A requirement-driven framework for cloud service provider selection using AHP, QFD, and TOPSIS

Cloud service provider (CSP) selection has become a complex decision-making task as users must evaluate multiple functional requirements alongside competing quality-of-service (QoS) attributes. Although existing multi-criteria decision-making approaches can rank cloud services effectively, most lack a transparent mechanism for translating user requirements into measurable evaluation criteria. To address this limitation, this paper proposes a requirement-driven cloud service selection framework that integrates the Analytic Hierarchy Process (AHP), Quality Function Deployment (QFD), and Technique for Order Preference by Similarity to Ideal Solution (TOPSIS). The proposed framework first prioritizes user-defined functional requirements using AHP, subsequently maps these requirements to QoS attributes through QFD, and finally ranks CSP alternatives using TOPSIS based on requirement-aligned attribute weights. Experiments were conducted using real-world cloud service performance data comprising ten cloud service providers evaluated across five QoS attributes, namely response time, cost, security, scalability, and availability. The AHP phase produced highly consistent requirement priorities with a consistency ratio of 0.0074, while QFD-derived weights identified response time (0.264) and cost (0.204) as the most influential QoS attributes. The proposed framework consistently ranked CSP5, CSP9, and CSP2 as the most suitable providers. Comparative analysis against AHP-TOPSIS, MOORA, and VIKOR demonstrated strong ranking agreement with Kendall's Tau values exceeding 0.86, while Wilcoxon signed-rank tests confirmed statistically significant differences in ranking outcomes. Furthermore, sensitivity analysis verified ranking stability under alternative QFD relationship scales, and scalability experiments showed execution times below one second for scenarios involving up to 50 cloud service providers and 10 QoS attributes. The results demonstrate that the proposed framework offers an interpretable, robust, and computationally efficient solution for user-centric cloud service selection by explicitly linking user requirements to provider evaluation and ranking.