Can software debugging be more accurately modeled? This research proposes a new class of software reliability models. These models incorporate practical insights about the failure process and the effect of debugging on software failure rates. The models assume that each correction removes some previous software faults, but also potentially introduces new ones. Several existing models fit within this class. With further assumptions, a simplified model is developed and its parameters are estimated using maximum likelihood equations. Applied to real-world data, the model's performance is compared to that of established software reliability models. This research offers a more nuanced approach to software reliability modeling, potentially leading to more accurate predictions and improved software development processes. The result are compared to usual software reliability models.
This paper, appearing in the International Journal of Reliability, Quality and Safety Engineering, aligns with the journal's focus on improving the reliability and safety of engineered systems. By developing a new software reliability model, the research directly contributes to the journal's core subject area.