Analysing the role of non-functional requirements in organizational energy consumption

Abstract

Energy efficiency is an increasingly critical concern in the design and operation of soft- ware systems. Non-functional requirements (NFRs), particularly those related to qual- ity attributes (QAs) and CIAP ratings, play a crucial role in shaping energy use. This thesis presents an automated framework for extracting NFRs from software architecture documents using pattern-based string matching by extended ISO 25010. By analyzing a dataset of architectural documents (containing the NFRs) and energy consumption, we identify statistically significant relationships between documented NFRs and reported en- ergy consumption. Significant findings include that QA ’Functional Suitability’ has the strongest association with energy consumption (linked to a 3% increase as per Gamma GLM Regression) while the QA ’Reliability’ has the strongest negative association with energy consumption (linked to a 28% decrease as per OLS Regression), and CIAP ’P’ has marginal effect on energy consumption. Overall, this work contributes to the field of sustainable software engineering by providing a replicable method for mining NFRs, un- covering quantifiable NFR-energy use relationships, and suggesting practical improvements to documentation and monitoring practices. The insights derived can inform both design decisions and runtime optimizations aimed at reducing energy consumption in software systems.