1. Participants and Collaborators
Nicolas Gorse
LOTOS Team
SITE - School of Information
Technology & Engineering
University of Ottawa

2. Objectives Specification Phase Testing Phase
Feature Interaction avoidance and detection in early stages of development of a system.
Improvement of the design/specification phase.
Testing Phase
Automatic generation of test suites.
Reduction of the testing time and cost.

3. Main Ideas Specification Phase Testing Phase
formulate logic specifications of features
potential interactions will show as contradictions
Testing Phase
test cases are constructed from information on the contradictions found
they will show if the potential interactions are real ones

4. FI Detection Using Predicate Logic, UCM and LOTOS (cont’d)
Derivation of a LOTOS specification from UCM (see companion project)
Provides an executable model - prototype
Provides information for scenario generation
Scenario generation for possible interactions identified
Using information on the structure of the feature
Based on possible interactions identified

5. FI Detection Using Predicate Logic, UCM and LOTOS (cont’d)
Feature Interaction scenario-based validation of the LOTOS specification
Allows to verify whether the possible interactions identified are present in the LOTOS spec
Method only identifies possible interactions, however experimental study showed very high hit rate
Scenarios derived can be reused at final system testing stage

6. Process Illustration

7. Process Illustration (cont’d)
First identification of possible feature interactions and requirements improvement
Graphical representation of the system
Formal specification allowing to perform Scenario-based testing and model checking
Generation of validation test suites Corresponding to the interactions identified
Detection of FI in the LOTOS specification

8. Feature Interaction Filtering Using Predicate Logic
Representation of features
Pre-conditions
CFA: {subs(B, cfa), concerns(B, cfb), cfa(C)}
CFB: {subs(B, cfb), concerns(B, cfb), busy(A), cfb(D)
Triggering Events
CFA: {call(A, B)} Same triggering events
CFB: {call(A, B)} for both features
Results
CFA: {call(A, C), connect(A, C)} Different results,
CFB: {call(A, D), connect(A, D)} non determinism

9. Process Illustration

10. Lessons Learned Mitel project Feature Interaction contest
22 feature descriptions (484 pairs), 4 users
43 possible interactions found in secs
Feature Interaction contest
97 feature descriptions (9409 pairs), 4 users
149 possible interactions found in secs
Fairly quick representation of features
Rapid method, accessible to non-specialists

11. Deliverables and Project Plan
Achieved
Method fully developed
Master Thesis Submitted
Future work
Improvement of the method for detection of new kind of Feature Interactions
Addition of a Graphical User Interface for the Design of features
Distribution of the algorithm over a network

12. References Use Case Maps Feature Interactions
Various documents about Use Case Maps, UCM web site,
Feature Interactions
K. Kimbler, L.G. Bouma, Feature Interactions in Telecommunication and Software Systems V, IOS Press, 1998
M Calder, E.H. Magill, Feature Interactions in Telecommunication and Software Systems VI, IOS Press, 2000