Research On Combinatorial Test Generation With Parameter Constraints

With the continuous development of computer technology, software testing has gradually become a resource-intensive activity, so people have been studying how to reduce the cost of software testing. Experience and practice show that the combinatorial testing is a practical and effective testing method. It assumes that the defects are often triggered by the interaction of a small number of parameters. Because combinatorial testing can use a smaller test set to achieve a scientifically valid test of the system under test, it has been widely studied and applied.Combinatorial test generation technology has been an important research in the field of the combinatorial testing. The second chapter introduces some basic concepts of combinatorial testing, as well as several common combination of test case generation algorithm, and describes several advantages and disadvantages of these algorithms.Traditional combinatorial test generation methods generally assume that the parameters in the software system under test are independent of each other. However, in practical applications, there are often a certain constraint relations among the various parameters of the software. The third Chapter introduces four constraint handling strategies, and describes the procedure that how these strategies handling constraints through examples. The chapter also analyzes a prerequisite for the use of these four constraint handling strategies, and gives a simple algorithm to reduce redundant test cases induced by three of those strategies.Because soft constraints have little effect on error detection,the paper mainly aims at hard constraints.Consider these constraints in the design of test cases, it’s possible to further reduce the size of the test suite under the premise of not affecting the error detection ability of the system,and it also have more practical value. Because of the complex relations among the parameters, it may occur that the given parameter constraints induce new constraints, that is implied constraint. If implicit constraint handling is ignored, test case generation algorithm may lead to abnormal. In order to solve this problem, the paper proposed an algorithm that is able to handle the implicit constraint. The algorithm handles the given constraints to obtain implied constraints and incorporate them into original parameter constraints. Then all constraint relations are obtained. Finally an example illustrates the effectiveness of the algorithm.The paper is based on PSST algorithm, adding parameter constraints. It can generate a smaller test suite under the premise of not affecting the coverage. The basic steps of the algorithm are as follows. First, check whether the constraint set given by the system under test contains implied constraints. If it contains, solve the implied constraints and incorporate them into the original constraint set. The combination of constraints in the constraint set can not appear in the final generated test cases; Second,Use backtrack algorithm to traverse the solution space tree, and at the same time, cut the branches of solution space tree utilizing the constraint relations among parameters to avoid invalid test cases and reduce the number of traverse. When leaf node is traversed, the path from the root to the leaf node is a test case; Third, check whether the test suite generated by the backtrack algorithm contains all pairs of parameter values. If there still have combination pairs that are not covered, then some test data may be added into the test suite by a greedy algorithm to cover all the uncovered pairs. The experiments in fifth chapter show that this method can also effectively reduce the number of test cases.