Cross-Layer Dependency Analysis for Safety-Critical Systems Design

Abstract:
Current safety standards in domains such as automotive, avionics or medical equipment, all require for applications with different assurance levels to be integrated into the same system, either that all applications be certified to the highest applicable level or that sufficient independence of implementations can be established. Therefore, the challenge in robust designs is to minimize the number of unintentionally dependent elements. However, existing solutions for determining the dependent elements of a complex system require a laborious and costly process that, in case of e.g. Fault Tree Analysis (FTA), is quite static and hardly applicable to dynamic and evolving systems. In this paper we motivate and propose a formal analysis for automated model-based detection of dependencies across different architectural model layers that could jeopardize or undermine the independence of system elements. In contrast to other approaches that use a specific system and failure modeling we propose to consolidate information from different models to automatically detect dependencies that are otherwise part of the failure model.