Elicitation of a Complete Set of Non-Functional Requirements.

Requirements engineering is the first activity in engineering a softwarebased product. Making mistakes in such an early phase has a strong impact on all subsequent software development phases. Especially nonfunctional requirements (NFRs) play an important role for the success of a project or product. In today's practice, essential information on a system's NFRs have often not been elicited properly and are thus incomplete. The NFR methodology described in this thesis provides a systematic approach for the elicitation, analysis and specification of a complete set of NFRs. To achieve this, the NFR methodology contributes an algorithm to elicit NFRs, which takes quality model information as input and systematically processes specific elements of the functional specification. Checklists and tool support are used to support and partially automate the NFR methodology. The NFR methodology was evaluated in a series of eight, mainly industrial, case studies. The evaluation showed that the NFR methodology is feasible and results in a more complete set of NFRs. The ratio of newly identified, important NFRs ranges from over 100% to 622%. In two case studies, the NFR methodology application resulted in an ROI > 2 and an ROI > 17, respectively.