Binary Black Hole-Based Optimization for T-Way Testing

Abstract

Software testing is an important process in software development life cycle, which aims to guarantee the quality of software and reduce the number of errors and bugs. In such a process, software inputs and parameters are used to create a set of testing cases. Nevertheless, the number of testing cases increases enormously when considering all combinations of those inputs. Although t-way testing can reduce the test cases, generating the minimum, yet representative t-way testing set is challenging due to the large search space, which renders finding the best solution computationally prohibitive. The extant solutions suffer from the sensitivity to the random initialization and the subjectivity to the local minima, which adversely affects the reproducibility of these algorithms and obstructs finding the optimal solution. To this end, this paper proposes a novel meta-heuristic searching algorithm called Binary Black Hole (BBH) optimization that formulates the t-way testing as a binary optimization problem. Experimental results show the superiority of BBH over the famous Binary Particle Swarm Optimization (BPSO) algorithm. The achieved improvement shows the capability of BBH in generating smaller covering arrays with the same t-strength compared to those generated by BPSO.