Fault Localization Based On Combinatorial Testing

Combinatorial testing as an important testing method has been wildly used. It can de-tect failure caused by some factor or their interactions while generates very few test cases. When a failure has been detected, identifying the factor combinations which cause the failure can help developers to make future improvements for system. So it is required to make fault localization efficiently after combinatorial testing. Combinato-rial testing has a strong competitive advantage in fault detection, but cannot accurately locate the factor combinations that cause the failure. Until now, there are several fault location methods based on combinatorial testing have been proposed, but most of them have deficiencies such as:imprecise locating result, high cost and lack of fully using information from combinatorial testing.We introduce Delta Debugging method to locate faults after study on the basic principle of combinatorial testing. And we propose a new locate method called RI (Repetitive Isolation) to solve the problem that, as one classical algorithm of Delta Debugging, Isolation algorithm can only locate one parameter value that relative to the failure. By using the information from combinatorial testing, an enhanced method SRI (Strengthen RI) will be proposed too. Simulation experiment has been conducted to compare the efficiency among RI algorithm, SRI algorithm and other existing methods. Experiment data has showed that RI algorithm has a higher efficiency over all other existing methods, and SRI algorithm which has used extra information improved the efficiency distinctly over RI algorithm. Empirical study about the compatibility testing of software on Windows platform has verified that our SRI algorithm has a very high practicability.