In August last year we reviewed the 5G Myth, a book questioning the rationale behind 5G. In recent weeks I’ve read several journalists and analysts reinforce that line of reasoning, asking which flavour of 5G would be deployed, and based on what business case.
Odd Generations seem to go awry
Looking back at the four Generations of cellular technology, it’s easy to see that 2G and 4G (the even generations) have been an enormous success. 1G referred to analogue systems such as AMPS, TACS and NMT. These regional systems weren’t interoperable and resulted in a fragmented market.
2G ended up being a battle between GSM and CDMA which GSM won hands down. The global mass market benefits of a common system drove down costs for infrastructure, handsets and the expertise for design and operation.
3G became a choice between three competing standards:
- in the US, CDMA2000 (the evolution of CDMA IS-95)
- in China, the mandated TD-SCDMA
- everywhere else, WCDMA (the evolution of GSM)
It quickly became apparent that WCDMA offered much faster speeds at lower cost and was by far the most widely adopted. A few CDMA networks swapped across from CMDA to WCMDA, such as Telus/Bell Canada, but most stuck to their 2G evolution. China Mobile was the only network which deployed TD-SCDMA and attempts to promote it elsewhere were unsuccessful.
4G is where the world once again became united and has adopted the same technology, benefitting greatly from mass market cost reduction of both infrastructure and handsets. Voice remains the Achilles heel and this still causes issues when roaming which is not quite as transparent or functional as it has been for 2G/3G. VoLTE is not universally adopted, causing voice quality problems in areas of poor coverage and limiting the technical choices for network evolution. Vast swathes of new spectrum have been allocated for 4G which can operate in both FDD and TDD modes, together with refarming of existing 2G/3G spectrum for use by more efficient 4G. It’s even been extended for use with unlicenced bands at 5GHz using LAA.
5G sees the industry once again at a crossroads. It isn’t a really a single standard, nor really even a single solution for a single business case. Regional differences vary widely which may lead to lower take-up and fewer benefits from mass production of common equipment.
The two flavours of 5G
- an upgrade to 4G in the sub-6GHz bands. There may be a little new spectrum allocated around 3.5-4GHz for 5G, but this will continue to be dwarfed by what’s already in use between 450MHz and 3GHz by 2G/3G/4G.
- a parallel network in the millimetre wave bands, say around 30GHz. This is almost entirely line-of-sight short range and will be very challenging to use for typical mobile applications. It won’t penetrate inside buildings, gets absorbed by human bodies radically varying signal strength/quality as you move and turn around.
5G promises low latency – but that’s feasible using 4G
5GNR, or 5th Generation New Radio promises further performance improvements over previous generations. A key benefit often highlighted is low latency – the round trip time for packets to be sent and received. But more important constraints for end-to-end latency are the transmission time to a remote data centre and the processing time of the application. Smartphone/device processing can also be considerable – just compare the time taken to render a webpage on an older smartphone with a newer one. LTE Release 14 and 15 promise much lower latency than commonly available today, with WirelessOne reporting that Ericsson’s roadmap for 4G round-trip latency going down to 4ms by next year. We risk being deflected into spending a fortune to save a few milliseconds of transmission time when ignoring multiple seconds of delays in handset processing or server response times.
Much of the activity I’ve seen at the higher frequencies has been around 30GHz which has previously been used for point-to-point microwave links, such as connecting rural basestations. This high frequency offers substantial bandwidths, say 400MHz at a time, and thus enables very high data rates of multiple gigabits per second.
Although I have seen demos of this working with real-world equipment and a tablet and been shown predicted coverage plots for neighbourhoods, I’m still fairly dubious about the viability of this as a reliable high speed mobile interface.
Verizon Wireless is one of the few champions of 5G mmwave, intending to use it widely as a fixed wireless access solution that provides broadband internet in residential areas. Each home would have their own dedicated receiver that rebroadcasts inside using Wi-Fi. The high cost of cable and wireline broadband may make this viable in the US but is much less attractive in other countries.
5G being tested at many different frequencies
This report from VIAVI (who provide testing gear to networks worldwide) highlights the huge variety of bands being actively tested worldwide. 5G may be a standard technology but it certainly doesn’t appear to be a standard frequency.
Who’s going to sell all that 5G kit?
Ericsson is just one of the major RAN vendors looking for a major revenue surge from 5G deployments. This remains a very limited marketplace, with few companies large enough to make the massive investments and grow the expertise required to develop such products.
Viavi also reveal the major players in 5G trials today, not surprisingly the same “Big 6” that currently supply equipment.
If you assume that Qualcomm and Intel are more likely to be priming the pump than intending to become a major RAN vendor themselves, you can see that 90% of the market is already taken.
There are a few academic projects there too. The OpenRAN initiative doesn’t feature strongly yet, being more focussed on 4G. Where operators do invest in 5G, they are most likely to buy it from their existing suppliers.
What are other analysts and journalists saying?
John Strand, StrandConsult, warns that "Operators are stuck in their old ways of doing things" and "Regulation deters product development and business innovation". He thinks that while it is clear that 5G will be exciting, it is less likely that mobile operators and infrastructure providers will be in a position to monetize the value created. The largest value creators will likely be the makers of connected devices and companies that create services on top of the mobile network.
Joseph Waring, Asian editor for Mobile World Live, gives his personal view questioning the business case for 5G. He quotes Dan Warren, Samsung UK’s head of 5G research, acknowledged at MWC initial 5G rollouts may not be as immediately transformative as expected. Dan said 5G was developed based on the belief mobile connectivity could transform industries outside the traditional telecoms space, but noted there’s been a problem turning that vision into a reality.
Rethink Research takes a more measured approach, warning that there won’t be a step change in investment for 5G. Much of the new technology such as vRAN, densification and Massive MIMO will all be used for 4G before 5G arrives. Significant investment in 5G RAN will depend on it being clearly significantly cheaper than 4G.
Iain Gillot, iGR, recently conducted a survey of 1,000 US consumers (most of whom have never heard of 5G), asking if they'd pay more for a 10x faster service that enables VR and AR.
- 7 percent said they'd pay 11-50% more than today,
- 24 percent said they'd pay up to 10% more,
- 46 percent said they'd expect to pay the same.
So for the typical US ARPU of $60/month (yes, it's much higher than other parts of the world), US operators can expect to recover at most just a few dollars more per month on average. I'd expect that much of the speed/performance improvement for mobile users will be down to the end-user devices and/or delivered using LTE Advanced Pro. It will be more challenging elsewhere with lower ARPUs, but perhaps some edgy marketing and hype might encourage wider takeup and/or higher prices. He thinks 5G will need to look elsewhere for significant revenues such as Enterprise, Fixed Wireless Access and IoT.