Automatic test generation from behavioral patterns

This paper presents a method for automatic test case and test sequence generation from behavioral patterns. A behavioral pattern represents a class of behaviors of the environment with which the software under test interacts. Behavioral patterns are provided by the human tester to complement functional testing based on random test data generation. Thus, the tester can bring all her knowledge of the situations which are worth being considered. In a manner, a pattern is a form of an operational pro le, since probabilities are associated with the events which occur in this pattern. Patterns can be expressed graphically or textually; they combine conditions on instants and intervals, each condition being a temporal logic formula. Patterns can be composed using di erent operators. This method is embedded in a tool, Lutess, which provides an automated and dynamic generation of test data sequences for testing synchronous software. An industrial-like case study is reported, which demonstrates the adequacy of the method when dealing with complex environment behaviors. The case study refers to the feature interaction problem, which hampers the rapid creation and integration of new features in telecommunication systems.