Architectural support for model-driven performance prediction of distributed real-time embedded systems of systems

Abstract

Systems of systems (SoS) are large-scale systems composed of complex systems with difficult to predict emergent properties. One of the most significant challenges in the engineering of such systems is how to predict their non-functional properties such as performance, and more specifically, how to model non-functional properties when the overall system functionality is not available. In this paper, we define an approach to SoS performance prediction based on the modelling of system interactions and their impacts. We adopt an Event Driven Architecture to support this modelling, as it allows for more realistic and flexible performance simulation, which enables more accurate performance prediction. We introduce a generic architecture and present its instantiation in a software architecture for the performance prediction of defence SoS. Our architecture allows for loose coupling, interoperability, and adaptability and facilitates sustainable evolution of the performance model of the SoS.