Optimal decisions on software release and post-release testing: A unified approach

Abstract

In this research, a novel approach is developed where a testing team delivers the software product first and extends the testing process for additional time in the user environment. During the operational phase, users also participate in the fault detection process and notify the defects to the software. In this study, a reliability growth model is proposed using a unified approach based on the expenditure of efforts during the testing process. Besides, debugging process is considered imperfect as new faults may enter the software during each fault removal. The developed model further considers that the developer’s rate of defect identification changes with a software release. Thus, the software time-to-market acts as a change-point for the failure observation phenomenon. It is asserted that the accuracy of a software reliability estimation improves by implementing the concept of change-point. The main aim of the paper is to evaluate the optimal release time and testing termination time based on two attributes, particularly, reliability, and cost. A multi-attribute utility theory (MAUT) is applied to find a trade-off between the two conflicting attributes. Finally, a numerical example is presented by using the historical fault count data. The behavior of two decision variables is measured and compared with the existing release time strategy.

Keywords

• change-point
• field-testing phase
• imperfect debugging
• multi-attribute utility theory
• optimal software scheduling
• software reliability assessment
• testing effort
• testing duration