A Reliability Prediction Method for AUTOSAR Architecture Considering Unreliable Platforms

With the trend of intelligence, automobile architecture has become more complex. It is necessary to predict and discover reliability-related issues to reduce the cost of correction in the later period. In AUTOSAR-based automotive architecture design, software and hardware interaction, middleware platform behavior, physical environment, and system usage profile affect the system’s reliability. It is necessary to comprehensively consider these factors to predict the system’s reliability reasonably. However, existing methods often overlook the influence of some factors, especially oversimplifying the control flow of the middleware platform in the system. Resulting in difficulty in effectively modeling the behavior of the AUTOSAR middleware platform in error propagation and its impact on system failure behavior. To analyze the impact of middleware platforms on failure behavior, this paper analyzes the impact of the AutoSAR middleware platform on application software faults based on error propagation methods. Then, expand the AUTOSAR meta-model to model reliability parameters and automatically convert the architecture model into a formal model for reliability prediction. Finally, the effectiveness of considering unreliable platform behavior modeling was verified through a car headlight design case study.