Enumeration Problem In Evolutionary Testing

Evolutionary testing is an emerging technology for automatic test data generation, it has been successfully applied in engineering practice. However, in white box testing, the search process of evolutionary testing will be hindered in the presence of Enumeration variables, and degenerates into random testing. We describe it as enumeration problem. There is no relative research on this problem recently, but the research on flag variable has become popular in evolution testing. As in this situation, we present a unified rule for fitness function calculation for enumeration problem.In this article, we analyze the similarity between enumeration problem and flag problem and extend the unified rule of flag variable to enumeration. According to their differences, we give the new iterative rule and finally present the fitness calculation rule for enumeration problem. Experiments showed that evolutionary search can find test data under the guidance of our unified rule in the case that previous approaches failed.When we extend the unified rule of flag variable in fitness calculation to enumeration variable, the problem of code redundancy and location selection will be encountered. These problems will affect the cost of embedding and performance of fitness calculation. This article analyzes these problems and presents the solutions of code embedded in the rule of enumeration variable fitness calculation. In the end, we present the design and implementation of algorithms in testing data generation.