Improved Delta Debugging Based on Combinatorial Testing

Software fault diagnosis is a process of locating the source of faults based on the testing result (pass or fail) of each test case. It plays an important role in software debugging. However, because of the continuous expansion in software size and complexity, it becomes more and more difficult to diagnose software faults quickly and effectively. Combinatorial testing (CT) is a widely used black-box testing method. Currently, there exist some fault diagnosis methods based on CT to locate the source of faults. But they have not made full use of all information resulted from the CT process, and thus have not been very cost-effective in fault localization. This paper studies how to locate faults based on the test results of CT using a method of Delta Debugging (called Isolation). Two isolation methods, Repetitive Isolation (RI) and Strengthened Repetitive Isolation (SRI), are proposed. These two algorithms differ in the amount of CT information utilized by them. A series of experiments show that comparing to existing debug algorithm, the SRI algorithm is more cost-effective.