Test Data Generation for Path Coverage using Mean Variance Mapping Optimization

Abstract: In Search Based Software Testing (SBST), the problem of generating test data for path coverage is first formulated as an optimization problem and then is solved by meta-heuristics. A notable trend of research in this field is to reduce computational time or the number of times a test program are executed, since running a test program may be time consuming. In this paper, a meta-heuristic named Mean Variance Mapping Optimization (MVMO) is adopted with appropriate settings (e.g. representation of test data, objective function) to solve the problem. Six popular test programs for path coverage testing in SBST are used to evaluate the method. In comparison with the most popular algorithm (Genetic Algorithm - GA) and the most recent and efficient algorithm that combines dynamic stopping criteria with GA (called DSGA) in path coverage testing problem, MVMO has outstanding efficiency and effectiveness. MVMO can save generally an average of 40% - 95% and more than 50% of total computation respectively. The number of target paths exercised by MVMO is also slightly higher than those by GA and DSGA in some test programs. This is the result of using a novel mapping function which implies an efficient strategy to allocate computational resources for search space exploration and exploitation. If the available computational resource is large, the method prioritizes to explore the search space otherwise it prioritizes to exploit the search space. It is expected that this strategy can achieve the same performance when applying to other meta-heuristic algorithms in generating test data for path coverage.

Keywords: Evolutionary Algorithms, Mean Variance Mapping Optimization, Path Coverage Testing, Search Based Software Engineering.

Faculty of Information Technology, Le Quy Don Technical University, Vietnam, v.v.pham2012@gmail.com

[PDF]

Cite: Viet Van Pham, "Test Data Generation for Path Coverage using Mean Variance Mapping Optimization," Progress in Intelligent Computing and Applications , vol. 5, no. 1, pp. 14-22, March 2016.