Structural test data generation based on harmony search

Abstract

Software testing has been validated as an effective way to improve software quality. Among all research topics in software testing, automated test data generation has been viewed as one of the most challenging problems. In recent years, a typical solution is to adopt some meta-heuristic search techniques to automatically tackle this task. In the paper, our main work is to adapt an emerging meta-heuristic search algorithm, i.e. harmony search (HS), to generate test data satisfying branch coverage. Fitness function, also known as the optimization objective, is constructed via the branch distance. In order to verify the effectiveness of our method, eight well-known programs are utilized for experimental evaluation. According to the experimental results, we found that test data produced by HS could achieve higher coverage and shorter search time than two other classic search algorithms (i.e. SA and GA). © 2013 Springer-Verlag Berlin Heidelberg.