Requirements Engineering for Large-Scale Agile System Development: A Tooling Perspective

Abstract

Context: Companies are using the agile methodology for implementing their softwareintensive projects. However, different studies have indicated that using the agile methodology in a large-scale context leads to different requirements engineering challenges. According to some studies, tool-support has shown positive results in addressing large-scale agile requirements engineering (RE) challenges. But the current tooling solutions have many limitations in addressing large-scale agile RE challenges. Objective: This study tries to investigate tool-support related large-scale agile RE challenges and proposes tooling solutions to those challenges by enhancing an existing RE tool called T-reqs. Method: The design science research has been used to conduct this study in three different iterations. Data has been collected using four workshops, two focus groups, one interview and literature. During the workshops, observation and survey questionnaires were used to collect data. Interview data has been analysed using thematic analysis. The functionality assessment results of the artifact which is produced as a result of this study have been analyzed by summarizing and visualizing them using bar-charts. To analyze the usability assessment results of the artifact, the system usability scale (SUS) was applied. For each item of assessment, SUS scores were calculated and interpreted. Results: The results of this study are a set of tool-support related large-scale agile requirement engineering challenges and their corresponding tooling solutions. The main tool-support related large-scale agile RE challenges identified in this study are difficulty in working with model-based system requirements, difficulty in propagating requirement changes to multiple teams and projects (repositories), difficulty in keeping track of requirement change history, difficulty in balancing the autonomy of agile teams and the need to control requirement documentation, and updating traces in a model and concurrent artifact ID generation. Tooling solutions have also been implemented for those challenges. According to the results of this study, the implemented solutions address the identified challenges.