Pairwise testing is a practical and effective approach for various types of software systems, since many troublesome faults are caused by the interaction of a few test parameters. In this thesis, we first introduce the methodology of combinatorial design of testing parameters and the effectiveness of pairwise testing. Then we define Pair-Combinatorial Graph (PCG) and Case Sub Graph (CSG) to analyze the pair-cover problem and indicate that to achieve a minimum test set for pairwise testing is an NPC problem, but the size of such test set is in a predictable range.After a comparative studying and analyzing three effective strategies for generating pair-cover test set by means of PCG, CSG and a series of sample systems' experiments, we present an improved algorithm, which has a rule for ordering parameters extension and also solves constraint parameters problem.Finally, based on the improved algorithm, we implement a testing tool for pairwise testing by Java Swing components. This testing tool solved constraint parameter and parameter alternation problems, which are very often seen in practical testing situation. |