NGMN, an Alliance of mobile operators, has produced a guideline paper aimed at 3GPP to ensure that any new Radio Access Technologies (RAT) proposed for 6G standards do indeed deliver cost, performance, customer experience and even revenue growth opportunities for mobile network operators.
Previously, NGMN has said that any 6G RAT “should be a graceful evolution of communications networks into the 2030s.” Aligned with that, it now wants to make double sure that any new 6G RAT should meaningfully surpass the capabilities of 5G Advanced networks as defined in 3GPP Release 18.
As researchers beaver away on new 6G radio technology candidates, NGMN is aiming to keep standards groups up to the mark with its expectations, the group has produced a Radio Performance Assessment Framework (RPAF) as a means of benchmarking new proposals against 5G Advanced capabilities.
The paper said, “This RPAF is primarily intended as a guideline to 3GPP so that their evaluation process ensures a new 6G RAT deployed provides tangible benefit over an equivalent 5G-Advanced evolution, especially considering new RAT deployment within already deployed bands.”
Not only that, but the paper also explicitly calls out that this benchmarking exercise will also show quite how capable 5G Advanced is, describing its potential as vast and yet to be exploited. In its words, “The ingredients of the benchmark should also be simultaneously viewed as a testament to the vast potential yet to be exploited by 5G/5G-Advanced.”
Back in February 2023 Vodafone’s Luke Ibbetson, who is a Member of the NGMN Alliance Board and Head of Group R&D, Vodafone, and also involved in the publication of the RPAF paper, said at the group’s MWC press conference,
“It’s very easy to take theoretical capabilities, make them all go faster and say that would be a good outcome. We are not doing that this time.
“We want to make sure that whatever 6G becomes – and it isn’t necessarily a new radio interface, it could be a bunch of capabilities built onto 5G – it has to mean something tangible to customers that allows us to deliver compelling new experiences, that get people spending more, allowing us to reinvest that in the services and fabric people reply on.”
And he added then, “We are only tapping into a tiny fraction of what 5G is capable of doing right now, partly because we are only in early stages of deployment, but also because 5G already specifies requirements that we don’t have enough spectrum to be able to deliver.”
The new paper, then, is a continuation of that view, with the added aspect that it gets into a little bit more detail about where 6G capabilties should start from.
What is NGMN looking for?
First, note that any improvements should be benchmarked against R18, the first 5G-Advance standard, not against previous iterations in R15-17, “to avoid artificially inflating 6G gains”.
The paper said that a new 6G RAT should demonstrate” significant benefits over and above a 5G-Advanced evolution” in the following key metrics.
- Spectral Efficiency
- User Experienced Data Rates
- Total Radiated Sensitivity
- Total Radiated Power (TRS/TRP)
- Total Energy Consumption
- Network Simplicity
But more than that, NGMN wants to do away with merely listing theoretical peak data rates and sub 1ms latency as metrics. It notes that using peak data rates as a comparison metric is outdated “since the best that 5G/5G-Advanced can offer already exceeds customer demand”, and does not reflect the need for network (or per-user) capacity and that the customer experience is dominated by “worst case”, e.g., cell edge, performance. Also, the paper argues, peak data rates can be artificially inflated without any improvements to spectral efficiency simply by assuming migration to higher frequency bands and greater corresponding channel bandwidths.
As for latency, it says a similar argument can be made for not using latency reduction beyond the sub-1ms order of magnitude to claim 6G gains – “since at this threshold, the server latency at the application layer dominates in shaping the user experience.”
Therefore the NGMN RFAP defines what its authors think are actually useful subsets of metrics. For example they would rather see measurements of average spectral efficiency and user experienced data rate, which would be more representative of actual network performance.
The AI-RAN and energy
When it comes to measuring energy efficiency the group argues that for each proposed radio feature enhancement, the baseline framework should include measurement of the difference in RAN energy consumption associated with that feature as a percentage change in total energy consumption relative to the total energy consumed in a 5G system without that feature.
Interestingly, the paper argues that this is particularly relevant when considering new AI features where the energy consumption is likely to be impacted by changes in processing requirements (including the energy used for model development and training).
The paper also points out that there should be “an assumption that the base station is already using AI/ML for many functions” – including positioning, CSI feedback, beam management. This use of AI would also already include neural networks to understand channel conditions and the existence of AI models to predict beam trajectory.
And the RPAF takes its AI-RAN benchmark not just from R18 but also R19. “Enhancements related to AI-RAN should also exceed the performance of AI features to be enabled in forthcoming Rel-19 specifications,” it states.
Baseline for 6G RAT
The NGMN paper outlines features that should be considered as 5G Advanced baseline benchmarks, which a 6G RAT would have to out-score.
For example, masive and multi-use MIMO should be compared to R18’s MU-MIMO baseline (e.g., 128Tx128Rx antenna configuration, up to 24 orthogonal DMRS ports), 5G-6G dynamic spectrum sharing should be compared to 5G systems that have no sharing in any bands, battery life and wake-up signalling should be compared to R19 proposals, and the radio performance baseline should include the most advanced CA features e.g. supporting SSB-less SCell for inter-band CA in FR1 bands.
Finally, the paper says that any increase in complexity – including the inherent complexity of adding a new 6G RAT – should be more than compensated for by performance improvements.
Raising the bar
So what’s this all about? Well, it’s about a group of operators raising the bar, or establishing the level at which they want 6G standards to play. They are directly stating that any new RAT should outperform the very latest standards, and also the upcoming advances on those standards, in a dynamic manner.
That, as we’ve said, sits with NGMN’s overall position on 6G, which is that they want any potential new “G” to show not just clear, tangible performance and cost benefits, but also benefits in terms of the business operations and revenue raising capability of the operator.
It also points out that 5G Advanced standards already provide a great deal of capabilities – some even exceeding user demand, and already integrate AI and ML technologies in the RAN.
Of course, on one level it might seem logical that operators would want 6G to go beyond 5G. But what we haven’t had before is such a clear statement of how operators would like to assess that, and in what areas. It would certainly seem to be part of NGMN’s existing position that 6G shouldn’t exist merely for the sake of having another G. They’re not interested in investing in that.
