Reliability analysis of open source software systems considering the effect of previously released version

Abstract

ABSTRACTIn this study, reliability analysis of open source software in an imperfect debugging environment has been discussed by considering the effect of detection rate and the remaining faults in successive releases. First, a modified non-homogeneous Poisson process model is developed in an imperfect debugging environment by assuming that new faults may be introduced in fault detection and correction process. Then, the detection rate and remaining faults of previous release are considered in the modeling and analysis of reliability of the current release. The optimal version-update time is computed using multi-attribute utility theory for these software systems, considering the two main utility factors, namely, rapid release strategy and level of reliability. The proposed models are verified on real data sets and are compared with other existing models. The proposed decision models may be helpful for the software developing management to decide the optimal version update time for open source software.