《绿色未来网络—网络能源效率白皮书》-56页_1mb

Summary of Green Future Networks - Network Energy Efficiency

Core Content

This document, titled Green Future Networks - Network Energy Efficiency, is a final deliverable by the NGMN Alliance, approved on 4th November 2021. It outlines the strategies and technologies to improve energy efficiency in mobile networks, especially in the context of 5G and beyond, with a focus on reducing energy consumption and carbon emissions.

Main Objectives

• To reduce energy consumption in mobile networks.
• To enhance energy efficiency on a per-bit basis.
• To support the sustainability goals of the mobile industry, including carbon neutrality.
• To explore the use of AI for intelligent energy management.

Key Points and Main Ideas

1. Energy Consumption Overview

• Mobile network energy consumption is a major operational expense (OPEX) and a significant contributor to CO₂ emissions.
• Base stations account for ~57% of total energy consumption in a typical cellular network.
• Energy efficiency improvements are essential to meet the rising demand for connectivity and reduce environmental impact.

2. Energy Efficiency in 5G

• 5G is more energy efficient than previous generations, thanks to:
  • Massive MIMO technology, which increases spectral efficiency.
  • Improved sleep modes based on traffic and load conditions.
  • Lean carrier design and advanced interference mitigation.
• However, 5G’s higher frequency deployment and network densification may lead to higher power usage compared to 4G.

3. Strategies to Decrease Energy Consumption

The document outlines multiple strategies for reducing energy consumption across three levels: equipment level, site level, and network level.

Equipment Level

• Base Station Hardware:
  • Power amplifiers are the largest power consumers, with ~59% of total consumption under full load.
  • Massive MIMO significantly increases throughput but also requires more digital processing.
  • Power-efficient hardware such as ASSPs, ASICs, and low-power FPGAs are recommended.
  • Virtualization of RAN allows for the use of COTS hardware (e.g., GPPs, NICs) to improve efficiency and reduce dependency on proprietary solutions.

Site Level

• Cooling Solutions:
  • Free cooling and liquid cooling are highlighted as important for reducing energy consumption in site infrastructure.
  • These methods can reduce the cooling load, which currently accounts for ~50% of network energy use in some cases.
• Renewable Energy:
  • Solar and other renewable sources are being adopted for both on-grid and off-grid sites.
• Smart Batteries:
  • New battery technologies are being integrated into 5G sites to improve energy management.

Network Level

• Flexible Cooperation between 5G and LTE:
  • Enables dynamic capacity planning and energy optimization.
• AI-Driven Energy Management:
  • AI can predict and optimize traffic patterns, enabling intelligent activation and deactivation of sleep modes and site functions.
  • AI is still in its early stages but has the potential to significantly improve energy performance.

4. Software and Sleep Modes

• Sleep Mode Functions:
  • Include symbol shutdown, channel shutdown, and carrier shutdown.
  • Symbol shutdown is the most effective, saving ~10% energy in less loaded scenarios.
  • Carrier shutdown can be used but may impact user experience.
• Sparse Antenna Arrays:
  • Can reduce power consumption by turning off unused antennas.
• Network Design:
  • Optimizing network layout and resource allocation helps reduce energy usage.

5. Impact of Terminals on Network Efficiency

• The performance of user devices impacts the network's energy efficiency.
• Higher receiver sensitivity leads to better throughput and less retransmission, reducing base station power usage.

6. Challenges and Future Directions

• AI Potential:
  • AI can enable dynamic and intelligent energy management without affecting QoE.
  • The document provides a methodology for calculating AI energy consumption, based on a canonical CNN configuration.
• Need for Collaboration:
  • Achieving sustainability goals requires end-to-end collaboration among MNOs, partners, and the entire ecosystem.
• Standards and Maturity:
  • Standardization and maturity of AI-based solutions are still under development.

Conclusion and Recommendations

• To achieve carbon neutrality, mobile operators must adopt integrated energy-saving technologies.
• Virtualization, AI, and advanced sleep modes are key enablers of energy efficiency.
• End-to-end optimization is crucial, including hardware, software, and site-level improvements.
• Renewable energy sources and efficient cooling should be prioritized to reduce the carbon footprint.
• Continued research and development are needed to fully realize the potential of AI in energy efficiency.

Key Figures and Statistics

• 57% of total energy consumption in a typical cellular network is attributed to base stations.
• 100 billion connections are expected by 2025, including 40 billion smart devices.
• Massive MIMO improves spectral efficiency by 3–5 times compared to traditional 4G.
• Symbol shutdown can save ~10% energy.
• COTS servers can achieve 1.42X performance gain and 15% energy efficiency improvement for 5G UPF.

Summary of Energy Efficiency Metrics

• Power Consumption: Measured in Watts (W), calculated as Voltage × Current.
• Energy Consumption: Measured in Watt Seconds (Ws) or Joules (J), often expressed as Kilowatt Hours (kWh).
• Power Efficiency: (Output Power / Input Power) × 100%.
• Energy Performance: (Service Delivered / Power Consumed), often expressed as Mbits/kWh.

References

• NGMN "Sustainability Challenges and Initiatives in Mobile Networks" White Paper [1].
• Analysis Mason [3].
• Three European universities [4].
• US Environmental Protection Agency and Energy Star [5].
• COTS server performance data [6], [7].

This White Paper serves as a comprehensive guide for MNOs to adopt best practices and technologies that support the transition to more energy-efficient and sustainable mobile networks.