6G – a generation too far?

Now that 5G has arrived with an advertising onslaught from operators like EE, what else could there be for the telecom industry to work on but 6G? Due out around the end of this decade, 6G would keep wireless communications on much the same time scale set since 3G first appeared. But questions are now being asked as to whether these generational shifts are working out the way the industry intended.

One of the early aims for 5G was to bring cellular wireless to a bunch of industrial applications that could not use its predecessors. Although 4G made some attempts to cut the round-trip latency for delivering packets over the air, 5G was designed to cut that to a millisecond or so. In principle, this makes it possible to coordinate robots without having to cable them up, assuming you do not try to have the control systems for those robots running in a data centre overseas.

In practice, the wireless operators are, as with 3G and 4G, pushing new phones to consumers and, by all accounts, are not doing that badly at it. Both 3G and 4G suffered from false starts and so took a while to get out of the gate. 5G was somewhat luckier and may have had a boost from the pandemic and the shift to remote working, and living.

At the recent 6G World Symposium, Andreas Müller, general chair of the 5G Alliance for Connected Industries and Automation and head of corporate-sector communication and network technology at Bosch’s corporate research centre, said there is still some work to do on the standard to get it ready for factories. “The problem for industrials is that a lot for them only comes with Releases 16 and 17,” he said. The 3GPP group’s Release 16 or 5G Phase 2 was finalised last summer, two years after the initial Release 15; Release 17 is due to become a finalised standard next year.

It is possible there is pent-up demand among factory owners to install their own 5G picocells to support robotic control and mobile augmented reality. But adapting their operations is nowhere near as simple as buying an upgraded handset. “It simply takes time,” Müller says. “We have different innovation cycles between the ICT and OT [operating technology] industries. You don’t buy a new robot every two years. You also have to take into account that the ecosystem has to grow from scratch. We don’t only need [5G-compatible] chipsets, we need robots, automated guided vehicles and so on.”

Though Müller stresses the need to get earlier buy-in from new markets before embarking on the standardisation process, something that happened later with the 5G process; the mismatch between requirements and real-world use-cases is causing some operators to question whether these generational revolutions are hitting the spot.

Maria Cuevas, principal manager for mobility research at BT, said on a panel on the network plans for 6G that the economics for wireless operators are not promising. They have spent a lot on the build-out, complicated by the recent switch away from Chinese suppliers because of geopolitical concerns. And they have seen user behaviour change significantly because of the pandemic. Operators like BT could do without another massive change to the infrastructure before they have finished recouping the investment in the previous generation.

“6G has to be more of an evolution than a [new] generation. We need as an industry for it to be a lot more modular and introduce features in a modular way without having to do things such as refarm spectrum. It needs to be a much more gradual process than it is today. Generations have happened on ten-year cycles and there were good reasons for that, but it doesn’t mean we have to follow that same pattern in the future,” Cuevas says.

Cuevas also cautioned over the big promises being made and the tendency for them to follow the rule of ten: that 6G should be ten times better. “We don’t really know whether it’s achievable or realistic. We should not be setting arbitrary and possibly unrealistic targets for ourselves.”

The advantage of doing more Gs is the motivational spur: not least to an investment community that, if it believes big profits will ensue, will plough money into agents of change. Without the prospect of a revolution in elements such as coding schemes and business models, the ability to sell the idea of change to those financing the components and subsystems becomes far harder. A generational change may also be needed to deliver on any of the promises.

Paul Hart, executive vice president and general manager of RF power components at NXP Semiconductors, says, “6G is really about ubiquitous connectivity. 5G was metrics-driven. 6G needs to be purposeful. This purposeful network needs to be sustainable: we can’t simply allow power consumption to grow exponentially.”

Hart points out that the upgraded 5G network consumes roughly twice the power of its predecessor. This is not because the hardware used to deliver data wirelessly is unnecessarily profligate. Far from it. “Increases in efficiency [per bit transmitted] have been offset by a faster rise in bandwidth,” Hart notes.

Addressing the energy consumption may only come with big changes in how the packets and bits are encoded, which implies a big upgrade in the base stations. On other hand, a big problem is that the bits are doing way too many miles in the first place. Far more processing needs to be done at the edge of the network, arguably in systems sitting right next to the base stations or at least in the same suburb or village. This is something that, potentially, can be accommodated in 6G, especially as it was a selling point for 5G originally and may slash overall energy consumption in its own right as the higher-bandwidth exchanges are made at short range.

Where 5G evolution and 6G revolution potentially overlap the most is in the management of the network. Already, some chipset designers and equipment designers have harnessed machine learning to control the complex parameters that determine how well a handset that moves around can adapt to changing interference conditions. 6G aims to spread machine learning throughout the entire networking stack: cloud servers are expected to monitor the flow of data to work out better ways to route it. Though there will be advantages in having AI operate on reworked protocols that accompany a new G, operators may decide 5.5G is what they really want.