7. An Evaluation Framework for Modeling Languages Supporting Predictable Vehicular Software Systems

Handling the software complexity of modern vehicular systems has become very challenging due to their non-centralized nature and real-time requirements that they impose. Among many software development paradigms for these systems, model-based development excels for several reasons including its ability to verify timing predictability of software architectures of these systems using pre-runtime timing analysis techniques. In this work, we propose a comprehensive framework that captures the timing related information needed for the modeling languages to facilitate these timing analyses. We validate the applicability of the framework by comparing two modeling languages and their respective tool-chains, Rubus-ICE and APP4MC, that are used for software development in the vehicle industry. Based on our results, both modeling languages support the design and analysis of vehicle software, but with different. Both modeling languages support time-, event- and data-driven activation of software components and modeling of single- and multi-rate transactions. Amalthea targets applications on single nodes with multi-core architectures while RCM focuses on single-core single-node and distributed embedded systems with ongoing work for supporting single-node multi-core architectures. In comparison to Amalthea, RCM provides a generic message model which can easily be re-modeled according to protocol-specific properties.