Coupling based integration testing: A fitness function

Test data generation is one of the most important and crucial phases in software testing. Software testing is not possible without adequate test data. Many automated and manual test data generation techniques have been proposed for software testing. Most of the work on automated software test data generation at unit level is by applying evolutionary approaches for test data generation. Evolutionary approaches, especially genetic algorithm, use fitness function for evaluation of individuals in different iterations. In this paper, we have proposed a novel fitness function for test data generation at integration level. Fitness function plays a vital in the success of evolutionary testing, without effective fitness function evolutionary testing is not effective for achieving required results. The success of evolutionary testing depends upon the success of fitness function. We have proposed a novel fitness function for coupling based integration testing. Up until now, there is no fitness function that caters coupling based integration for test data generation. Up till now most of the work for test data generation is at unit level and fitness function also cater only unit level test data generation. We have implemented our fitness function in a prototype tool `EE-COUP´ and performed different experiments for test data generation for some sample programs containing coupling relationship.