AI-ENHANCED PHYSICAL LAYER FOR ENERGY-EFFICIENT WIRELESS SENSOR NETWORKS (WSNS)

Abstract

WSNs are key technology in smart cities, environment management and medical fields, but are restricted in energy by their battery requirements. To fix this problem, the paper recommends using an intelligent AI-powered transmission control framework to boost energy efficiency and increase network lifespan. The system uses both Machine Learning (ML) and Reinforcement Learning (RL) approaches. The Random Forest model accurately forecasts energy consumption at nodes (MAE = 0.00226) and real-time data from sensors and the environment allows the Q-learning-based RL system to adjust power use on the channel between 0.5W and 2.0W. Results are verified with information provided in the Intel Berkeley Research Lab WSN dataset. Simulated results reveal that the suggested AI framework helps save 33% more energy than running on a fixed schedule. Because the simulations adjust well to different environments, the approach is suitable for situations in real-world WSNs.