Search Based Combinatorial Testing

Combinatorial testing is an effective method to detect the failures triggered by fac-tors and their interactions. This dissertation focused on the search based combinatorial testing, which aims to use the search techniques to generate, minimize and prioritize the covering array (i.e. test suite) in combinatorial testing. The followings list the main contributions of this dissertation.Firstly, the performance of search techniques is significantly impacted by their parameter settings, and these settings should be adapted for different problems. To further enhance particle swarm optimization for covering array generation, we explored its optimal parameter settings, and proposed an adaptive search strategy. This new adaptive approach does not need specify the settings by hand, and can generate smaller covering arrays than the standard version of particle swarm optimization.Secondly, except by mathematical construction, covering arrays generated by all other methods still contain many redundant elements. Current minimization algorithms can reduce the size to some extent, but their random search strategies are less effective. To further reduce testing cost, we refined the existing post optimization process, and proposed an estimator to predict the extent to which covering arrays can be improved. This new approach can reduce more test cases within a reasonable time, while still maintaining full combination coverage.Thirdly, current researches in combinatorial testing mainly focus on prioritizing test suite by combination coverage, while reasonable time and effort is required to change configuration settings when switching test cases for highly configurable soft-ware systems. To address this issue, we proposed a switching cost based test suite prioritization strategy, and developed a hybrid search approach to find the best execu-tion sequence of a given test suite. This new strategy can not only reduce the whole execution cost, but also improve the fault detection ability.At last, we present an empirical study of the functionality testing of mobile phone voice call. The above covering array generation, minimization and prioritization ap-proaches are applied to generate its test suite. The results further demonstrated the effectiveness of search based combinatorial testing.