Research On Fault Location Based On Error Explanation

With the development of computer system, demands of software become more and more higher, so we have become increasingly concerned about the research project of enhancing software quality. The reliability, accuracy, security of software are the key factors of ensuring software have higher quality. Now, how to locate software failure after it was wrong is the hot spot in the area of software failure.Locating of the cause of software failure is the heart of procedure diagnose. When the software is failure, a model checker will automatically produce a counterexample, this counterexample show some abnormal behavior. However, the user must decide if the counterexample genuinely shows erroneous behavior or it is just caused by incorrect specification. In the event that the error is real, there remains the challenging task to isolate and modify the faulty aspects of the system. The research of this paper is based on similarity path measurement and error explanation. We describe a path similarity measurement algorithm for data flow difference and a fault location method based on error explanation which based on Groce's method. Finally, we use a simple program to do the experiment of fault location, with contrast and analysis of the test results , it show that the method of procedure fault location mentioned in this article in handling the data flow errors can be effectively located fault.In our research, the following aspects have been researched:1. The research on the criterion for similarity measurement based on data flow.In this paper, it defines a similarity measurement criterion based on data flow differences; it gives a specific algorithm about path distance, and applies the algorithm to the fault location algorithm based on an error explanation.2. Research on Fault location Based on Error explanationIn this paper, with the use of static slicing in the prophase of the fault location, a fault location method based on error explanation is proposed. Groce proposed a fault location method with the use of dynamic slice (A slice) in the late period, so the code amount in the pre-processing is more, and needs dynamic tracing the history of procedure executing, the cost is much higher. In order to solve above problems, we use the static slice (the interprocedual slicing of KJOttenstein and LMOttenstein) in the prophase, so it reduces the code amount in the late period, making error explanation better. On this basis, a method of fault location based on error explanation is proposed.3. The analysis of experimental resultsAccording to the fault location method mentioned in this article, using a specific C program as an example and on the basis of the distance measurement algorithm mentioned in this article, we do a fault location experiment, and give the analysis of experimental results. The results show that the proposed method has high accuracy in localizing the program faults related to program data flow errors.