Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Charting 5G Energy Efficiency: Flexible Energy Modeling for Sustainable Networks

Anderson L de Araujo, Luc Deneire, Guillaume Urvoy-Keller, André L F de Almeida

TL;DR

This work addresses the challenge of estimating end-to-end energy consumption in 5G networks by introducing an operation-count-based framework that spans both mobile terminals and the base-station PHY. The method discretizes the signal-processing chain into eight blocks (A–H), maps each block's processing to CPU-cycle counts, and translates those cycles into energy using the per-cycle energy $\epsilon = \kappa f_{pc}$. Validation against a MATLAB-based Intel target shows overall agreement, with specific blocks exhibiting over- or under-estimation due to algorithmic and architectural differences, highlighting opportunities for optimization. The proposed initial flexible, fine-grained framework enables cross-context energy comparisons and serves as a foundation for optimizing 5G network energy efficiency across diverse technologies and deployments.

Abstract

Despite the rapid advancements in 5G technology, accurately assessing the energy consumption of its Radio Access Networks (RANs) remains a challenge due to the diverse range of applicable technologies and implementation solutions. Designing a versatile power model for estimating the 5G RANspecific power consumption requires extensive data collection and experimental studies to capture the diverse range of technologies and implementation solutions. The objective is to outline a versatile energy model capable of estimating RAN-specific energy consumption, encompassing both mobile terminals and the physical layer (PHY) of base stations. In this paper, we focus on the computational complexity of the baseband part of the model. The developed (part of the) model is compared with the estimation of the number of cycles (and energy per cycle) used by a specific implementation (here a Matlab code ported on an Intel target), enabling the assessment of the model with the estimation of energy consumed on a real target. The study's results show a good agreement between the model and the implementation, even if some parts need to be refined to take specific algorithms into account. The key contribution is the development of an initial flexible energy model with finer granularity, enabling comparisons of energy use across various applications and contexts, and offering a comprehensive tool for optimizing 5G network energy consumption.

Charting 5G Energy Efficiency: Flexible Energy Modeling for Sustainable Networks

TL;DR

This work addresses the challenge of estimating end-to-end energy consumption in 5G networks by introducing an operation-count-based framework that spans both mobile terminals and the base-station PHY. The method discretizes the signal-processing chain into eight blocks (A–H), maps each block's processing to CPU-cycle counts, and translates those cycles into energy using the per-cycle energy . Validation against a MATLAB-based Intel target shows overall agreement, with specific blocks exhibiting over- or under-estimation due to algorithmic and architectural differences, highlighting opportunities for optimization. The proposed initial flexible, fine-grained framework enables cross-context energy comparisons and serves as a foundation for optimizing 5G network energy efficiency across diverse technologies and deployments.

Abstract

Despite the rapid advancements in 5G technology, accurately assessing the energy consumption of its Radio Access Networks (RANs) remains a challenge due to the diverse range of applicable technologies and implementation solutions. Designing a versatile power model for estimating the 5G RANspecific power consumption requires extensive data collection and experimental studies to capture the diverse range of technologies and implementation solutions. The objective is to outline a versatile energy model capable of estimating RAN-specific energy consumption, encompassing both mobile terminals and the physical layer (PHY) of base stations. In this paper, we focus on the computational complexity of the baseband part of the model. The developed (part of the) model is compared with the estimation of the number of cycles (and energy per cycle) used by a specific implementation (here a Matlab code ported on an Intel target), enabling the assessment of the model with the estimation of energy consumed on a real target. The study's results show a good agreement between the model and the implementation, even if some parts need to be refined to take specific algorithms into account. The key contribution is the development of an initial flexible energy model with finer granularity, enabling comparisons of energy use across various applications and contexts, and offering a comprehensive tool for optimizing 5G network energy consumption.

Paper Structure

This paper contains 12 sections, 18 equations, 3 figures.

Table of Contents

  1. Introduction
  2. Related Works
  3. Method
  4. Block A
  5. Block B
  6. Block C
  7. Blocks D and E
  8. Block F
  9. Block G
  10. Block H
  11. Results and Discussion
  12. Conclusion

Figures (3)

  • Figure 1: Number of cycles per block and the total number for a single slot transmission on downlink.
  • Figure 2: Number of cycles per bit estimated by the model for the blocks and total number of cycles of modulations QPSK, 16QAM, and 64QAM for a single slot transmission on downlink.
  • Figure 3: Number of cycles per bit estimated by the measurements for the blocks and total number of cycles of modulations QPSK, 16QAM, and 64QAM for a single slot transmission on downlink.