Assessing the Effect of Failure Severity, Coincident Failures and Usage-Profiles on the Reliability of Embedded Control Systems

摘要

The increasingly ubiquitous use of embedded systems to manage and control our technologically (ever-increasing) complex lives makes us more vulnerable than ever before. Knowing how reliable such systems are is absolutely necessary especially for safety, mission and infrastructure critical applications. This paper presents a structured compositional modeling method for assessing reliability based on characteristic data and stochastic models. We illustrate this using a classic embedded control system (sensor-inputs | processing | actuator-outputs), Anti-lock Braking System (ABS) and empirical data. Special emphasis is laid on modeling extra-functional characteristics of severity of failures, coincident failures and usage-profiles with the goal of developing a modeling strategy that is realistic, generic and extensible. The validation approach compares the results from the two separate models. The results are comparable and indicate the effect of coincident failures, failure severity and usage-profiles is predictable.