A technical look at 5G mobile device energy efficiency
5G enables new functions and major performance improvements, but it also puts tougher energy requirements on mobile devices. Below, we ask the question: Is it possible to drive down energy consumption in mobile devices and continue to improve the 5G user experience?
Reducing the level of energy consumption associated with cellular network operations is a strong focus area for Ericsson, and a key sustainability improvement goal. Our post on 5G energy consumption highlights advances in network energy efficiency, but the energy consumed by individual mobile devices also needs to be considered.
For example, many components contribute to the energy consumption of a modern smartphone, such as apps, operating systems, and the screen. However, cellular radio also contributes as a critical factor. With 5G, the range of device types is extensive, including built-in modems, smart wearables, and wireless sensors as just a few examples.
Improvements in 5G New Radio (NR), particularly the sparser transmission of always-on signals, have significantly reduced energy consumption on the network side. These changes also require updates to radio processing solutions in NR mobile devices. Since the very beginning of NR technology development, we have worked towards ensuring device energy efficiency via both standardized and proprietary features. Below, we will elaborate on the origins of device energy efficiency challenges and how NR releases and network configurations mitigate these concerns.
Device energy consumption in the first NR release
NR enables higher data rates and lower latency, which allows user data sessions to be terminated faster than in LTE. This inherently reduces the associated energy consumed by device per transmitted bit. However, since data arrival patterns are not deterministic, the device also monitors the physical downlink control channel (PDCCH) for possible data scheduling information during periods when data is not scheduled. To learn more about this, read our colleagues’ earlier article about the 5G NR physical layer.
With similar settings, NR and LTE device energy consumption for control channel monitoring in connected mode does not differ significantly. However, invoking new performance-enhancing features in NR (wider BWs, shorter slot times, multiple scheduling events per slot, etc.) can also increase the energy cost of control channel monitoring. Therefore, the first NR specification Rel-15 includes numerous tools for saving device power and energy, such as inactive state, connected-mode discontinuous reception (cDRX), and customized control channel monitoring:
- The device spends most of its time in an idle or inactive state, relieved from data monitoring and staying in a low-power mode state known as “deep sleep” most of the time. Only when data arrives, the device enters the connected mode for data transmission.
- Connected mode DRX (cDRX) is a key feature for device energy saving. cDRX provides two levels of monitoring granularity via the short and long DRX configurations. It allows the device to only monitor scheduling messages during well-defined monitoring intervals e.g. during 10ms on-durations once every 160ms in long DRX. The rest of the time the device can remain in sleep mode.
- Customizing control channel monitoring patterns allows the device to stay in an intermediate low-power state (micro-sleep) part of the time without any performance loss. Customized “search spaces” may be set to allow the device to skip monitoring certain slots. The device’s operating bandwidth, as well as other operating parameters, can also be adapted with low latency and signaling load via the bandwidth part (BWP) switching mechanism.
3GPP companies have developed a mutually-agreed device power model that captures the relative powers associated with different active operations, in relation to the lowest power, deep sleep mode. Clearly, reducing the fraction of time for the device to perform unnecessary PDCCH monitoring and enabling the device to be in a sleep state instead offers a high potential for energy conservation.
Figure 1 (below) presents the accumulated energy consumption profile of a typical eMBB device operating in a variety of states in a mix of traffic events over 24 hours. The left-hand bar graph shows, starting from the top, the fraction of energy consumed while performing control channel monitoring, data reception, periodic activities in connected and inactive modes, and deep sleep. The right-hand bar graph indicates the fraction of total time spent performing the respective operations.
Despite the relatively short time fraction spent in the connected mode, energy spent waiting for additional data arrival dominates. In this example, control channel monitoring without data transmission comprises over half of the total NR radio energy consumed. This is the cost of maintaining the required responsiveness to data arrival in the baseline Rel-15 framework. As illustrated in the figure, by avoiding unnecessary control channel monitoring, we can achieve savings in the device’s total consumed energy.
Network vendors can configure the cDRX and search space (SS) parameters to control the trade-off between the energy efficiency of the device and system performance KPIs such as throughput and latency. Ericsson has, in cooperation with device partners, identified enhanced system configurations and parameter settings to achieve an improved performance trade-off on both sides. While proprietary improvements can enable energy reductions compared to the baseline, additional mechanisms for energy savings beyond those in the Rel-15 NR specification can offer benefits.
Energy consumption in mmW deployments
An important advantage of NR compared to LTE is that it allows deployments over a wide frequency range – from below 1 GHz to over 50 GHz (millimeter-wave, mmW). Frequency ranges up to 7 GHz and above 7 GHz are referred to as frequency range 1 (FR1) and frequency range 2 (FR2) respectively. While FR1-only deployments are common in some regions such as Europe, FR2 bands are commonly used in early carrier aggregation and LTE-NR dual connectivity (EN-DC) deployments such as the US.
Using FR2 bands, the NR device can exchange data over a significantly wider carrier bandwidth and achieve very low scheduling latencies, realizing the promise of multi-Gbps data rates of 5G for eMBB devices.
There are power aspects that may pose challenges in FR2 compared to FR1:
- wider operating bandwidth and efficiency differences of radio frequency circuitry,
- shorter slot times that raise the required number of decoding operations per time unit,
- additional beam management measurements to maintain connectivity between narrow transmission and reception beams of the base station and the device.
Table 1 (below) captures some of those differences. For example, the instantaneous power associated with control channel monitoring in FR2 bands is 75 percent higher than in FR1. However, since the data transmission can be completed much faster, the total energy consumed for transmitting a data burst of a given size utilizing FR2 may become lower than in FR1-only setups.
A decisive aspect that affects the overall impact on energy consumption by adding FR2 carriers is therefore the network’s ability to dynamically activate the FR2 carrier only when it is needed for actual data transmission, and to instruct the device to monitor the FR2 control channel only when data may be present there. Opportunities to temporarily deactivate the carrier or pause the FR2 PDCCH monitoring are constrained by the time required to restart active operation once data resumes. Rel-15 provides baseline mechanisms for FR2 carrier activation/deactivation and controlling the monitoring patterns, but the dynamics are relatively slow, on the order of multiple tens of milliseconds.
To leverage the data performance gains available from mmW, the FR2 carriers in initial network configurations could remain activated over longer time periods when data transmission is expected. In such cases, complementing a traditional 20 MHz LTE carrier with one or more 100 MHz NR carriers at FR2 enables devices to operate with six to forty times wider bandwidth compared to LTE – and reach correspondingly higher data rates – but it can also increase device energy consumption.
Since it is important for user experience to preserve the battery lifetime, we have worked closely with device and chipset vendors to improve cDRX configuration settings and introduce novel uses of available Rel-15 mechanisms for FR2 deployments. This will lead to significantly reduced energy consumption for FR2-capable devices while securing data performance. Nevertheless, there remains room for enhancing the management of FR2 carriers, to be addressed in pending 3GPP releases.
|Power consumption (relative units)|
|Device power states and operations||FR1
(below 7 GHz)
|PDCCH monitoring only||100||175|
(depends on TX power level)
|250 - 700||350|
Table 1: Device power consumption in different operations and states (TR 38.840)
New solutions in Rel-16 for further reduced device energy consumption
Similar to previous mobile technology generations, once the basic NR functionality in the first release, Rel-15, was completed, development of additional features and optimizations has continued to further reduce device energy consumption in subsequent releases. Ericsson has been driving numerous improvements for device energy savings in Rel-16, which is planned to be finalized in the first half of 2020. Some important new features include:
- Improved cross-slot scheduling: The network can inform the device that a guaranteed minimum time interval of K0 slots exists between the downlink control channel PDCCH and the data packet it schedules. The device can thereby omit unnecessary radio frequency (RF) operation to buffer the data channel if no data is scheduled. It may also be able to use a more efficient receiver configuration for PDCCH reception. The signaling is dynamic so the reception mode may be adapted to match the instantaneous data traffic pattern. The resulting connected-mode energy reduction can be up to 15 to 20 percent, depending on the use case. Figure 2 illustrates device receiver processing for K0=2 slots.
- Dynamic Search Space (SS) adaptation: Although the BWP switching mechanism in Rel-15 functionally allows switching between dense and sparse patterns of control signal monitoring, the associated mode switching delays make the feature less attractive. Rel-16 introduces a new, low-overhead and low-latency signaling mechanism for SS switching to match instantaneous traffic arrival events. This can reduce the connected mode energy further by up to 10 to 15 percent for most device types without impacting their data performance.
- Connected-mode Wake-Up Signal (WUS): A Rel-15 device is expected to monitor all ON-durations in its cDRX pattern. In Rel-16, a wakeup signal can be transmitted to the device ahead of an ON-duration if the network intends to schedule the device in that ON-duration. Thus, if the device does not detect the WUS during the monitoring occasion (MO), it can skip the upcoming PDCCH monitoring. This can provide up to 10 percent additional connected mode energy savings for infrequently scheduled devices, depending on the cDRX settings.
- Secondary cell dormancy: When a secondary cell (e.g. an FR2 carrier in carrier aggregation) is activated, the device is traditionally expected to perform link quality measurements and reporting, as well as control channel monitoring on that carrier. The latter is the main source of FR2 energy consumption discussed earlier. Rel-16 introduces an additional “dormancy-like” operating mode where the device reports link quality but does not monitor the control channel on the secondary cell. When data is to be scheduled on the secondary cell, it can be rapidly brought out of dormancy via a downlink control message on the primary cell at a negligible energy cost for the device.
- Device assistance for secondary cell release: Rel-16 allows the device to indicate to the network that it does not expect further large data transmissions during the current connection whereby the secondary carrier can be immediately inactivated by the network until a future data session.
Towards an increasingly energy-efficient cellular ecosystem
Ericsson has already started planning for implementation of new device energy saving features to further enhance device energy efficiency with minimum impact on performance. We also continue to improve network configurations and introduce advanced mechanisms in upcoming product generations, using mechanisms available in the standard, based on both our own network operation analysis and information from device partners.
Advances will also happen on the device front as device architectures and implementations become more mature. The combined effect of advanced network mechanisms and improved chipsets is expected to further enhance NR device energy efficiency in FR1 and FR2 setups, as well as in FR1-only configurations. We foresee that this can be achieved without compromising traditional network performance KPIs such as throughput and latency.
However, the standard work to further improve device energy efficiency will only continue. In Rel-17, mechanisms to facilitate additional energy savings beyond the Rel-16 level are anticipated in both single-carrier (or FR1-only), and carrier aggregation (CA) and dual connectivity (DC) (FR1 and FR2) deployments. Besides the traditional eMBB traffic patterns, energy consumption improvements for reduced-capability devices and other use cases with infrequent data transmissions will also be considered.
We continue to be at the forefront of sustainability improvements in the cellular ecosystem. Reducing device energy consumption is an integral part of that work, simultaneously increasing user satisfaction, enabling new 5G use cases, and, by facilitating efficient device operation in all bands, encouraging continuous NR uptake across the spectrum.
Learn more about the energy efficiency of 5G networks.
Read about the focus areas of the latest 3GPP release 17.
Take a broader view of the climate impact of ICT devices in our digital carbon footprint report.