Software Reliability Growth Model Based on Weibull Distribution Introduced Faults

Software debugging is a complex process and affected by many factors, such as debugging resources, debugging tools, debugging skills, etc. When detected faults were removed, new faults may be introduced. Therefore, it plays an important role to research an imperfect debugging phenomenon in the software debugging process. How to model fault introduction in building an imperfect debugging model is still an unresolved issue. So far, numerous software debugging models are developed by researchers, for example, assuming the fault content function is a linear, exponential distribution or proportional to the number of removed faults, etc. However, they can not entirely satisfy the realistic needs due to fault introduction complicated changes over time. In this study, an NHPP software reliability model is proposed based on Weibull distribution introduced faults and the fault content function following Weibull distribution is considered. The related experiment is carried out which validates the fitting and predictive power of the proposed model. The experimental results also show the proposed model has much better fitting and predictive performance than other models using two fault data sets, as well as better robustness. © Copyright 2019, Institute of Software, the Chinese Academy of Sciences. All rights reserved.