based) has been proposed to address the problem. However the growth in the size of the required memory storage capacity and associated efficient access to the distributed knowledge still presents a problem. This paper presents a novel approach to maintaining a situation space, within a situation calculus approach, based on proving that the space conforms to scale-free connectivity from a certain perspective. The special features of the topology may then be used, via an efficient strategy for memory clearance, based on immunisation methods in such networks. This results in a concise ecosystem evolutionary description provided by a core set of action histories.