Distributed Machine Learning Cuts Energy Use by 70% in 6G IoT Networks

A new study analyzes energy consumption in AI-powered 6G Internet of Things networks, comparing centralized and decentralized architectures. Researchers deployed a testbed within Germany's railway infrastructure, using sensor data for machine learning-based predictive maintenance. The comparative analysis found that distributed learning models achieve predictive accuracy around 90% while reducing total electricity consumption by up to 70% compared to centralized approaches. This research addresses critical energy efficiency challenges as sixth-generation technologies create new demands for machine learning applications in IoT systems. The work highlights how model training and data transmission significantly impact energy usage in these networks. Optimization of these processes has become essential for sustainable system design in emerging 6G environments. The findings demonstrate distributed machine learning's potential to improve energy efficiency in IoT networks supporting predictive maintenance applications.