Special Report: Towards 6G
The drum beats about 6G are getting louder, even though work on official 3GPP standards will not start for several years – there are three releases of 5G-Advanced specifications to get through first.
It may seem rather premature to be having serious conversations about 6G, at least outside the rarefied realms of academic or industry research labs. Yet if the mobile industry continues to follow its usual pattern, we could expect 6G equipment and devices to start to be deployed at the end of the decade, just 7-8 short years away.
The big question is, will operators deploy another generation of networks? One argument says that they will be financially incapable of bearing the cost of yet another build-out, together with the associated architecture changes that will certainly underpin 6G, including fully distributed RAN-capable edge cloud. The diminishing returns of each successive network upgrade, as evidenced by the first years of commercial 5G, suggest that, if 6G is to be real, it will have to be evolutionary, fully software-upgradeable from 5G.
The big vendors are pledging full backwards compatibility, but at the same time, they are talking about sub-terahertz spectrum (additional spectrum bands are one of the few areas of consensus regarding what 6G may look like). Those would certainly require a radically new air interface and probably a belated move beyond OFDM-based modulations and towards some of the more dramatic approaches that were proposed, and rejected, for some aspects of 5G.
Surely, 6G will have to consist of a brand new architecture if it is to deliver experiences significantly different and superior from those enabled by 5G (including the upcoming three 5G-Advanced releases). There is a valid argument that 5G has, so far at least, failed to fulfil any of the promises made for it in terms of new use cases and experiences, and may continue to do so. This could drive a rerun of 3G, which creaked under the weight of expectations it was ill-equipped to meet, and led to an accelerated race to all-IP 4G to deliver full mobile broadband.
Would it be better for the mobile industry to stop trying to extend 5G in order to shoehorn it into the cloud architectures and new spectrum bands that will be required to enable the full extended reality, industrial metaverse kind of vision? 5G could be retained and enhanced to support conventional business models and operators, and a 4G-plus world. Meanwhile, 6G – with new cloud-native architectures and air interfaces – could be developed from today, and ready to support truly innovative and new business models by the late 2020s.
Just as we have argued that consumer mobile broadband networks do not have to share the same ecosystem and architecture with small cell-based industrial and enterprise networks, so it is not inevitable that emerging ‘metaverse’ use cases, and service providers, will run on the same networks as consumer broadband.
A new breed of network deployers and providers could inject the new money into 5G/6G – which will be essential for 6G to get off the ground as a new generation of networks – and accelerate the development of the fully cloud-native, dynamic, intent-based and user-centric architectures that 5G is unlikely to deliver in full, because of its inherent technical compromises, and because of the legacy networks and applications that need to be maintained. Some of those ‘new generation’ providers may be current operators – NTT Docomo’s 6G trials show that the Japanese MNO is determined to be at the heart of the next wave of cellular technology as it as been since 3G.
Some of the 5G innovations we highlight in this week’s edition, related to healthcare and robotics, will certainly provide intelligence and important learnings for the 6G era, especially in terms of how new technology works in reality in different enterprise environments. And the faster the ecosystem starts to define 6G roadmaps, the sooner the usual race for patents power will be kickstarted (with many geopolitical implications, as highlighted today by Huawei’s IPR power in 5G).
But it is time for the ecosystem to give some deep thought to how 6G will work. Like 5G, it will go a lot further than the air interface and involve hyperscale, IT and semiconductor players from many sectors.
The latter will be particularly influential – from the development of chips that can use high frequency spectrum without excessive power consumption, to the evolution of processors that can support the intensive compute associated with 5G/6G Layer 1 tasks. Vendors such as AMD/Xilinx have immediate targets, such as 5G Open RAN, in view for their new products, but also a roadmap for accelerator chips that looks very 6G-like in performance and flexibility. Meanwhile, Nvidia is leading the way towards ‘superchips’, which are targeted at current high performance computing applications, but which are likely to be required to support 6G RAN if that really does become an essential enabler of the future digital world.