International Journal of Computing and Digital Systems | 2026
Authors: Aitamar Y.; Lamin S.A.L.; Oubaha J.; El Abbadi J.
DOI: 10.12785/ijcds/1571143435
Journal: International Journal of Computing and Digital Systems
Year: 2026
Publisher: University of Bahrain
Document Type: Article
Open Access: All Open Access; Gold Open Access
Cited by: 0
Wireless Sensor Networks (WSNs) stand as a cornerstone of the Internet of Things (IoT), enabling a multitude of applications across various domains. However, the inherent energy constraints of individual sensor nodes pose significant challenges to the scalability and long-term viability of these networks. To address these limitations, this paper presents a novel hybrid clustering methodology designed to enhance energy efficiency and scalability in IoT-integrated WSNs. Our approach synergistically combines the strengths of divisive hierarchical clustering and Bayesian classification. Initially, the network nodes are partitioned into cohesive clusters using a divisive hierarchical algorithm that considers both the spatial proximity and residual energy levels of the sensors. This initial clustering phase aims to create geographically compact and relatively energy-homogeneous groups. Subsequently, a Bayesian classifier is employed to refine these clusters and strategically select optimal Cluster Heads (CHs). This selection process leverages probabilistic reasoning, taking into account crucial node characteristics such as residual energy, node degree, and centrality within the cluster. By integrating probabilistic decision-making, the Bayesian classifier ensures a more robust and adaptive CH selection, mitigating the risks associated with relying solely on deterministic metrics. The subsequent Bayesian-based CH selection optimizes energy consumption by choosing the most suitable nodes to manage cluster activities, thereby distributing the energy burden more effectively across the network. This balanced energy expenditure contributes significantly to extending the overall network lifetime. Furthermore, the adaptability of the Bayesian classifier to varying network conditions and node behaviors enhances the scalability of the proposed method, making it suitable for deployment in diverse and dynamic IoT environments. Extensive simulation results demonstrate the superior performance of our hybrid approach compared to existing state-of-the-art clustering protocols, showcasing significant improvements in energy efficiency, network longevity, and overall network performance metrics across different network densities and topologies. © 2026, University of Bahrain. All rights reserved.
Bayesian Classifier; CH selection; Energy Efficiency; Hierarchical Clustering; Internet of Things; Machine Learning; Wireless Sensor Network