Agile requirements engineering in globally distributed software development projects

Abstract: Requirements engineering remains an important discipline to reduce costs, development times and improve quality in software engineering projects. With Agile methods gaining prominence in a rapidly globalized world, many requirements engineering efforts are today made in distributed contexts, with both teams and stakeholders being separated by physical and organisational distances. At the same time, it is not well understood how agile methods for requirements engineering apply to distributed contexts. This thesis investigates the implementation and use of agile methods for requirements engineering in distributed software engineering contexts. Observations made over a three-month study of the CHAMP project, a joint IT and process development effort between major European truck manufacturers Scania and MAN, are used to assess how commonly practices agile methods perform when implemented over distances. The case study of the CHAMP study suggests that the implementation of agile methods is highly context-sensitive, with limited current opportunities to formulate general heuristics for successful applications. The results of the CHAMP study indicate that distributed contexts hamper team communications when compared to co-located efforts, making it more difficult to implement an overall agile project model. However, individual methods, particularly the use of work backlogs, are found to offer increased structural flexibility beneficial to distributed workflows. Additionally, the CHAMP observations suggest implementing agile methods in new contexts requires an organisational mandate, as agile workflows are less predictable than linear models and can expose the surrounding organisation to higher uncertainty.