← Back to Posts

Sharing has to be designed into the standard: Peter Rysavy on the Pentagon's 5G idea

Sean Roberts•October 18, 2020•4 min read

In September the Department of Defense published a request for information on sharing spectrum with commercial 5G networks. I asked Peter Rysavy what to make of it. The document itself, he said, is very thin on detail. It asks what spectrum-sharing and capacity-sharing technologies might exist. But read between the lines, and look at the ambitions of some of the companies behind the scenes, and it describes a network the Defense Department would control, serving its own needs and offering any excess capacity to commercial carriers.

Two things worry him. The first is that the whole idea depends on technology that does not exist and would take many years to develop. The second is the spectrum it touches. The request covers 3.1 to 3.55 gigahertz, which includes the 100 megahertz from 3.45 to 3.55 that the Defense Department itself, the NTIA, the FCC and the White House spent at least six months agreeing to open for commercial 5G, and that the FCC plans to auction next year. Taking that band off the table for an unproven approach would, he said, be an extremely bad idea. The carriers planning to bid are extremely nervous, and several industry associations have told the White House that a network the government controls is not the way to go when commercial networks are already working and simply need the spectrum.

The comparison everyone draws is with China. I put the figure of about 600 megahertz of mid-band spectrum allocated there to him, and he called it an extremely good place to start, a foundation for national networks that everything from autonomous cars to AI systems will be built on.

I tried a loose analogy on him. ARPANET was built and run by the federal government, and once its commercial uses were obvious, control was released and it eventually became the commercial internet. Is the Pentagon's idea like that? Partly, he said.

I think a great role for government here is to provide funding, provide expertise in, for instance, spectrum sharing approaches. There's nothing inherently wrong with spectrum sharing. It just has to be designed correctly. 5G, in fact, does have spectrum sharing built into it.

His examples were concrete. A 5G radio channel can carry 4G and 5G traffic at the same time, a feature the industry calls dynamic spectrum sharing. And 5G can operate in the same unlicensed bands as Wi-Fi. In both cases the sharing was designed at a very low level and baked deep into the standards, a process that took years, and equipment is now built against those standards. More intelligent sharing makes sense in the future, because no system uses all of its spectrum all the time. The way to get there is research now, folded into future standards such as 6G.

Then he said the sentence I would hand to anyone running a large program. Sharing with military radar raises the complexity of a network exponentially, and the system could eventually be made reliable.

But what complexity usually means is that things take longer than you expected.

Taking longer is exactly the problem in a deployment race. And there is a second cost. Commercial networks get their reliability from a portfolio: low bands for coverage, mid bands for capacity, high bands for extra speed, combined through carrier aggregation. A network confined to one mid-band slice would be less reliable unless it were built out to the same depth.

The ARPANET analogy holds where government funded research that industry later built on. It breaks where the government would be running a network that competes with commercial carriers for a scarce resource. The sharing that works is the kind every device implements because the standard requires it. That takes a standards cycle, not a request for information. The CBRS story shows how long even the simple version took.

From Lincoln Shorts, the interview series I hosted for Lincoln Network, recorded 18 October 2020.

Share this post