On the performance of EvoPSO: A PSO based algorithm for test data generation in EvoSuite

Abstract

Nowadays software has a major role in our everyday life. Many critical tasks are done by software systems. The increasing complexity of software systems compels providing techniques and tools to design correct and well-functioning software in safety-critical systems. Up to 50% of the total software project costs are devoted to testing; hence, increased concern of automated software testing in recent years. The automation of software testing reduces costs and improves the effectiveness of tests that are generated in order to detect defects in the software under test. Various techniques are adopted for automated software testing including metaheuristic search algorithms. In this paper, we propose the EvoPSO algorithm using swarm intelligence paradigm. The algorithm is implemented in EvoSuite tool for the purpose of test data generation. The performance of EvoPSO has been investigated on SF110 dataset. The promising performance shows that EvoPSO is efficient and can give competitive results.