Research On Mutation-based Software Fault Localization

In recent years, with the progress of science and technology, computer software has been used in various fields, and becomes an indispensable part of people’s work, study and life. With the expansion of the software size, software failures are also constantly generated, find and repair these failures will spend a lot of effort and time. Therefore, locating faults quickly and accurately is an important stage in software develop.However, most of the automated software faults localization methods are not accurate and stable, and the effect of coincidence test cases is not considered. The mutation-based fault localization has high accuracy and the results are not affected by coincidence test cases. But if the types of mutation operators are not enough or mutation operators are selected improperly, it will affect the localization accuracy. In addition, the large number of mutations will also increase the localization cost.To solve these problems, the main achievements of this paper are as following:First, put the test procedure into lexical and syntax analysis to obtain the abstract syntax tree, use statement level program instrumentation on the syntax tree, and capture the execution path of the test cases, then obtain the mutations of the target statements.Secondly, design the mutation operators. We analyzed the open source program faults, and proposed four kinds of new mutation operators on the basis of the current mutation operators, the using the mutation operator on the target statements in syntax tree to generate a series of mutations.Then, in order to reduce the execution cost, the filtering method of redundant passed tests is put forward, and the selection and execution optimization of test cases and mutations are implemented. Then, calculate the suspicious degree of statement according to the mutations executive information, get the suspicious degree list, and gives an example of the mutation-based fault localization method.Finally, the approach this paper presented was tested using the Siemens Suite as the test data set, the case studies shows that the new mutation operators can improve the accuracy of fault localization to a certain extent. Comparing the method this paper presented with other fault localization method, the experimental results show that, this method has higher precision.