University of Ottawa (CANADA)
ELUDO (Environnement LOTOS de l'Université d'Ottawa)
LOLA (LOTOS Laboratory)
CADP (Construction and Analysis of Distributed Processes)
In the design of telephone systems, adding new features (e.g., call
forward, caller name display, incoming call screening, etc.) with
respect with a given system consists of modifying the system such
that it incorporates the modifications while retaining its previous
functionality, to the extent set by the new features. However, the
introduction of new features may cause undesirable effects, called
feature interactions, on the behaviour of the telephone system:
for instance, a feature like incoming call screening can, under certain
circumstances, prevent a call from being terminated. These problems
have motivated an important research effort in order to detect feature
interactions (FI). On the occasion of the 5th International Workshop on
Feature Interactions in Telecommunications and Software Systems
(FIW'98), an FI detection contest was proposed. The goal of the contest
was to compare the efficiency and the adaptability of different methods
and tools in the detection of FI.
A group of the University of Ottawa, leaded by prof. Luigi Logrippo, participated to the contest. The approach used by the team was based upon the use of LOTOS for specifying the basic system and the features, along with the feature properties. Several LOTOS-related tools were used to assist the FI detection process: ELUDO was used to develop the LOTOS specifications, CADP was used to construct and explore finite LTS representations of the behaviour of the system, and LOLA was used to validate the specification against test cases.
After obtaining the LOTOS specifications, a validation activity was performed in order to ensure that various components of the system were properly specified individually, and each feature by itself has its own desired properties. Then, two different methods were used to detect FI. The first method consists of assembling the LOTOS specification of the telephone system with a pair of features selected among the features under study, and to search in the resulting state space an execution sequence leading to an error action (which corresponds to an interaction between the pair of features considered). The second method works similarly, except that it focuses on specific scenarios, i.e., execution sequences that are manually chosen for each feature and that lead the system in desired states where problems are more likely to occur. The first method allowed to discover 39 pairwise interactions in the set of 12 features proposed, and the second method allowed to identify 154 interactions, of which 73 were in the contest committee list.
The LOTOS language has two main assets in the area of specifying
telephony systems and their features: it is capable of representing
system structure clearly, and it has a good set of validation tools.
This enabled to obtain good results in using a LOTOS-based approach for
detecting feature interactions. The methodology proposed appears to be
sufficiently mature for experimental use in industrial environments.
Q. Fu, P. Harnois, L. Logrippo, and J. Sincennes.
"Feature Interaction Detection: a LOTOS-based Approach".
Computer Networks 32(4):433-448, 2000.
Available on-line at: http://LOTOS.csi.UOttawa.ca/ftp/pub/Lotos/Papers/FI_Detect_LOTOS_Based.pdf
or also from our FTP site in PDF or PostScript
Prof. Luigi Logrippo
School of Information Technology and Engineering (SITE)
University of Ottawa
150, Louis Pasteur
PO Box 450, Stn A
MacDonald Hall, Room 310-B
Canada, K1N 6N5
Tel (direct): (613) 562-5800 x 6704
Tel (secretary): (613) 562-5800 x 6676 or (613) 562-5826
Fax: (613) 562-5187 or (613) 562-5185
|Further remarks:||This case study, amongst others, is described on the CADP Web site: http://cadp.inria.fr/case-studies|