CIPRNet training lecture: Hybrid simulation of distributed large-scale critical infrastructures

Abstract

Modern critical infrastructures represent the pivotal assets upon which the current society greatly relies to support welfare, economy, and quality of life. Nowadays, the trend is to re-organize these infrastructures by applying a System of Systems concept, where the sparse islands are progressively interconnected by means of proper middleware solutions through local or wide-area networks. The huge complexity of such systems makes the integration task among components extremely challenging. Indeed, it may introduce unexpected system behaviors, mainly affecting dependability and performance, that usually become evident only during systems operations and, in particular, in presence of stress or unexpected conditions. Additionally, as they cannot be detected earlier, these problems require complex on-site operations resulting in increased maintenance costs and overspending in terms of personnel resources. A promising way to cope with these new complex systems and to reduce maintenance costs, is to reproduce such distributed systems locally, and let them run prior to the actual execution on-site, in order to get knowledge about their real behavior and define mitigation means and improvement actions. On the other hand, the evaluation of this systems requires sophisticated modeling, simulation, and experimentation infrastructure, which needs the integration of existing simulation environments, real sub-systems, and experimental platforms, which have to interact in a coordinated way. Therefore, hybrid and distributed simulation strategies, supported by novel technologies for resources virtualization and working environment reproduction, represent the most promising way to define the needed strategies to actually support such complex paradigms [1,2].