Your browser is not supported. Please update it.

24 May 2022

Edge-based smart traffic management can help 5G meet energy targets

Energy consumption remains a controversial topic for 5G, with agreement only over its greater efficiency per bit of data transmitted than 4G. The two bones of contention are firstly whether this will translate into overall energy savings when continuing increases in data traffic are taken into account, and secondly even if this not the case whether the ability of 5G to elicit savings in other sectors such as smart grids will offset any net increase.

There is certainly no doubt that the scale and capacity of wireless communications enabled by 5G will help deliver energy savings in diverse industry sectors. But there is a quite reasonable assumption in the light of ambitions to achieve net zero that 5G should itself deliver energy savings, irrespective of benefits achieved elsewhere.

Despite industry claims it is by no means clear operators can guarantee these savings on their own with the current architectures. A paper published in March 2022 by the Centre for Research into Energy Demand Solutions (CREDS) argued that it was yet to be determined whether 5G will cut overall energy consumption, or indeed cause it to rise.

The paper identified shortcomings over evidence presented so far, which it argued was not based on firm foundations or peer review, with a failure properly to disclose key assumptions and data used in the analyses.

This belief in the need for 5G to deliver energy savings on its own in the light of inconsistent evidence is driving ongoing efforts to achieve further savings beyond those brought by 3GPP standards themselves. We have in the past touched on some of these developments, such as Huawei’s outdoor cabinet, the APM5950, launched late in 2021 to allow equipment rooms to be deployed without air conditioners, or with reduced need for them. This claims an average 30% saving in electricity consumption per site, while also being able to capture solar power and use high-cycle lithium batteries to replace diesel generators, reducing carbon emissions by an even higher factor.

There are various other initiatives that fall outside the ambit of 3GPP standards, one rather along the lines of the Huawei effort, to reduce power consumed in cooling data centers within core networks. Nokia’s Bell Labs is working on this as part of a $2m project for the US Department of Energy (DoE), seeking to develop more thermally efficient cooling methods that reduce power, while also capturing waste heat that can then be used in principle to yield further savings.

There are also efforts more directly related to the mobile traffic itself, essentially aiming to save money by cutting overall bandwidth consumed. The 5G standards themselves enable power savings by shutting down components more frequently when idle for shorter time periods than was possible under 4G. 5G NR goes further through use of OFDM multiplexing, which combines multiple sub-channels in a single channel, primarily to increase resilience against interference. But the ability to add and subtract subcarriers on the fly to increase or decrease a channel’s capacity as required also saves power by avoiding need to maintain higher bandwidth than is needed.

But while 3GPP standards are designed to exploit variations in data traffic they do not address further energy savings that could be achieved by reorganizing traffic on a large scale. This is where edge compute comes in, motivated partly to reduce latency for demanding applications but also to reduce core bandwidth consumed, delivering energy savings in turn.

Again though, as with 5G itself, it is not immediately clear if multi-access edge compute (MEC) would save or increase power consumption. After all, replicating centralized computational resources around the edge could well increase the total amount of hardware required for a given service or network. On the other hand, by saving overall bandwidth since a lot of data no longer has to traverse core networks or content delivery networks (CDNs), the need for associated processing is reduced.

There is now mounting evidence that edge computing does cut net energy consumption in networks, by reducing the total amount of data traversing the network. There also two other factors, one being that edge compute systems tend to be smaller and less likely to require sophisticated cooling systems, consuming less electricity per unit of computation. This so-called ‘free cooling’ is even more applicable in cooler climates, again just like Huawei’s outdoor equipment rooms.

Secondly, given their smaller scale, edge compute systems are likely to experience greater fluctuations in demand, which brings more scope for power savings by putting resources to sleep when dormant. However, exploitation of such capabilities requires investment in management of distributed edge data centers, which is ongoing work. In the end this could help 5G meet more ambitious overall targets for energy savings.