Simulation Based Comparative Analysis of Traditional and Hybrid GWO – ML Approaches for Energy Efficient Cluster Head Selection in Dynamic IoT Networks

Energy efficiency is a crucial issue in the Internet of Thing network due to the sensor nodes limited battery life span. The scalable and sustainable implementation of IoT solution depends on IoT energy efficiency. The changing patterns of energy usage in the IoT environment often exceed the capabilities of traditional metaheuristic algorithms. To address this, we propose a hybrid approach that integrates Grey Wolf Optimization with machine learning for energy efficient Cluster Head selection. The Random Forest-based machine learning model predicts the energy consumption patterns of IoT nodes. Simultaneously, as part of its fitness assessment, GWO uses these predictions to optimize CH placement. The proposed framework was implemented in Python and tested through simulations involving 100 IoT nodes over 50 iterations. Our analysis shows that the proposed model reduces overall energy consumption by approximately 12.44%, enhances load balancing, speeds up convergence, and extends network lifetime compared to traditional GWO. These results indicate that this framework can serve as a scalable solution for sustainable IoT network design and underscore the benefits of combining metaheuristic optimization with predictive intelligence.