Autonomous Traffic-Aware and QoS-Constrained Capacity Cell Shutdown for Green Mobile Networks

Energy efficiency of Radio Access Networks (RANs) is increasingly becoming a global strategic priority for Mobile Network Operators (MNOs) and governments to attain sustainable and uninterruptible network services. In this work, we propose an autonomous Machine Learning (ML)-based framework to maximize RAN energy efficiency via underutilized radio resource shutdown while maintaining an adequate network capacity with a preset Quality-Of-Service (QoS) level. This is achieved by dynamically switching radio resources on and off according to service demand. Training on a live network dataset and applying back the ML-advised parameters, the proposed framework is shown to save 10.3% of the overall RAN energy consumption while maintaining the imposed QoS level.