Energy Efficient Cooling in Networking Equipments: A Neural Network Based Predictive Approach

This study applies machine learning algorithms to predict optimal cooling and fan power consumption in networking equipment, specifically in systems with fixed power supplies and fans. We developed a DNN-based supervised cooling algorithm to improve energy efficiency while maintaining optimal system temperatures. ML model was trained with data containing ambient and component temperatures across system environmental sensors and fan speeds to establish a behaviour baseline for a system. The neural network model predicts optimal fan speeds based on the system load and environmental parameters. The proposed system achieved energy savings of approximately $40 \%$ through optimal fan speeds without compromising on optimal cooling. These findings demonstrate the potential of machine learning to improve energy efficiency and reliability in the cooling systems of network equipment, offering significant economic and environmental benefits.