Radio Revolutions on the Road to 5G

Rupert Baines

From 1G to 4G, analogue to digital, TDMA to CDMA to OFDMA, and circuit-switched to packet, cellular technology has always evolved in generations. With the arrival of each new technology work starts on the next generation, bringing with it a host of significant innovations.

Despite this trend, however, for the last few years there has been a creeping sense that innovation had stopped. After the widespread re-invention of the entire network architecture that LTE delivered, there’s been a sense that 5G might just be a stepping-stone, another incremental step from LTE-A to LTE-B and beyond. There might be much denser modulation, higher-order MIMO, and yet more frequency bands, but few could see where radical development could happen.

Inevitably, that feeling was utterly wrong.

In the last few months there's been a series of announcements of new technologies that discovered where radical improvements could be made. In doing so, they've almost guaranteed that 5G will be a significant development over previous generations.

To take just three examples...

Full-duplex wireless
This is familiar in the wired telephony world, where the POTS hybrid allows two people to talk and listen at the same time via both TX and RX pairs of wires.

Kumu claims to have developed the equivalent for wireless: enabling two-way duplex via a common frequency and channel at the same time.

As a result, we can almost double our spectral efficiency. Whether that’s back-hauling a small cell on the same band as it's using, switching to P2P backhaul, or doubling data rates, the end result is the same: a huge leap forward, in comparison to the modest gains of recent generations.

The concept itself is not hard to comprehend -- think of how a pair of noise-cancelling headphones subtracts the unwanted signal out to leave just the desired received one. But it is the implementation that will prove tricky, and we don’t yet know how well it works in reality. But no doubt about it: Even if Kumu doesn't succeed, others will have this working and in real systems soon.

A bonus: It might end the annoying quasi-religious division of the industry over FDD and TDD.

A slightly more mysterious invention, but with the potential to be even more radical, comes from Magnacom. They claim to have developed an alternative to the QAM coding used in virtually all digital communication systems. The new, patented technology is called WAM and is being touted as a drop-in replacement for QAM circuitry, requiring no changes to be made to the analogue or RF components and interfacing via the same "I/Q" interface as QAM.

The inventor claims that, when compared to QAM, the new technology delivers 10 dB better performance. That could mean it uses 50% less power, can reach up to four times the distance, and doubles spectrum efficiency -- all while providing a major speed increase.

The Magnacom website is currently short on real details -- though it does link to the patents themselves -- but the team behind these bold claims is impressive, with a strong track record worth taking seriously.

This made a lot of waves when it was announced by creator Artemis, delivered amongst a wave of press activity.

If early demonstrations are to be believed, pCell could potentially solve the impending "wireless spectrum crunch." The technology allows each mobile device to receive full wireless bandwidth from congested base stations, with the potential to increase our total wireless bandwidth by 1,000 times.

To lay my cards on the table, I was initially very suspicious, and was quoted on this. Real innovations usually appear slowly with patents, technical presentations, or papers, not in a burst of mainstream press and PR.

However, as more details have emerged, pCell's claims have started to seem more credible.

The technology makes use of "beam shaping." Instead of covering a whole area with a single, uniform signal that all users receive, it uses many transmitters coordinated to create interference patterns in such a way that a particular place gets a unique signal. That means each user gets his own very localized "bubble of capacity." This is an extension of the CoMP concept that featured in LTE-A, and closely related to ideas like "massive MIMO" or "distributed MIMO." Indeed, it appears pCell is based on a technology called Distributed-Input-Distributed-Output (DIDO), which creator Steve Perlman first mentioned publicly in 2011.

Again, this is a concept that has been around for a while, but making it work in a real-world situation is a significant challenge. We will need to wait and see how scalable and practical pCell is, but it certainly gives a clear indication of where the future is heading.

When will 5G be a reality?
Cellular generations do seem to move in a conveniently tidy 10-year cycle: The GSM MOU was signed in 1987; WCDMA was formalized in 1999 and commercialized in 2003; LTE was frozen as Release 8 in 2008 and commercialized a few years later. Using this very loose approximation, we would expect 5G to be specified around 2018 and commercialized in the early 2020s.

This time, however, some people think it might be a little later. Partly this is because of the mileage still remaining in LTE, and partly because many of these ”big step” innovations are currently at a stage where they still need time to develop and mature.

Bell Labs’ head of wireless research, Theodore Sizer, agrees with this view and envisions a timeline that culminates in 2025. Significantly, he has also highlighted a fascinating point about users of the next generation -- in 2025, 5G will serve a generation of users not born in 2000:

    They will have never known a corded phone, never shared a phone with their parents and siblings, will never have memorized a phone number and known anything other than an untethered world where they were wirelessly connected.

This is the generation of users for whom 5G must be designed.

— Rupert Baines is Chief Marketing Officer of Real Wireless,

(7)  | 
Comment  | 
Print  | 
Newest First  |  Oldest First  |  Threaded View        ADD A COMMENT
User Rank: Moderator
5/16/2014 | 3:15:14 PM
Re: Well done, but does pCell scale
And to have multiple antennas covering the same area from multiple angles means a lot of places needing permits, lease space, power, backhaul etc. So while this is perhaps a technically interesting solution it does have some practical limitations.
User Rank: Blogger
5/16/2014 | 12:08:07 PM
Re: Well done, but does pCell scale

Amsuingly, Simon Saunders did some research on exactly this area about ten years ago...

I think pCell is massive MIMO is CoMP on steroids: if so, it scales linearly with the number of radios.
So, yes, you do need a lot.

I just read something yesterday about Nutaq demoing a 100x100 MIMO system !
Kit Kilgour
Kit Kilgour,
User Rank: Light Beer
5/16/2014 | 11:52:30 AM
Well done, but does pCell scale
Nice article. One of my concerns about the pCell approach is that it seems to need independent enough paths for each UE served - which implies a lot of antennas if the design is to scale
User Rank: Lightning
5/16/2014 | 11:02:35 AM
Well done Rupert
Interesting article.  Very much in sync with my research.
Gabriel Brown
Gabriel Brown,
User Rank: Light Sabre
5/16/2014 | 4:55:17 AM
Re: Radio Revolutions
Absolutely agree on the startups. I think it's fair to say the industry needs next-gen radio startups to push it forward.
User Rank: Blogger
5/16/2014 | 4:51:20 AM
Re: Radio Revolutions
Yes, I would agree.


I mentioned the startups partly because they are more open, partly because that's more fun ;)

I think both beam-shaping / massive MIMO and full duplex do have programs going in the big vendors, although I'm not sure of their status. 

For example, the DoCoMo trial annouced last week included Ericsson doing massive MIMO.

I'm not aware of anyone else doing anything similar to Magnacom but it wouldn't surprise me if there were things in parallel: it is rare for invetions to be cmpletely unique & unexpected.

Philosophically, in major corporate R&D or in academia these concepts are known:  the challenge is implementation and how reliably they work in the real world. These demos area step forward but far from csolid commercialization.

Incidentally, there were two I missed out:

- Orbital angular modulation. It isn't as significant as the hype.

- New error correection. I've heard rumours of "better than turbo code" but not got details. 
Gabriel Brown
Gabriel Brown,
User Rank: Light Sabre
5/16/2014 | 4:25:40 AM
Radio Revolutions
Good piece Rupert. What I wonder is how far developed these techniques are inside the R&D departments of the big radio vendors?

The sense I have from asking those guys is the techniques are more or less known in princple, but that the implementation (as you point out) is the hard part.
5G Poll
Twitter Feed
Supporting Partners