Constraint test cases generation based on particle swarm optimization

Although Artificial Intelligence (AI)-based algorithms have made some achievements on t-way testing strategies and perform better than greedy algorithms, there still exist a challenging problem in t-way constraint covering arrays generation. Only few AI-based algorithms can handle constraints currently compared with greedy algorithms. In this paper, we first demonstrate two algorith ms to generate a t-way covering array with constraints based on particle swarm optimization. Particle swarm-based Constraints Test Generator with Avoiding strategy (PCTG-Av) uses the strategy of avoiding the selection of conflicting test cases. It selects the optimal particle which satisfies the constraint validity as the global solution after per iteration, and guides the evolutionary direction. Particle swarm-based Constraints Test Generator with Replacing strategy (PCTG-Re) uses the strategy of replacing conflicting test cases. PCTG-Re verifies the constraint validity of the global optimal solution after the iteration process. If the global optimal solution doesn't satisfy the constraint validity, then replace the parameter values related to conflicting. Finally we evaluate the availability of the two approaches with some excellent existing strategies. The results show that our algorithms have considerable competitiveness.