Performance Comparison of Model Based Test-Suite Optimization Using Evolutionary Algorithms and Greedy Heuristic Algorithm

摘要

Model Based Test-Suite (MBT) is an automated approach for generating test cases automatically from the model under test. Optimization techniques help the tester to generate good test cases that satisfy the coverage criterion specified and that can detect faults easily. The main aim of this study was to generate and optimize test suites that are relatively easy for testers to use and which can meet the needs of industrial model based testing. This paper discusses performance of evolutionary computation-based minimization and prioritization techniques for model based test suites compared to greedy heuristic technique. Evolutionary algorithms used for this purpose are simple, steady state, struggle genetic algorithms, evolutionary comma strategy, evolutionary plus strategy and evolutionary programming. Test suite minimization and prioritization was achieved using evolutionary and greedy heuristic algorithms. An empirical study was performed to analyze the performance of the algorithms with the proposed greedy heuristic technique. Greedy heuristic was compared with evolutionary algorithms to find the performance in terms of time, runs, priority Average Percentage Block Coverage (APBC), reduction APBC, and reduction percentages. To analyze the experimental data, GraphPad Prism curve-fitting software was used to investigate the statistical significance of any differences observed in the experiments. To measure the performance of the algorithms best-value metrics are used to compare the performance in terms of reduction rates. One way ANOVA (analysis of variance) test was performed to measure the statistical significance of the data. The results show that evolutionary algorithms performed better than the greedy heuristic algorithm for model based test suite optimization.