Should we develop a 6G NR waveform? — Madden

The mobile industry spends a huge level of effort on each “G.” Fifty companies contributing in non-stop standardization activities for 10 years. This investment has resulted in a huge payoff through the years, building up a market with over a trillion dollars of revenue every year.

With a nod to Seizo Onoe from NTT DoCoMo, I like the joke that only “even number  generations” can succeed in mobile. 2G was a winner, but 3G didn’t come through with much value. Then, 4G enabled great applications with serious revenue. So far, 5G has been a complete flop from a revenue point of view.

Sure, the wide spectrum blocks are useful, and massive MIMO is awesome. But 5G has not paid for the $100 billion or more in R&D investment from 2011 through 2020. All those engineers sitting in boring 3GPP meetings for countless hours! There’s zero revenue tied to URLLC or any new 5G apps so far.

So, maybe “6G” will be the consolidation that is needed for URLLC and enterprise applications to start creating hundreds of billions in revenue. I think so. In fact, I would generalize to say that every second generation of mobile technology succeeds because people need a generation — 20 years — to change their behavior. The 50-year-olds in the industry did a great job in creating wireless phones. Now it’s time for the 30-year-olds to step up and blow us away with something really different.

Today, my question is whether it’s a good idea to send hundreds of engineers to spend thousands of hours in boring 3GPP RAN meetings over the next 10 years. Should we revisit the OFDMA frame structure? Or should we simply tweak it for the new spectrum and focus our attention on new use cases instead?

There are multiple answers to this question:

• People that focus on 140 GHz as the “6G band” will say that a new frame structure and RF waveform are necessary. James Clerk Maxwell and Enrico Marconi would agree. Semiconductor physics in this band makes OFDMA problematic, and we will need to change to a constant-power approach. But wait a minute. There’s no market at 140 GHz and I don’t hear any credible market expert that would advocate investing billions in standardizing something that can’t penetrate a wet paper bag.
• If the “6G” band will be somewhere between 6 GHz and 15 GHz — the so-called “golden band”— then OFDMA will work out pretty well. The semiconductors will support OFDMA transmissions, and we can extend Massive MIMO for long range and very high capacity. I’m expecting 128T solutions to come to market at 7 GHz and 256T solutions at 10-13 GHz over the next five years. This is a solid way to keep building capacity for our existing applications — a known market with a proven ROI.
• Keep in mind that 6G will be refarmed into our existing bands. We’ve done this in many previous generations, and this will be no different. There’s a lot of value in refarming 6G using Dynamic Spectrum Sharing via software upgrade. That means that we should keep the OFDMA structure and the waveform very similar to the 5G arrangement. Backward compatibility means that we don’t have to yank radios out of the field.
• Open RAN is a great idea, and I support the concept of an open RIC with applications for specific use cases. But this approach requires stability in some fundamental things. Not everything can be changing constantly. If developers are going to innovate with AI and xApps, then I submit that the waveform should remain fairly stable. The 5G NR waveform already has the flexibility that is needed.
• I have spoken with dozens of enterprises about their plans for private 5G networks. Nobody is looking for anything beyond 5G URLLC, and in fact most applications are well served by private LTE. In our research, there is no driving force for further reductions in latency or higher speeds. Instead, the focus is clearly on software maturity in the use case and integration with their IT systems.

In the end, I would give the following advice to all of the leading network vendors and operators considering 6G standardization: Scale it back. We can probably squeeze another 1-2% more spectral efficiency out of small tweaks to OFDMA, so let’s do that.

Let’s not allow the 3GPP committees to run on “autopilot” and create completely new numerology or complex multiple-access schemes to squeeze 5% more spectral efficiency from a “6G NR” type of change. Heck, we will get 40% more capacity from the use of AI to optimize within the complexities of 5G NR and beamforming anyway, so let’s focus our attention there and leave the radio frame structure alone.  

At the same time, let’s focus the attention of our brilliant engineers on the private wireless market. Our recent research in industrial markets and “carpeted enterprises” illustrate a private-cellular equipment market of $2.5 billion this year, and at least 10 years of growth at 20% or more. Do the math. The private 6G market could be bigger than our existing RAN market in 10 years.