Declarative Dependency Specification for Inter-connected Large-scale Cyber-physical Systems

Abstract

Cyber-Physical Systems (CPS) are inter-connected software-hardware artefacts ranging from small-scale bio-electrical implants, medium-scale smart cars and intelligent production lines, to large-scale critical infrastructures with power grids, telecommunication networks and railway systems. Large-scale CPS plays a central role in modern society by delivering essential services for human daily life. Depending on the delivered services, any disruption or mal-function of these systems with a duration of hours or days will have severe societal and economic impacts and consequences, in particularly when the inter-dependencies between these systems are considered: the total consequences resulted by cascading effects is usually much larger than the one caused by the original failure. Federated simulation provides an effective means to analyse the overall dynamics of inter-connected large-scale CPS. In a typical federated simulation environment, dedicated domain-specific simulators, both commercial and open-source, are applied and inter-connected together to provide realistic system behaviours of the complete CPS networks. Specifying the dependencies among different simulator models is however a challenging task, especially for networks involving multiple large-scale CPS. In this paper, a declarative approach for dependency specification based on data stream processing is proposed. Preliminary results show the benefits comparing to the traditionally imperative approaches in terms of flexibility, scalability and usability.