DEEP LEARNING-DRIVEN ADAPTIVE CLUSTERING FRAMEWORK FOR ENERGY-EFFICIENT AND HIGH-PERFORMANCE INTERNET OF THINGS NETWORKS

Abstract

The rapid proliferation of Internet of Things (IoT) devices has intensified the need for intelligent network management strategies capable of addressing challenges related to limited energy resources, constrained processing capabilities, and restricted communication bandwidth. Clustering has emerged as an effective approach for enhancing network scalability, reducing communication overhead, and improving resource utilization in IoT environments. This study presents a comprehensive evaluation of conventional and deep learning-based clustering techniques for optimizing IoT network performance. Traditional clustering algorithms, including K-Means, Density-Based Spatial Clustering of Applications with Noise (DBSCAN), and Fuzzy C-Means, are compared against advanced deep learning approaches such as Variational Autoencoder (VAE)-Based Clustering and Recurrent Neural Network (RNN)-Based Clustering. Simulations are conducted using the Contiki operating system to emulate realistic IoT network conditions characterized by heterogeneous and resource-constrained devices. Performance evaluation is carried out using key metrics including latency, throughput, energy consumption, cluster stability, and communication efficiency. Experimental results reveal that deep learningdriven clustering techniques significantly enhance throughput and energy efficiency by effectively capturing complex data patterns and dynamic network behavior. However, these improvements are accompanied by increased computational complexity and processing overhead compared with conventional clustering methods. The findings highlight the trade-offs between performance gains and resource requirements, providing valuable insights for selecting suitable clustering strategies across diverse IoT deployment scenarios. The proposed framework contributes to the development of intelligent, scalable, and energy-aware IoT networks capable of supporting next-generation smart environments and connected systems.