A UML Model Based White Box Reliability Prediction to Identify Unreliable Components

In general, existing reliability analysis techniques have a limitation in that they normally require run-time information to assess a software system's reliability. So, reliability analysis has been delayed until system testing or operation where principal design decisions had been already made. Also, they normally focus on estimating the reliability of the whole system rather than finding out unreliable sources or components. In this paper, we propose a UML model based white box reliability prediction methodology. The proposed methodology provides the capability of reliability prediction in the early phase by utilizing design artifacts like UML model, and by analyzing the probability of failure in three different levels of the system. Hence, it makes possible to rank the unreliable parts of the system quantitatively not only at the component level but also at the user's system usage level and system's activity level. In order to validate the proposed method, we performed a case study with safety-critical embedded software and compare the results with traditional reliability models. The result of proposed methodology shows the better prediction accuracy as well as the more detailed analysis of the unreliable part of the system.