As 5G networks continue their global expansion, a critical challenge looms large: growing power consumption generating inordinate operational expenses along with associated carbon output. Considering various projections, it is possible that by 2030, mobile networks could potentially end up consuming 5% of the world’s total electricity usage if current trends persist, with base stations responsible for approximately 80% of that consumption. This staggering figure underscores an urgent need for innovative solutions in next-generation wireless infrastructure.
Thanks to funding from the U.S. National Telecommunications and Information Administration (NTIA) through the Public Wireless Supply Chain Innovation Fund, researchers at the Open Networking Foundation/Aether Project and Rutgers University WINLAB, in collaboration with Keysight Technologies, are pioneering groundbreaking work in O-RAN (Open Radio Access Network) energy efficiency. Their latest research focuses on one of the most critical components of the network from the energy perspective: the O-RAN Radio Unit (O-RU). These units are particularly important because of their sheer numbers in deployed networks and their significant contribution to overall network power consumption.
The POET platform: A game-changer for energy testing
At the heart of this research is POET (Platform for O-RAN Energy Efficiency Testing). This sophisticated testbed represents a major advancement in how we measure and understand energy consumption in disaggregated wireless networks (see a previous RCR article for details on POET). This comprehensive setup enables researchers to conduct repeatable, precise measurements across a wide range of operational conditions, something that has been sorely lacking in the open literature.
Breaking new ground with multi-vendor testing
One of the most significant contributions of this research is its multi-vendor approach. The team evaluated four commercial O-RUs representing different deployment scenarios such as Low-power units (24 dBm) suitable for small cell deployments, Medium-power units (37 dBm) for typical urban scenarios, and High-power units (47 dBm) for macro cell applications. The units tested included both TDD and FDD configurations, with 4-antenna and 8-antenna MIMO setups, providing a comprehensive view of the O-RAN ecosystem.
Key findings: Where the power really goes
The research yielded several critical insights that could shape future energy-saving strategies as explained below:
The idle power issue: Perhaps the most striking finding is the dominance of idle power consumption, especially in lower-power scenarios. For example, one medium-power O-RU consumed about 59W in an active-idle state with 0% utilization. Even when generating just 1W of RF output, the idle power constituted approximately 91% of total energy consumption. This means that traditional approaches to energy saving, like reducing transmission power during low-traffic periods, may have limited impact and reveal a massive opportunity for energy savings through intelligent sleep mode strategies.
The power amplifier problem: Power amplifiers are responsible for boosting signals for transmission. At high RF power levels, power amplifiers dominate overall energy consumption, making their efficiency a critical parameter that requires thorough testing and characterization to assess potential energy saving opportunities. High-power RUs use power amplifiers, and it is observed that their power consumption varies nonlinearly with respect to RF power. But the most remarkable observation is that power amplifiers exhibit low efficiency at low power levels (often below 15%) and dramatically improved efficiency at higher power levels (up to 28% for high-power units) and showed significant efficiency variations among different hardware designs. This non-linear behavior explains why O-RUs consume substantial power even when transmitting very little as the amplifiers operate with the least efficiency at low loads. Understanding these types of power utilization behaviors is crucial for optimizing network operations.
Energy efficiency varies dramatically among RUs: Maximum energy efficiency, defined as the ratio of total transmitted RF power to consumed power, varied significantly across O-RU categories, ranging from just 1% for low-power units to nearly 20% for high-power units. This disparity highlights the importance of proper equipment selection and deployment strategies.
Traffic management strategy affects energy consumption: The research also revealed that traffic load management (which varies among RU designs) makes a significant difference in power consumption. For a particular RU, we observed the following:
- Frequency-domain loading (increasing/reducing active PRBs): Power scales proportionally with load
- Time-domain loading (increasing/reducing active transmission time slots): Power consumption remains high regardless of load level
This finding has profound implications for network operators trying to implement energy-saving strategies.
No need to worry too much about MIMO: The most significant finding regarding MIMO impact is that total RF power output is the dominant factor driving O-RU power consumption, rather than the number of antenna chains themselves. This means:
- A 4-antenna (4×4 MIMO) O-RU transmitting 40W total RF power consumes nearly the same power as a 1-antenna (1×1) O-RU transmitting 40W
- The power consumption curves for 1×1, 2×2, and 4×4 configurations are remarkably close when plotted against total RF power
- Although there is a measurable but relatively small incremental energy cost for each additional antenna chain, what matters most is how much total RF energy you’re transmitting, not how you’re distributing it across antennas.
In general, MIMO is energy efficient. The capacity gains far outweigh the power increase, making higher-order MIMO attractive from an energy-per-bit perspective.
A Validated power model for the future
Beyond measurements, the research team successfully parameterized and validated a component-based power consumption model. This model breaks down O-RU power into:
- Static baseline power from processors and power supplies
- Idle RF chain power for active but non-transmitting components
- Dynamic power that scales with transmission load and PA efficiency
This model provides network operators and researchers with a practical tool for predicting power consumption under various operational scenarios, which is essential for developing and evaluating energy-saving algorithms.
Looking ahead: Real-world deployment and AI optimization
The research team plans to extend their work to the NTIA-funded ORCID Test and Evaluation Lab, where they will perform measurements using commercial O-DUs supporting multiple multi-band O-RUs in scenarios that replicate field-deployed operational systems. In addition, future work will explore machine learning approaches for real-time optimization of energy consumption.
Why this matters
The open and disaggregated nature of radio access networks creates both opportunities and challenges for power management. The lack of standardized energy efficiency metrics across heterogeneous O-RUs has complicated power management efforts. This research fills a critical gap by providing:
- Detailed empirical data that has been missing from open literature
- A validated modeling framework for power consumption prediction
- Practical insights for developing energy-saving algorithms
- A methodology that other researchers and operators can build upon
The work represents a significant step toward sustainable 5G and NextG networks, demonstrating how independent research can drive innovation in critical infrastructure challenges. As wireless networks continue to evolve and expand, research like this, combining rigorous testing methodology, multi-vendor collaboration, and open science principles, will be essential for ensuring that our connected future is also a sustainable one.
The path forward: Implications for the Industry
These research findings point toward several critical areas for future mobile network radio development:
1. Hardware Design Evolution
Manufacturers need to focus on improving power amplifier efficiency across the entire operating range, not just at peak power levels.
2. Intelligent Power Management
Advanced Sleep Mode (ASM) technologies that can progressively shut down different components become even more critical given the dominance of idle power consumption.
3. Load Balancing Strategies
Network operators need sophisticated algorithms that consider both traffic patterns and power consumption characteristics when distributing load across O-RUs.
4. Standards Development
The industry needs standardized energy efficiency metrics and testing methodologies to enable meaningful comparisons between vendors and technologies.
Beyond the lab: Real-world implications
While these findings come from laboratory testing, their implications extend far beyond the research environment:
- Economic Impact: Energy costs represent a significant portion of network operating expenses. Understanding true power consumption patterns enables better cost modeling and optimization.
- Environmental Responsibility: As the telecommunications industry faces increasing pressure to reduce its carbon footprint, accurate power models become essential for meaningful sustainability initiatives.
- Network Planning: Deployment strategies, site selection, and infrastructure planning all benefit from precise understanding of O-RU power requirements.
- Innovation Catalyst: Detailed power consumption data enables the development of more sophisticated energy management algorithms and AI-driven optimization systems.
Looking ahead: The future of energy-efficient networks
The telecommunications industry stands at a crossroads. The exponential growth in data demand shows no signs of slowing, but the environmental and economic costs of traditional approaches to network scaling are becoming unsustainable.
This research represents a crucial first step toward data-driven energy optimization in O-RAN networks. By understanding exactly how O-RUs consume power, engineers and researchers can develop targeted solutions that maintain service quality while dramatically reducing energy consumption.
The future likely holds:
- Machine learning algorithms that dynamically optimize power consumption based on real-time traffic patterns
- Advanced hardware designs with dramatically improved amplifier efficiency
- Sophisticated sleep mode implementations that reduce idle power consumption
- Network architectures designed from the ground up with energy efficiency as a primary consideration
The bottom line
The transition to 5G/NextG represents more than just a technological upgrade. It is a fundamental reimagining of network architecture. As this research demonstrates, understanding and optimizing energy consumption is a business imperative that will determine the long-term viability of next-generation networks.
The detailed power consumption data and models presented in this research provide the telecommunications industry with the tools needed to make informed decisions about energy efficiency. As we continue to push the boundaries of what’s possible with wireless communications, studies like this ensure we’re doing so in a way that’s both economically viable and environmentally responsible.
