Summary of the workshop on "Use of Description Techniques" |
Geneva, 23 November 2002
By: Arve Meisingset
Twenty-six participants attended the Workshop. The
presentations at the Workshop are accessible at the ITU web.
Mr. Meisingset welcomed the
audience and presented the objective of the Workshop, to determine needs for
improvements to the description techniques for various application domains.
Mr. Schumacher and Mr. Mainwaring informed that there is some
scepticism towards use of formal description techniques within Study Group 11.
They identified a need for professional support to develop and use formal
specifications, both for behaviour and data, within the Study Group.
Mr. Campos Neto
presented many usages of description techniques (mostly informal) within Study
Group 16. In SG 16, the most used description techniques are ASN.1 (in the
protocol Recommendations) and ANSI-C (in the media coding Recommendations).
Several other techniques are used, many times as a function of the experience
of the contributors and/or of the "legacy" of a Recommendations series. In
particular for the latter cases, specialized support is frequently needed for
some description techniques, e.g. ASN.1.
Mr. Johannessen presented activities in Study
Group 4 and work on methodology within this Study Group. The application area
of SG4 is data intensive as opposed to SG11 and 16, which are focused on
protocol development. SG4 asks for support from SG17 on
- Approaches to provide requirement specifications;
- Use of UML, for example to provide the paradigm independent specifications;
- Means for mappings between specifications, such as the paradigm independent and
dependent specifications;
- Means for reverse engineering from paradigm dependent to independent specifications;
- Methods and guidelines for using the UTRAD methodology.
Mr. Skelton provided a view on language
standardisation as seen from industry. He stated that tools are felt to be
complex and expensive. There is a need to reduce the necessary skill profiles
both for the languages and for the supporting tools. He proposed that the
standardisation work should focus on communicating embedded systems, covering
signalling and state machines. He pointed out that some lack of harmonisation
between SDL, MSC and TTCN is a major obstacle to their combined use. To overcome this obstacle,
SOLINET has an in-house tool supporting an enhanced sub-set of SDL including
many of the test features of the TTCN language.There is a need for very efficient development
features. He also added that UML may become a wrapper for SDL. Following his
speech, there was a discussion on static versus dynamic instantiation.
Mrs. Schieferdecker provided a tutorial on the new
Testing and Test Control Notation, TTCN-3. The scope and application area of
this description technique is considerably enlarged compared to previous
version of TTCN, which basically covered protocol testing only. She noted that
TTCN is not state based, like SDL.
Mr. Sanders pointed out that the family of
languages is large and complex for students, so that a selection of
language elements has to be made. He informed that within teaching there is a
competition with an "all UML approach", but that UML is not
yet able to replace SDL, MSC, ASN.1 and TTCN. Finally, he emphasised the
importance books and tool support has for teaching, and that this may
cause teachers to turn to UML, despite advantages of the other languages.
Mr. Amyot provided a tutorial on the User
Requirement Notation, URN, which has a focus on specifying goals and scenarios.
The URN descriptions can provide a first step of system development, but it was
commented that traceability to more formal specifications is
not yet supported.
Mr. Reed stated that SDL is felt by many to be too large and complex, and he
proposes to identify core features of SDL, which concentrate on the
computational viewpoints. He characterised SDL as a system design and
implementation language, allowing code generation from SDL specifications. He
mentioned that SDL with extensions could cover TTCN. He showed that the state
machine aspects of SDL could be isolated to STATE TYPE diagrams on which
PROCESS TYPE, PROCESS BLOCK TYPE, BLOCK , SYSTEM TYPE would depend. An actual
system description would be based an a SYSTEM TYPE. UML Class symbols could be
used to reference the TYPE descriptions. He argued that the core features of
the language should be separated from the advanced features, for example in
separate Recommendations.
Finally, Mr. Sarma summarised the Workshop and thanked the participants.
Download the document: [ MS-Word |
Zip ]
|
|
|