Software Reliability Analysis Using Prediction Models
Ali Imran Haidry, Tahir Jameel, Ayesha Naveed, Raaiha Riaz, Laraib Razzaq · 2023
Software Reliability analysis plays a vital role to ensure quality and dependability of software systems. In order to achieve desired reliability of a software product, reliability prediction in early phases of software development life cycle is beneficial for effective resource allocation, risk assessment, and decision-making. These early decisions benefit in twofold, firstly, they supplement software reliability directly and secondly, reduce rework cost associated with bug fixes at later stages. In this research, we studied different approaches of software reliability prediction and applied Neufelder's Shortcut Model on real projects to evaluate its effectiveness and usefulness. The prediction models are based on historical failure data capturing complex dependencies and relationships among various factors directly influencing software reliability. Both Reliability Prediction and Reliability Growth Estimation models (can be used after testing phase) have their own advantages but the prediction models have an advantage over estimation models that they can predict reliability early in project life cycle enabling corrective decision in a proactive manner to achieve the desired reliability in a more deliberate way. We collected real software metrics for Neufelder's shortcut model of different projects to predict defect density and software reliability. Their effectiveness is evaluated using the designer's assessment and growth estimation based on the failure data of the software products. The results demonstrate that software reliability prediction models can be useful for developer, QA engineer, reliability engineer and at the same time for customers for pre-assessment of software reliability of the desired product. Overall, our research contributes to the advancement for software reliability analysis and empowering organizations to build more reliable and robust software system.