Pattern-Based Approach for Designing Fail-Operational Safety-Critical Embedded Systems

Abstract

To deal with fail-operational (FO) requirements in today's safety-critical networked embedded systems (SCNES), engineers have to resort to concepts such as redundancy, monitoring, and special shutdown procedures. Hardware-based redundancy approaches are not applicable to many embedded systems domains (e.g., automotive systems), because of prohibitive costs. In this scenario, adaptability concepts can be used to fulfill these FO requirements while enabling optimized resource utilization. However, the applicability of such concepts highly depends on the support for the engineering during system development. We propose an approach to cope with the challenges of fail-operational behavior of SCNES in which engineers are supported by design concepts for realizing safety, reliability, and adaptability requirements through the use of architectural patterns. The approach allows expressing FO concepts at the software architecture level. This lowers the effort for developing SCNES by utilizing generic patterns for general and reoccurring mechanisms.

Automatic Code Generation of Safety Mechanisms in Model-Driven Development

“…There also exist some completed research projects that target an increase in safety in general. The EU project SafeAdapt [37] provides a model-driven approach for self-adaption in safety-critical systems in the context of the automotive sector [38][39][40]. While our work focuses on the automatic generation of error detection mechanisms, theirs is more focused on achieving a safe state of the system after an error has occurred.…”