With the introduction of 5G systems and the deployment of new 5G Radio Access equipment, it is expected that 5G RAN energy consumption will also experience a significant increase over the next years. This increase will impact not only the energy levels but also the overall costs of operating a 5G site.
RAN Energy Consumption is an essential component of OPEX costs for any Mobile Operator. 5G MNOs and vendors are currently analyzing their strategies regarding potential 5G RAN energy-saving techniques to help reduce energy consumption. This analysis takes place both from a technical and a cost perspective in order to find the optimum solution per Operator.
5G RAN power consumption factors
The overall cost is affected by system technology. Different hardware elements such as the Radio Unit (RU), the Baseband Unit (BBU), and the transport unit, may add to the overall power consumption. In the case of a 5G system with an active antenna, the antenna unit will also consume power. The rest of the base station infrastructure also plays an important role. Elements such as the power supply unit (PSU) and the cooling or heating systems also contribute to the power consumption levels.
5G RAN energy efficiency
One method to reduce 5G RAN energy consumption is equipment re-engineering. Depending on the hardware lifetime, the Operator may use new low-energy consumption equipment to swap existing hardware.
Another approach is network planning. The 5G Operator may optimize the physical placement of equipment or change the geographical position of the site. The latter can be applied to a lesser extent since it is usually challenging to find new places to host new base stations.
Other various energy-saving techniques can also be used. One of them is the dynamic adaptation of network resources to traffic load variations. Scheduling policies can be used to adapt the bandwidth, combined with adaptive power amplifiers, BBUs, and antennas. Other aspects include Radio Resource Management (RRM), Artificial Intelligence (AI), and Machine Intelligence. Sleep mode is another power-saving technique where active hardware can be turned on and off according to network resource needs. These techniques will be described further in the following paragraphs.
5G RAN Energy-saving techniques
RF Unit efficiency improvement
RF unit efficiency improvement is targeting innovations on both HW and SW level. Telecom vendors are working on continuous development and new product generation introduction processes. Hardware improvements include Power Amplifier (PA) efficiency, RF and IF chip-level efficiency. New HW generation is introduced approximately every 5-8 years. RAN vendors’ goal is to improve the total energy efficiency of each new HW generation by approximately 5-10%.
One Radio for three sectors
Usually, a standard 3-sector site requires three radios. In this example, one physical equipment is configured to serve all three sectors. However, as the output power of one RF unit is shared within three sectors, there is an expected limitation in the downlink. This solution is, therefore, suitable for low capacity sites. The gNodeB can then be upgraded to the standard configuration when traffic exceeds certain predefined thresholds.
Symbol and enhanced symbol power saving
Power Amplifiers require static power consumption even when no signal is transmitted. Symbol power-saving feature quickly shuts down PAs of symbols that do not contain any data. This selective powering off of certain elements reduces static overall power consumption. In time for data to be resent, the PA is switched on again.
In Enhanced Symbol power saving, shaping of downlink Physical Resource Blocks (PRBs) to same Transmission time intervals (TTIs) occurs. Therefore, more TTIs are available for Symbol power saving. The advantage of this technique is the reduction of network interference. In theory, there should be a small impact on latency since the maximum delay is the length of one TTI.
Dynamic voltage adjustment
Adaptive power adjustment adjusts the PA working voltage based on the traffic load, therefore, increasing PA conversion efficiency. This technique is used to reduce PA power consumption and improve overall e/gNodeB energy efficiency.
M-MIMO array-related sleep mode
For low traffic scenarios during night time such as stadiums and shopping malls, some channels of the RF module can be shut down. These channels can be turned on dynamically based on traffic load and configured thresholds. When the channels are turned off, the transmitting beam waveform will widen. This will, in turn, increase the interference of neighboring cells, and affect the performance of the edge users. The system may lose some dBs of coverage if just M-MIMO channels are turned down from 64TRX to 32TRX or from 32TRX to 16TRX. The coverage will remain unchanged only after compensating by increasing the transmit power.
Capacity layer shutdown
Capacity layer shutdown feature transfers User Equipment (UE) from a lightly loaded capacity layer to a basic layer of other cells. Then the capacity layer is shut down to save energy. When the traffic load meets the power-off condition threshold, UEs from the capacity layer are handed over to the basic layer. After the capacity layer shutdown, the gNB monitors the traffic load on the basic layer. A shutdown layer can be reactivated when traffic exceeds the defined threshold.
M-MIMO RF deep sleep mode
In the case of low traffic, when an Active Antenna Unit (AAU) is in sleep mode, the PA is turned off, and power reduction is estimated at around 15-20%. With RF deep sleep mode, almost all components in the AAU can be shut down, and power reduction would be even up to 75%. This functionality of the entire RAT shutdown could be again applicable in individual cases with specific capacity needs such as stadiums, concert venues, etc.
MIMO sleep mode (8×8)
In the case of 8×8 MIMO, the radio can be reconfigured to 4×4 or 2×2 MIMO mode during low traffic to save energy. After some transmit channels are off, the gNB monitors the load, and shutdown antennas can be activated when traffic exceeds the defined threshold. Impact on coverage can again be compensated by increased RF power which reduces possible energy saving. Throughput degradation is expected by up to 75% while energy-saving only around 12-15%.
Multi-RAT shutdown
In the case of different Radio Access Technology (RAT) networks, if the cells of all RATs have low traffic load, the Multi-RAT Shutdown feature may take over. This feature shuts down the cells of one or more RATs based on the overall traffic changes. The coverage and service are provided by the cells of other RATs. For example, RAT 2 and RAT 3 cells are shut down when the traffic volumes on all RATs are low. The system restarts RATs 2 and 3 when traffic volume on RAT 1 increases.
Artificial Intelligence
AI increases energy saving based on traffic prediction. AI is using traffic monitoring and machine learning to automatically adjust the time when PAs can be turned off during carrier shutdown and MIMO sleep. The setting of a possible turn on and off is automatically configured by AI and updated daily to reflect changes in traffic.
BBU Sleep mode
In the case of capacity layer shutdown or M-MIMO sleep mode during a low-traffic period, part of the RF equipment is inactive. With BBU sleep mode, the associated digital processing components in the BBU are put into sleep mode as well. Estimated energy saving with this technique is about 40–50W.
5G RAN energy-saving summary
Investing in power-saving technologies will result in reducing 5G site operating costs for Mobile Operators. However, the Operator should not jeopardize the overall 5G user experience. All the above techniques should, therefore, still guarantee the 5G end-user with the required throughput and coverage. The capacity during night hours could decrease gradually, and active elements turned off only when system performance is ensured. At the same time, fast wake-up techniques should guarantee that the customers will still maintain their user experience levels and SLAs in case of required traffic bursts.
