AI radio and 6G at Mobile World Congress 2022
Joe Barry, VP wireless Communications business unit for Analog Devices at Mobile World Congress this week talks to Nick Flaherty on plans for adding AI to its radio chips
5G OpenRAN (O-RAN), AI and 6G are key themes for the Mobile World Congress, back this year in Barcelona, Spain after two years’ break during the pandemic.
“If you look at 5G with release 17 and 18 from 3GPP, then Release 19 is when you will probably start to see 6G,” said Joe Barry, VP wireless Communications business unit for Analog Devices, who is based in Limerick, Ireland. Work on Release 18 is going on now, with the key protocol coding set to freeze by March 2024, which means the Release 19 work will start to emerge later in 2024 and 2025. European 6G projects such as Hexa-6G have been underway for the last year.
“Of course there’s an insatiable appetite for data, and more spectrum, but then you look to the use cases, is it around augmented reality (AR) and virtual reality (VR) or are there potentially others that could help to shape what the network needs to look like? Some people talk about terahertz spectrum, but when I look at the advantages and disadvantages of spectrum then there’s the golden bands of 6-8GHz and 8-12GHz opening up for 6G.”
“I don’t think the industry has figured out yet, there’s a lot of change in the radio access network (RAN) side and in the core network. One thing that 6G can do more on is the sustainability and energy consumption,” he said.
“There’s a lot of good things happening but I think there’s more we can do. 5G has tried to address this. Power consumption peaks in urban areas, so if you can use dynamic allocation of resources and consumption, looking at how much processing you can move to the edge versus the core, there’s lots of ways to optimise it. There’s intelligent caching, more AI in predictive applications, predicting the needs of resources, there’s a lot of good research going on there,” he said.
Related 6G articles
- Major 6G project starts in Europe
- Paper identifies 6G use cases
- Project to develop 6G transceiver
- Who is winning in 6G?
“We are actively developing AI machine learning into the radio in the future, to deliver performance and tackle the power efficiency on the radio side,” he said.
“The other thing about AI is you can go out from the core. You can put different components of AI at different places in the network. Its more about distributed AI that will have an impact on sustainability,” he said. “This is about the RU [radio unit] and interaction with the DU [distributed unit] and the interaction with the standards bodies, and how and where things get done. That’s not a done deal, it’s an on-going discussion.”
“You will see neural network processors in our coming products,” he said.
- RadioVerse chip looks to drive FPGAs out of 5G radios
- Machine learning boosts OpenRAN 5G spectral efficiency
- MWC 2022 sees O-RAN chip deals
Part of this drive to distributed networks is Open RAN, with a range of announcements at the show.
“When you look at what we are doing today around O-RAN I think about this is two ways,” said Barry.
“In macro cells we are implementing those with a combination of ADI SoCs and FPGAs, and that’s what we are showing at the show. In the future the aim is to eliminate the FPGAs to get to the cost, power and size that these radios need.”
“When you look at the other area, massive MIMO radios, these are built around custom ASICs and we are collaborating with Marvell on that. We are seeing the 7.3 split [in the Open RAN structure] standardising there.”
“Talking to ecosystem partners this week is reinforcing what’s happening in private networks, where IoT will be a key component in 5G. Our O-RAN radios already support IoT and all of the traditional 5G basestations are set up for IoT so it’s more a question of when they chose to switch those on. In private networks it’s a new area. The growth in IoT is significant with around 30% CAGR on that business so it’s the fastest growing segment.
Open RAN is a key technology rolling out for cellular networks, particularly for new 5G systems, he says. “With virtualisation and disaggregation, there’s tremendous progress,” he said. “I think its unstoppable in terms of the openness of the networks.”
“A lot of the technology is in its infancy but there’s so much effort that means this will mature quicky in the next couple years and we are starting to see real brownfield deployments with this architecture in the networks,” he said.
“When I compare 4G with 5G, 5G has such a rate of change, it’s a breakneck pace in driving down the cost and weight and I think that will continue for several years to come,” he said.