Dependability and Performance Assessment of Dynamic CONNECTed Systems

In this chapter we present approaches for analysis and monitoring of dependability and performance of connected systems, and their combined usage. These approaches need to account for dynamicity and evolvability of connected systems. In particular, the chapter covers the quantitative assessment of dependability and performance properties through a stochastic model-based approach: first an overview of dependability-related measurements and stochastic model-based approaches provides the necessary background. Then, our proposal in connect of an automated and modular dependability analysis framework for dynamically connected systems is described. This framework can be used off-line for system design (specifically, in connect, for connector synthesis), and on-line, to continuously assess system behaviour and detect possible issues arising at run-time. For the latter purpose, a generic, flexible and modular monitoring infrastructure has been developed. Monitoring is at the core of the connect vision, in order to ensure run-time observation of specified quantitative properties and possibly trigger adequate reactions. We focus here on the interaction chain between monitoring and analysis, to allow for on-line continuous validation of specified dependability and performance properties. Illustrative examples of applications of analysis and monitoring are provided with reference to the connect Terrorist Alert scenario.