Efficient backtracking strategies in test generation

This thesis addresses the problem of backtracking strategies in test generation. First, a methodology which uses status of absolute dominators as a means for causing backtracking during the test generation process is presented. Then, different heuristics that force the test generation to execute the backtracking procedure are investigated. Experiments which generated test patterns for over 30,000 faults have been used to evaluate these heuristics. According to the experimental results, we recommend a new backtracking strategy that has the best performance among the six strategies explored in this thesis