A General Imperfect Software Debugging Model Considering the Nonlinear Process of Fault Introduction

In general, an assumption of perfect debugging is used to build a software reliability model in the software testing process. However, it is not in accordance with the realistic testing situation. In practice, software testing is a complicated process and includes a lot of impact factors, such as testing environment, testing skill and testing tool, etc. Therefore, a new fault can be introduced when the detected fault is removed. It is a well-known imperfect debugging process. When the code size is larger and complex, more faults can be introduced. Furthermore, the cumulative number of new introduced faults shows a nonlinear growth over time. In this paper, we propose a general imperfect software debugging model considering the nonlinear process of fault introduction and use three historical fault data sets to validate our proposed model. The experimental results show our proposed imperfect debugging model has a better fitting and predicting performance compared with other software reliability models. The assumption considering the nonlinear process of fault introduction is in line with fault introduction changing over time in the practical software testing process. Moreover, our proposed model can effectively fitting the historical fault data and accurately predict the software failure behavior in the actual software testing process.