DefinePK

Abstract

Software development projects often face escalating costs associated with identifying and rectifying faults as they progress through different phases. The expense of addressing faults post-delivery is significantly higher than when detected during earlier development stages. The choice of development methodology influences the cost disparity between phases, with agile practices claimed to yield a less steep fault cost curve. However, for systems with intricate interaction requirements, defects may evade detection at the unit or component level, emphasizing the need for robust integration and system-level testing. In the realm of increasingly complex software requirements, Model-Based-Testing emerges as a technique generating test cases from a system model. Despite its potential benefits, there exists limited understanding of the advantages and challenges associated with Model-Based-Testing at various testing levels. Bridging this knowledge gap is crucial for successful technology transfer in industrial software development. This thesis delves into the application of Model-Based-Testing at the system level, commencing in the early stages of development. The study also employs Model-Based-Testing on a subsystem of a message gateway product to enhance early fault detection. Through this exploration, a set of challenges and recommendations for system-level, Model-Based-Testing are presented based on gathered experiences. Preliminary findings suggest that Model-Based-Testing significantly enhances fault detection during system testing. The research too employs a convergence of quantitative and qualitative data, utilizing fault-slip-through as a metric to gauge the relationship between early application of Model-Based-Testing and the number of undetected faults until customer acceptance testing. The study focuses on black-box functionality testing, encompassing both positive and negative scenarios, using a combination of manual and automated test cases. Furthermore, the research delves into the perceptions of managers and developers regarding the impact of Model-Based-Testing through observations and qualitative interviews. By shedding light on the challenges and benefits of implementing Model-Based-Testing at the system level from early development, this thesis contributes to the advancement of effective testing practices in the ever-evolving landscape of software development.