Dortmund Technical University, Potsdam University (GERMANY)
Uppsala University (SWEDEN)
DFA (Deterministic Finite Automata)
CADP (Construction and Analysis of Distributed Processes)
Telecommunications protocol specifications are often defined in natural
language text documents, and there is no formal link between the
specification and the implementation, making any formal validation or
model checking difficult. Automata learning techniques have been
proposed to overcome this situation, by allowing one to construct and
later update behavioral models automatically. LearnLib is a framework
of tools for automata learning, which is explicitly designed for
the systematic experimental analysis of the profile of available
learning algorithms and corresponding optimizations.
This case study examines the use of automata learning by means of LearnLib, to construct behavioural models tailored to specific contexts and learning scenarios. In particular, the study uses three protocols from the CADP demos, applying LearnLib to "learn" each protocol behaviour, and experimenting with various options for optimization and tailoring.
The experimental results show the value of using tailoring in automata
learning algorithms, including in one example an increase in performance
of one order of magnitude. Though this case study involved
telecommunications protocols, it demonstrates an approach that could be
applied to the more general setting of reactive systems that are
component-based, addressing the problem of under-specification.
Harald Raffelt, Bernhard Steffen, Therese Berg, and Tiziana Margaria
"LearnLib: a framework for extrapolating behavioral models".
International Journal on Software Tools for Technology Transfer (STTT)
Volume 11, Number 5, 393-407, Springer, 2009.
Available on-line at: http://www.springerlink.com/content/7581g372n82206v4/
or from our FTP site in PDF or PostScript
Chair of Programming Systems,
|Further remarks:||This case study, amongst others, is described on the CADP Web site: http://cadp.inria.fr/case-studies|