Search Based Test Data Generation: A Multi Objective Approach using MOPSO Evolutionary Algorithm

Abstract

Search based test data generation plays an important role in software testing. Several search based evolutionary algorithms are used to find the optimal test data. Among these algorithms, a meta-heuristic algorithm called Particle Swarm Optimization (PSO) algorithm is adopted for finding the optimal test data for the given Software Under Test (SUT) due to its simplicity and fast convergence. The success of PSO as a single objective optimizer in the literature has motivated to solve multi objective optimization problems. Hence, Multi Objective Particle Swarm Optimization (MOPSO) is adopted for solving more than one objective. This research work consider two objectives which attempts to maximize the branch coverage and reduce the test suite size. A benchmark program is used for the experimental analysis using MOPSO algorithm. The experimental analysis was performed using MOTestGen tool to extract the results. The extracted results portraits the convergence and coverage performance in producing the optimal test data as the population size increases.