On bringing object-oriented software metrics into the model-based world - Verifying ISO 26262 compliance in Simulink

Abstract

For ensuring functional safety of electrical/electronic sys- tems, it is necessary to exclude malfunctions from hardware and software as well as from the interaction of both. In today’s passenger vehicles, more and more safety critical functionalities are implemented in software. Thus, its importance for functional safety increases. The dominating safety standard for the automotive domain (ISO 26262) con- siders the software part and defines requirements for safety critical soft- ware. However, applying and fulfilling the standard is a major problem in industry. In this context, the paper presents a novel metric-based approach to evaluate dataflow-oriented software architectures used in many model-driven processes regarding the fulfillment of requirements defined by ISO 26262 (in particular part 6). The core idea is to derive metrics for model-based software from already existing, well-performing metrics elaborated for other programming paradigms. To link metrics to requirements fulfillment of ISO 26262, we briefly sketch the factor-criteria- metrics paradigm for this problem. Technically, this paper presents a generic meta-model for dataflow systems, which is used to define the met- rics. We implemented this meta-model and the metrics as a prototype for Matlab Simulink. As examples, two models of a 400 kW full Drive-by- Wire experimental vehicle with all-wheel-steering, all-wheel-drive, and electro-mechanical brakes are analyzed using this prototype.