The first rule of 6G club is… we talk only in cautious and customer-centric terms about 6G club.
Here’s Mike Irizarry, Executive Vice President & CTO, UScellular, talking at a press conference held by the NGMN (Next Generation Mobile Networks alliance) at MWC.
“We don’t know what it is, we don’t know how much it is going to cost, and we don’t know what problems it can solve.”
But, says Irizarry, we do know usage overall is increasing and will continue to do so – and that if apps like telepresence, assisted and autonomous driving and immersive experiences do emerge, we know we need to be able to support them.
The thing is, as Irizarry says, “we are not sure yet if 5G can support them.”
“We are not saying it can’t, but we want to be ready to do that and technology transitions can take 7-10 years. So we need to start thinking now about what industries, verticals and consumers want to see, and if 5G and 5GA need enhancements we are ready for that. And maybe that is what 6G ends up being – some enhancements – we just don’t know yet.”
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
Also at NMGN’s press conference was Luke Ibbetson, Head of Group R&D, Vodafone. He is part of the team at NGMN that has produced a paper that outlines design considerations for 6G.
The first key message of that, he says, is that “everything we want to do has to have a very tangible customer outcome, has to be customer-centric.”
The group tried to think in terms of customer outcomes and then design around that, not to assess what is possible technically, and then retrofit some use cases.
“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.”
For Ibbetson, although 6G picks up where 5G leaves off, or enhances 5G, he is keen to point out, “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.”
So NGMN’s focus, Ibbeston says, is to be realistic about 6G’s technology scope, and to think in terms of solving some long standing customer problems.
That means design considerations will reflect tradeoffs “in a judicious and informed manner” between energy consumption and bit rates, bit rates versus area coverage, in-building versus wide area.
The team is also looking to make sure that it doesn’t fall into the trap of previous generations by concentrating on performance led metrics versus what is “useful”.
“We will bring fresh thinking to address the perennial frustrations of customers – like in-building, reliability – so we are not just bringing technologies forward that makes things go faster in ideal conditions, but that have more tangible benefits.”
It all sounds a bit like the operators at NGMN are trying to tap the brake pedal on 6G, to take stock and work out exactly what long standing customer problems they want to solve. It stands against the sense of inevitability that because we have a new G every ten years or so, we must have 6G too.
Ibbetson: “We want to make sure that we shape technology ahead of time to make sure we are able to deliver on that. But we can’t put an arbitrary date on that nor can we say we need to invest in something simply because we have always invested in a new generation of technology in a certain time. We are breaking that cycle and looking at it more from, “what do we need to build for customers?”
Irizarry: “I would say this can’t be tech led. It must be led by consumers and verticals, and NGMN is making sure we are moving at the right pace. If it looks like we have to accelerate then we will, and if it doesn’t then we won’t. It’s got to be led by the consumer industry and the business verticals.”
Is the rest of the market listening?
Rajesh Pankaj is InterDigital’s EVP, Chief Technology Officer. He is responsible for leading the company’s technology roadmaps. 6G is very much on that roadmap. So what does he think 6G is about?
“What we are thinking about is things that 5G already but maybe there are a few faults that we can do better. And then also there’s new things that we can do.”
A RIS mode (above) and RIS panel (below) on the KT booth at MWC23.
An example of the first is increasing overall power efficiency of mmWave systems. Pankaj sees Reconfigurable Intelligent Surfaces (RIS) as a means of increasing PA efficiency in systems that are operating in higher bands. RIS are panels that reflect signal, but they are controllable rather acting as static repeaters, so that they can be altered as user demand alters.
Interdigital doesn’t build these panels, but it is looking at how to control them to get the overall system to work better.
“Ultimately it comes down to finding a way for the mobile to indicate presence – and to control the RIS through the network via the base station. We are thinking about what signals need to be sent back and forth in order to make sure that the base station has enough information, and then has what it needs to send to the RIS so that it can focus in the right direction.”
Another area that Interdigital is looking at that might fall within the “enhancing 5G” bit of 6G, is looking at distributed and edge computing. Pankaj sees a world where low latency services will require compute mobility to be controlled between far edge locations – essentially handing off compute as a user moves between edge areas.
“The role we play is very much analgous to how mobility is handled from a mobile network point of view. It’s a similar idea to making sure connectivity remains strong, but for compute – the same sort of algorithm but with different metrics.
Edge compute that follows users around may seem like an extreme vision, especially as on-device compute capability is also likely to increase over time, but Pankaj disagrees.
“People think there is competition between compute and bandwidth – but the reality is when we give people more compute power and more bandwidth they are able to find applications to take advantage of that. I don’t see that trend ending in the foreseeable future. No matter now good inference is on the phone, it is never as good as when you have more compute power available, so not being able to take advantage of that will always be a disadvantage. There are all sorts of XR and AI compute available if you still want it on the phone, but it could be be much better.”
Sensing
Like others, Pankaj is also looking at sensing applications. This essentially uses the high frequency signals like a sonar or radar, bouncing back very accurate signals to build a cm-level positioning model of a person or object.
“A decade ago I thought this looked like sci-fi, but since then signal and processing capabilities, combined with AI, gives the possibility to use RF signal that is also being used for communications to position and sense.”
(Warning, forklift approaching: sensing demo on Nokia booth at MWC23)
Other companies too are looking at sensing. At MWC Nokia had a demonstration on its booth of using RF for sensing. A user walked down a walkway and triggered musical notes on a scale, depending on their location. There was also a simulation where a pedestrian was sensed and located as being in the path of a moving autonomous vehicle, and delivered a warning. For this live demo, an antenna was transmitting over a 200 MHz wide bandwidth at 28 GHz, with the signal reflected back off the body in the runway.
But developing combined sensing and comms capability is in the early stage. Test company Rohde & Schwarz has developed a testing set-up to enabled companies to develop combined communications and sensing applications – in other words combining the ability to transmit data as well as do environmental sensing. It devised a sensing capability test by modelling various objects in a new emulator – creating reflective profiles of the objects. It then used a signal generator and a signal analyser to create and visualise reflections from the objects – with those reflections checked against known object profiles. That means that developers can test different waveforms and see how well they work in sensing applications. This is already done in other frequency bands in automotive use cases, but this capability is for joint comms and sensing in a range of different frequency bands.
The AI ai
Nokia and Rohde & Schwarz were also amongst companies working on an intelligent air interface.
Working with Nvidia, Rohde & Schwarz is developing a neural receiver demonstrator, replacing conventional signal processing blocks with neural models (trained ML models).
The use of AI/ML for signal processing tasks, such as channel estimation, channel equalisation, and demapping might create an ai-native air interface in 6G. R&S said that its simulations suggest that a neural receiver will increase link-quality and throughput compared to the current high-performance deterministic software algorithms used in 5G NR.
In the absence of live data models for ML training, test and measurement equipment generates various data sets with different signal configurations to train machine learning models for signal processing tasks.
In the setup at the Rohde & Schwarz booth, the signal generator emulates two individual users transmitting an 80 MHz wide signal in the uplink direction with a MIMO 2×2 signal configuration. Noise is applied to simulate realistic radio channel conditions (in the demo we saw a bottle of water was placed between the two antennas in the picture above). A satellite receiver acts as the receiver, capturing the signal transmitted and then providing that as data via a real-time streaming interface to a server. There, the signal is pre-processed, with this data set serving as input for a neural receiver implemented using NVIDIA Sionna, a GPU-accelerated open-source library for link-level simulation. Sionna enables prototyping of complex communications system architectures, supporting the integration of machine learning in 6G signal processing.
As part of the demonstration, the trained neural receiver is compared to the classical concept of a linear minimum mean squared error (LMMSE) receiver architecture, which applies traditional signal processing techniques based on deterministically developed software algorithms (see image above).
On Nokia’s booth, it was showing its demo of an AI-native AI (air interface) – using AI so that the transmit and receive functions learn together to create more efficient data transfers.
Dani Korpi, Senior Specialist, Nokia Bell Labs, said the Proof of Concept is its first implementation an ai native air interface, using AI solutions to learn how to convey information. Allowing the network to learn what to do is a “fundamentally different way of implementing the network”, he said. Using a spectrum analyser it can see the performance, signal to noise ratio and BER of the AI AI, compared to a 5G interface. Nokia is finding that throughput may improve by 20-35%, even up to 40% in certain use cases, and sees almost no errors.
That might mean that an AI ai in a factory, say, could be configured to support sensors, then robotics, then video surveillance applications over the same air interface. “A flexible Ai-native air interface could change the face of network optimisation,” Korpi said.
Does all of this R&D conform to the wishes of the operators expressing their desired outcome at the top of the article? That will depend, as ever, on industry engagement with its key customer bases running in tandem with the creative innovation departments at operators and vendors. Whatever the eventual criteria for membership of the 6G club is, or is not, you’d have to think that more efficient and more intelligent networks, and real-life applications such as sensing and positioning, would seem to stand a good chance being accepted.
