Consistency Is Key: Can Your Product Line Realise What It Models?

Nowadays, automotive systems are modelled as cyber-physical product lines. However, often it is not clear whether a modelled configuration can be realised as a physical product. The combination of software- and hardware-artefacts harbours the risk of non-functioning products due to incompatible resource demands and provisionings of the components installed (realisability), thus can result in high financial loss for manufacturers. With new business models, such as over-the-air updates, they also face this risk for products already in the field, because it remains unclear whether a vehicle is still functioning after an update where resource demands have changed (update-ability). Manually analysing realisability and update-ability is infeasible in practice, as the number of product variants in a product line grows combinatorially (with respect to the number of configuration options). In this paper, we approach this challenge by proposing a novel baseline approach for the analysis of realisability and update-ability in automotive cyber-physical systems. We formally model resource demands and resource provisionings as well as the construction of a resource allocation problem per product variant based on constraint satisfaction problems. In the evaluation, we apply our method to an automotive case study by modelling realisation artefacts and investigate its feasibility and performance. Our results show that a non-realisable configuration of the product line can be identified in 206 ms median time. Finally, we discuss limitations and extensions of our ongoing work.