A Model-Based Framework for System-Wide Plug-and-Play with Flexible Timing Verification for Automotive Systems

The integration of hardware and software components into today’s vehicles from a variety of suppliers is a complex process and becomes more and more challenging. The amount of code and data as well as the number of interconnections increases rapidly and pushes the complexity of the on-board electronic systems and the involved infrastructure to new limits. This stands in contrast with the constantly growing demand for new functionalities to enhance safety, comfort, and efficiency. Integration expenses are an increasing problem during design time and the addition of hardware and software to a vehicle after sale is limited today, because the systems are developed in a static manner. To ease the integration process and to allow the addition of functionality after sale, a model-based framework is proposed in this work that combines the plugand-play concept with an automatic timing verification to fulfill the real-time requirements of automotive systems. The idea is to divide the functionality of a vehicle into individual features that can be freely composed. Each feature consists of a set of hardware and software components as well as communication and timing requirements, which are automatically matched and verified in the resulting system setup. Addition of further features and automatic reverification is possible at any point in time with an adjustable approximation level. The approach is capable to process event-based communication patterns and is based on the data-centric design principle, i.e., data senders and receivers are loosely coupled. The contributions of this work comprise the introduction of the system-wide plug-and-play approach, the definition of a minimal set of suitable models, transformation patterns for a mapping to exemplary technologies, and the introduction of a method to specify timing requirements for unknown setups. They further include the development of a performance verification tool based on the Real-Time Calculus framework, its enhancement for the automatic processing of cyclic resource dependencies, and the design of an approach to control the approximation of the analysis and trade tightness of the derived timing bounds for computation time. The feasibility of the approach is shown by a running example based on the electric vehicle demonstrator (eCar) throughout this work. The performance of the extensions and approximation approaches for the verification process are examined in detail by a series of experiments.