An Effective Reliability Goal Assurance Method Using Geometric Mean for Distributed Automotive Functions on Heterogeneous Architectures

Functional safety is aimed at avoiding unacceptable risks and safety damages due to system functional failures, and it is a critical demand for the automotive embedded systems. For safety-critical distributed automotive functions, reliability is an important functional safety requirement and reliability goal should be assured. In general, the key of reliability goal assurance method is to transfer the reliability goal of a distributed function to that of each task. This study proposes an effective reliability goal assurance method called RGAGM for automotive functional safety. The core idea of this method is defining two kinds of geometric mean for tasks and function, respectively, and preassigning geometric mean-based reliability values for unassigned tasks, thereby saving more resources for systems. The correctness of the proposed RGAGM method is proved. Experiment results on the real-life automotive function and the randomly generated distributed automotive functions show that the proposed method can effectively ensure the reliability goal and reduce resource consumption cost compared with the state-of-the-art MRCRG method.