Loading…

Feature specification and automated conflict detection

Large software systems, especially in the telecommunications field, are often specified as a collection of features. We present a formal specification language for describing features, and a method of automatically detecting conflicts ("undesirable interactions") amongst features at the sp...

Full description

Saved in:
Bibliographic Details
Published in:ACM transactions on software engineering and methodology 2003-01, Vol.12 (1), p.3-27
Main Authors: Felty, Amy P., Namjoshi, Kedar S.
Format: Article
Language:English
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Large software systems, especially in the telecommunications field, are often specified as a collection of features. We present a formal specification language for describing features, and a method of automatically detecting conflicts ("undesirable interactions") amongst features at the specification stage. Conflict detection at this early stage can help prevent costly and time consuming problem fixes during implementation. Features are specified using temporal logic; two features conflict essentially if their specifications are mutually inconsistent under axioms about the underlying system behavior. We show how this inconsistency check may be performed automatically with existing model checking tools. In addition, the model checking tools can be used to provide witness scenarios, both when two features conflict as well as when the features are mutually consistent. Both types of witnesses are useful for refining the specifications. We have implemented a conflict detection tool, FIX (Feature Interaction eXtractor), which uses the model checker COSPAN for the inconsistency check. We describe our experience in applying this tool to a collection of telecommunications feature specifications obtained from the Telcordia (Bellcore) standards. Using FIX, we were able to detect most known interactions and some new ones, fully automatically, in a few hours processing time.
ISSN:1049-331X
1557-7392
DOI:10.1145/839268.839270