While I was in Germany for the Automobil Elektronik Kongress, the results of the 5G auctions were just announced. Four companies paid €6.5B for spectrum. These were in the 2GHz and 3.5GHz bands.

A few weeks earlier, the US Federal Communications Commission (FCC) concluded auctions for a lot of spectrum in the 24GHz bands.

Historically, the best bands for mobile have been from 600MHz up to 2GHz. These can send signal long distances without excessive attenuation, and the signals can pass through walls and so provide cellular service indoors from basestations outside. These used to be the only bands, but are now known as the low band. These bands were used for the original analog phones, then 2G (GSM in much of the world), 3G, and LTE. Some countries, such as Germany, have allocated additional spectrum like that 3.5GHz band, known as medium-band (or mid-band). Different countries have different spectrum available, but in the chip world they are all grouped together as "sub-6GHz". The chips that go into 5G smartphones need to work in any country—semiconductor manufacturing is a mass production system, and we are way beyond the days when it made any sense, if it ever did, to use different technologies, and thus different chips, in different parts of the world. (For a primer on 5G, see the two posts I wrote recently, What Is 5G? and Why Is 5G Such a Big Deal?)

Look at the frequencies that the FCC just auctioned. Those are at a much higher frequency, and that band is known as mmWave (or rarely high-band). There are two things about mmWave, one good, one bad. The good is that there is lots of bandwidth available. The bad is that almost anything attenuates mmWave signals drastically:

• Oxygen, meaning that it is limited to about 300m in air (although current trials are showing more like 300 ft so far)
• Walls, so in-building coverage requires in-building basestations
• Hands, so handsets require multiple antennas to use the parts you are not holding
• Basically anything solid, like trees, cars, people. There are tricks that can be played with multiple beams and reflections, but basically mmWave is tricky to handle

Spectrum

Between 6GHz and 24GHz, the airwaves are used for other things. To give you an idea of just how complex this is, this chart shows the spectrum allocation, with mobile the light pink color (you can click on the diagram for a 6K-pixel version where you can read the text):

The rest of the world is using the low and mid bands to build out 5G, and will later add mmWave in limited amounts to add additional capacity for dense coverage in places like sports arenas and convention centers. I opened this post with some news about the German auction because I happened to be there, but they are not alone. China, Korea, Spain, Italy, Japan, and Australia, and more all allocated/auctioned mid-band spectrum recently. But the US has not made any mid-band spectrum available, and so has not had any auctions (and has none planned). Note that mmWave doesn't go far so is no use in rural areas. This is not just a problem for rural residents, who won't see any upgrade from 5G, it is a problem for anyone outside dense cities. If it is really true that autonomous driving depends on 5G, as the mobile operators like to insist, then you'll only be able to autonomously drive on freeways outside the US. Luckily, I think the operators are overstating the case. The spectrum currently used for 3G and LTE will get gradually repurposed to 5G anyway, but that can't go too fast or it risks stranding people overnight with existing phones that are unusable in practice.

The US Military

The difficulty in the US is that for ages a lot of that mid-band spectrum has been used for other things such as military radar. At GOMAC a couple of years ago, this was actually the focus of the conference. (See my post GOMAC: A Conference that Starts with the National Anthem.) As I said in that post:

• The experience with WiFi has shown that cooperating devices can share a band successfully, in a very dynamic way, as other devices and interference come and go, without any central organization handling the frequencies.
• Static bands are very inefficient because the military generally wants spectrum in places that industry does not. For example, the mobile carriers want lots of capacity in New York City, while naval radar wants its bandwidth mostly at sea.
• In a conflict, the enemy isn't going to stick to its color on the frequency chart, so military communications must be robust.
• National security is not all about fighting wars, it also has something to do with keeping an economic and technological edge, which may mean giving up spectrum for commercial use. This is called this "balancing economic and national security."

In April, the Defense Innovation Board (DIB) produced a report, The 5G Ecosystem: Risks and Opportunities for DoD (30-page pdf).

In the introduction, the difficulty is made explicit:

U.S. carriers are primarily focused on mmWave deployment for 5G because most of the 3 and 4 GHz spectrum being used by the rest of the world for 5G are exclusive Federal bands in the United States, extensively used by DoD in particular.
...
The rest of the world does not face the same sub-6 spectrum limitations as U.S. carriers, and is subsequently pursuing 5G development in that range. As a result, the United States may find itself without a global supply base if it continues to pursue a spectrum range divergent from the rest of the world.

The tradeoffs are really detailed on page 21, where the report says:

The United States may choose to continue down the path of mmWave, but the rest of the world is focused on building out sub-6 infrastructure, with China in the lead. As a government entity that operates overseas, DoD will ultimately have to learn to operate on that sub-6 infrastructure, regardless of how the United States chooses to implement 5G domestically. For this reason, the United States must invest in sub-6 capabilities and take steps to share its spectrum. However, there are legitimate concerns within DoD that opening up sub-6 spectrum will create a number of operational issues, from spectrum optimization to security vulnerabilities. If DoD operators are forced to share their bands of the spectrum, there are concerns that this may temporarily or permanently reduce the performance of systems. The addition of commercial users would also increase the overall congestion of the sub-6 spectrum, increasing the risk of connectivity interruptions for DoD operators.

The tradeoff is that the DoD is blocking sub-6GHz spectrum since some of it is dedicated to the DoD. But it needs 5G to work on sub-6GHz for its own uses, and it knows it won't get that if there is not the demand that only comes from the spectrum being opened to consumer mobile. On the other hand, it is actually using some of that spectrum, so they can't just give it up. And on the other other hand, they would really like to run their own military sub-6GHz 5G network without the additional complication of us pesky consumers sharing the band and using up their bandwidth.

Recommendations

The report ends with four recommendations, although each has a dozen bullet points, and the fourth is classified. The first is the most relevant to this post, and I'll list a few of the most relevant bullet points, too:

Recommendation #1: DoD needs to make a plan for sharing sub-6 GHz spectrum to shape the future 5G ecosystem, including an assessment of how much and which bandwidths need to be shared, within what timeframe, and how that sharing will impact DoD systems.

• DoD and the FCC must flip their prioritization from mmWave to sub-6 GHz spectrum for 5G.
• DoD must bear in mind that the status quo of spectrum allocation is unsustainable. DoD stands to significantly benefit if it shares some of its sub-6 GHz spectrum.
• Without aggressive action as outlined in this report, we believe there is a high likelihood that the United States will be unable to convince the rest of the world to adopt mmWave technologies as the standard 5G pathway.

Summary

If this doesn't happen, and to some extent even if it does since it's already too late, then the US will have an inferior 5G compared to the rest of the world. It will be limited to the already overloaded low-bands, which will only gradually free up as LTE migrates to 5G, plus mmWave in the densest cities. Like the cobbler's children who have no shoes, the US risks having invented cellular technology only to have no 5G shoes to wear.

There is a piece in Wired written by one of the commissioners on the FCC Choosing the Wrong Lane in the Race to 5G.

Once again, a link to the DIB study. It is very readable, although quite long. If you have anything to do with 5G, I recommend reading the whole thing, or at the very least the three-page executive summary.

Sign up for Sunday Brunch, the weekly Breakfast Bytes email.