Test Case Prioritization and Reduction Using Hybrid Quantum-behaved Particle Swarm Optimization

Abstract

Regression testing is an integral part of the software evolution and maintenance phase as it ensures that the modified software is working correctly after any upgrades. Test case prioritization and reduction minimize cost and effort needed for retesting by scheduling critical test cases before the less critical ones and removing redundant test cases. The criticality and redundancy of the test cases depend on several testing criteria. This paper empirically analyzed the effect of different testing criteria like code and fault coverage on the techniques' performance. This paper proposed a discrete Quantum-behaved particle swarm optimization (QPSO) for enhancing efficiency of test case prioritization. The algorithm is improved by replacing the random distribution with Gaussian probability to escape from the local optima. The evolution stagnation issue is further resolved by hybridizing it with genetic algorithm (QPSO-GA). In addition to prioritizing the test cases, the algorithm also reduces the test suite size through the test suite reduction approach. The experiments are conducted on different versions of three pro-grams from the open-source software infrastructure repository. The performance is compared with the average percentage of statement coverage, fault detection, and their combinations with the cost. Consequently, suite reduction, fault detection capability losses, and coverage loss percentage are also drawn for test suite reduction. The proposed algorithms outperformed the random search, ant colony optimization, differential evolution, GA, PSO, and adaptive PSO for all the evaluation metrics.