Information Technology Reference
In-Depth Information
The DeSCAS Methodology and Lessons
Learned
on Applying Formal Reasoning to
Safety Domain Knowledge
Jan Gačnik
1
, Henning Jost
2
, Frank Köster
1
, and Martin Fränzle
2
1
German Aerospace Center, Lilienthalplatz 7,
38108 Braunschweig, Germany
{jan.gacnik, frank.koester}@dlr.de
2
University of Oldenburg, Ammerländer Heerstr. 114-118,
26129 Oldenburg, Germany
{henning.jost, martin.fraenzle}@informatik.uni-oldenburg.de
Abstract. Functional safety has become an important aspect for
engineering activities in the automotive domain due to the upcoming
introduction of the safety standard ISO 26262. This paper proposes
a methodology to guide the safety related requirements engineer-
ing process by means of OWL (Web Ontology Language) ontologies.
These ontologies formalize necessary domain knowledge and serve
as reference models to support semi-automated requirements discov-
ery and to ease the certification process. Using OWL's logical base,
knowledge inference is applied to reason about safety measures for
ensuring compliance with the reference process (guidance). The pro-
posed methodology has been implemented in a prototype toolchain
and applied to a simple lane departure warning system as an example
assistance and automation system. Lessons learned refer to concep-
tual (expressiveness) and technical (tooling e
ciency) issues.
Keywords: Certification, ISO 26262, Domain Knowledge, Ontology, Process
Framework, Assistance and Automation System, Semantic Reasoning
1
Introduction
Safety critical systems like assistance and automation systems (AAS) in the
automotive domain demand a clearly defined proceeding during development,
especially to support certification and qualification processes. In order to re-
duce the risk of a hazardous system failure, standards have been defined which
propose a certain proceeding, requirements and associated methods and mea-
sures during development. One of these standards is the upcoming ISO 26262
for functional safety in the automotive domain [6]. Due to the informal rep-
resentation of such standards in natural language text, there is an inherent
