Modelling Resilience of Data Processing Capabilities of CPS

Modern CPS should process large amount of data with high speed and reliability. To ensure that the system can handle varying volumes of data, the system designers usually rely on the architectures with the dynamically scaling degree of parallelism. However, to guarantee resilience of data processing, we should also ensure system fault tolerance, i.e., integrate the mechanisms for dynamic reconfiguration. In this paper, we present an approach to formal modelling and assessment of reconfigurable dynamically scaling systems that guarantees resilience of data processing. We rely on modelling in Event-B to formally define the dynamic system architecture with the integrated dynamically scaling parallelism and reconfiguration. The formal development allows us to derive a complex system architecture and verify its correctness. To quantitatively assess resilience of data processing architecture, we rely on statistical model checking and evaluate the likelihood of successful data processing under different system parameters. The proposed integrated approach facilitates design space exploration and improves predictability in the development of complex data processing capabilities.