Use Neural Network to Improve Fault Injection Testing

fault injection is an effective technique in software testing. By introducing faults to software under test, fault injection can improve the coverage of a test, as the same time, the fault injected in software contributes significantly to find true fault related to fault injected. In this paper we propose a software testing method based on fault injection. In this method, we first use neural network to calculate the value of fitness function that describe the proximity of two paths, and then use simulated annealing algorithm to generate test data. In this method we use the value of fitness function as the criteria in simulated annealing algorithm. Experiment shows that this method can improve the efficiency of generating test data obviously.