QoE-aware network energy savings in 5G

In this work, we study the problem of reducing the network energy consumption in a 5G deployment while avoiding to negatively impact the experience of the end users. Quality-of-experience (QoE) metrics from the mobile users are not presently available in the base stations of a real 5G system, and thus it is necessary to estimate them. We focus on QoE scores related to video streaming, and we employ these scores to optimize a machine learning (ML) algorithm for network energy savings. The performance of our approach is tested using a comprehensive 5G NR system-level simulator.