Measurement-Based Energy-Efficiency Optimization for On-Demand Data Rate Provisioning in 5G RAN

Abstract

As network traffic grows, reducing energy costs and $\mathbf{C O}_{2}$ emissions is essential for sustainable telecom operations. Energy efficiency (EE) is therefore a key design objective in 5G Radio Access Networks (RAN), where computing resources in the Central Unit (CU) and Distributed Unit (DU) must be managed while meeting Quality of Service (QoS) requirements. This work focuses on the CU/DU computing platform, a major contributor to RAN power consumption due to continuous baseband processing. We present a measurement-based framework that dynamically adjusts CPU core count and clock frequency to minimize the energy consumption while meeting the data rate requirements of User Equipment (UE). The proposed algorithm selects the most energy-efficient CPU configuration based on experimentally derived power-performance profiles from a practical 5G RAN testbed. Experimental evaluation on two RAN platforms shows power savings of up to 16% and 35%, respectively, compared to a baseline configuration with all CPU cores active at the maximum base clock frequency. The results demonstrate the effectiveness of the proposed framework for energy-efficient operation in practical 5G RAN deployments.