A Method of Statistical Process Control for Successful Open Source Software Projects and Its Application to Determining the Development Period

A software development paradigm for open source software (OSS) project has been rapidly spread in recent years. On the other hand, an effective method of quality management has not been established due to the unique development characteristics such as no testing phase. In this paper, we assume that the number of fault-detections observed on the bug tracking system tends to infinity, and discuss a method of statistical process control (SPC) for OSS projects by applying the logarithmic Poisson execution time model as a software reliability growth model (SRGM) based on a nonhomogeneous Poisson process (NHPP). Then, we propose a control chart method based on the logarithmic Poisson execution time model for judging the statical stability state, and estimating the additional development time for attaining the objective software failure intensity, i.e., the target value of the instantaneous fault-detection rate per unit time. We also discuss an optimal software release problem for determining the optimum time when to stop OSS development and to transfer it to user operation. Further, numerical illustrations for SPC are shown by applying the actual fault-count data observed on the bug tracking system.