Combinatorial testing data generation based on bird swarm algorithm

Combinatorial test data generation is a research hotspot in combination testing. Evolutionary algorithm has been applied successfully into generating covering arrays that are competitive in size. In this paper, Bird Swarm Algorithm (BSA) is introduced to explore the effect of covering array generation. However, no suitable parameter configurations are available to guide BSA to search solutions. In order to determine the optimal configuration of BSA for this problem, parameter tuning makes an operation on it. Moreover, this paper also does three improvements containing the Levy flight, the bird reinitialization strategy, and the dynamic flight frequency on the original BSA to boost its ability to jump out of the local optimal. Experimental results present that BSA for combinatorial test data generation becomes an effective method and that Enhanced Bird Swarm Algorithm (EBSA) can produce smaller covering arrays than the original BSA.