QoS-aware multi-objective SaaS selection system in cloud computing using intelligent learning algorithm

• Published in: Cluster computing Vol. 28; no. 14; p. 939
• Main Authors: Kouchi, Sana, Nacer, Hassina, Slimani, Hachem
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.11.2025; Springer Nature B.V
• Subjects: Availability; Cloud computing; Clustering; Complexity; Computer Communication Networks; Computer Science; Consumers; Customer services; Customers; Decision making; Fuzzy logic; Linear programming; Machine learning; Multiple criterion; Multiple objective analysis; Operating Systems; Optimization; Performance evaluation; Processor Architectures; Quality of service; Search algorithms; Software services; Cloud computing; Cuckoo search algorithm; QoS; Customer profile; Metaheuristic algorithm; SaaS selection
• ISSN: 1386-7857; 1573-7543
• DOI: 10.1007/s10586-025-05642-0

Cloud computing offers businesses increased flexibility and scalability. This evolution has led to the emergence of numerous providers offering services that are similar in terms of technical functionalities but differentiated by their Quality of Service (QoS) attributes, such as performance, security, or availability. This diversity has caused an exponential increase in the number of available services. When a client requests a service, they are faced with a wide range of options, making the selection of the most appropriate service a complex problem, often classified as NP-hard. The complexity is further compounded by the need to consider individual client preferences and requirements, turning the selection process into a true multi-criteria optimization challenge. In this study, we propose a Multi-Objective SaaS Selection System (MO3S) designed to address this issue by processing customer requests and identifying optimal services based on predefined QoS criteria (qualitative or quantitative) and customer preferences. The system leverages a novel mathematical model to facilitate the multi-objective optimization of service quality, ensuring the selection of the most appropriate services. To enhance this process, we implement the Multi-Objective Cuckoo Search Extended Algorithm (MOCSEA), which combines local and global search techniques with a similarity-based search space reduction module and an advanced population initialization mechanism supported by clustering learning. To evaluate our MO3S system, we utilize a prototype in a real OpenStack cloud and a local environment, using two types of datasets. The obtained empirical results demonstrate that MO3S significantly outperforms existing approaches in terms of execution speed, precision, and overall quality of the solutions achieved.