High Dependability Computing Program: Evolving a Dependability Requirements Elicitation and Modeling Framework Based on Use

Abstract

Correctly identifying and expressing dependability requirements for software systems has wide-ranging consequences for planning and conducting software development as well as for the final system success. Yet crucial difficulties exist, many stemming from the fact that definitions of “dependable” will vary not only from system to system, but will be perceived differently by different stakeholders of the same system.

UMD is a requirements engineering framework for eliciting and modeling dependability requirements that has been devised, to mitigate such difficulties.

In this report, we introduce UMD and describe an empirical study designed to shed some light on the feasibility of the ideas behind UMD and to identify which aspects of the framework could be improved, in the perspective that software technology transfer from research to industrial use should proceed iteratively and empirically. Subjects in the study consisted of 7 students in a graduate-level class.

Empirical qualitative and quantitative results show that the UMD approach is feasible but also allowed us to identify important missing aspects, confirming our assumption that it was not yet mature enough for a rigorous industrial study. The contributions of this study have been twofold: Demonstrating the usefulness of the tech transfer approach which we have followed as well as the feasibility of the UMD approach.