Title
Employing early model-based safety evaluation to iteratively derive E/E architecture design.
Abstract
ISO 26262 addresses development of safe in-vehicle functions by specifying methods potentially used in the design and development lifecycle. It does not indicate what is sufficient and leaves room for interpretation. Yet the architects of electric/electronic systems need design boundaries to make decisions during architecture evolutionary design without adding a risk of late changes. Correct selection of safety mechanisms from alternatives at early design stages is vital for time-to-market of critical systems. In this paper we present and discuss an iterative architecture design and refinement process that is centered around ISO 26262 requirements and model-based analysis of safety-related metrics. This process simplifies identification of the most sensitive parts of the architecture, selection of the best suitable safety mechanisms to reduce thereby failure rate on the system level and improve the metrics defined by the standard. To support the defined process we present the metamodels that can be integrated with existing DSL (domain-specific language) frameworks to extend them with information supporting further extraction of fault propagation behavior. We provide a framework for architecture model analysis and selection of safety mechanisms. We provide details on the model-based toolset that has been developed to support the proposed analysis and synthesis methods, and demonstrate its application to analysis of a steer-by-wire system model and selection of safety mechanisms for it.
Year
DOI
Venue
2014
10.1016/j.scico.2013.10.005
Science of Computer Programming
Keywords
Field
DocType
Automotive systems,Embedded systems,Model-driven engineering,Quantitative safety analysis,ISO 26262
Architecture,Software engineering,Digital subscriber line,Computer science,Model-driven architecture,Failure rate,Systems design,Theoretical computer science,Enterprise architecture framework,System model,Defined process
Journal
Volume
ISSN
Citations 
90
0167-6423
5
PageRank 
References 
Authors
0.48
15
6
Name
Order
Citations
PageRank
Vladimir Rupanov1101.45
Christian Buckl233938.31
Ludger Fiege355552.04
Michael Armbruster4597.33
Alois Knoll Knoll51700271.32
Gernot Spiegelberg6184.95