Schedulability analysis support for automotive systems: from requirement to implementation

Abstract

Modeling and analysis of precise non-functional properties, such as energy and timing constraints, is key to the correct development of automotive systems. Automotive applications development cost, in particular, is impacted by incorrect design made at the early development phases but only detected later, often after implementation. This late detection of design errors leads to additional cost. In this paper, we propose a model driven approach to perform non-functional properties verification and to enable scheduling analysis of automotive systems at the very early design level. The different phases of a design range from the requirements to a model allocated on a specific execution platform: East-adl and Marte are used together to specify the structure and energy/timing constraints of the software, as well as the hardware parts of the system. To prove the correctness of specification and perform the scheduling analysis, the semantics of the constraints is given as mapping to a formal interchange format Xfg (eXtended Function-block Graphs) language. The Xfg models are then automatically translated into priced timed automata for model checking. This later transformation is supported by a tool chain called A-BeTA. We demonstrate the applicability of our approach on the Brake-By-Wire case study.