Pairwise Test Generation Based on Parallel Genetic Algorithm with Spark
Rongzhi Qi, Zhiting Wang, S.Y Li · Advances in computer science research · 2015
Pairwise testing is an effective combinatorial test generation technique that can generate tests covering all pairs of parameter values.Genetic algorithm has been used for pairwise test generation by researchers.It can often produce smaller test suite, but typically require a longer computation.To solve this problem, in this paper we use spark, an in-memory and iterative computing framework, to parallelize genetic algorithm for pairwise test generation.We propose fitness evaluation parallelization, which evaluates each individual's fitness value on spark's workers.A preliminary evaluation of the proposal algorithm is conducted to verify the effectiveness compared with those of other algorithms published in the literature.Experiments show that the proposed algorithm can generate better results among these algorithms.