A Discrete Particle Swarm Optimization for Covering Array Generation

摘要: Software behavior depends on many factors. Combinatorial testing (CT) aims to generate small sets of test cases uncover defects caused by those factors and their interactions. Covering array generation, a discrete optimization problem, is the most popular research area in field CT. Particle swarm (PSO), an evolutionary search-based heuristic technique, has succeeded generating covering arrays that are competitive size. However, current PSO methods for generation simply round particle’s position integer handle search space. Moreover, no guidelines available effectively set PSOs parameters this problem. In paper, we extend set-based PSO, existing (DPSO) method, generation. Two auxiliary strategies (particle reinitialization additional evaluation gbest ) proposed improve performance, thus novel DPSO developed. Guidelines parameter settings both conventional (CPSO) developed systematically here. Discrete extensions four variants developed, order further investigate effectiveness Experiments show CPSO can produce better results using settings, smaller than other algorithms. promising improvement