The Pentagon’s Spectrum Dilemma

The Washington Post has uncovered some misappropriation of funds on the part of the Pentagon related to COVID-19: “A $1 billion fund Congress gave the Pentagon in March to build up the country’s supplies of medical equipment has instead been mostly funneled to defense contractors and used to make things such as jet engine parts, body armor and dress uniforms.” This comes as no surprise to spectrum analysts as a similar swindle has been underway with respect to DoD’s massive spectrum holdings.

In the 2010 National Broadband Plan, the FCC recommended the re-allocation of at least 500 MHz of legacy spectrum to fixed and mobile use from a variety of sources. We know how to reallocate spectrum: we upgrade or replace legacy applications to reduce their spectrum footprints and we auction of the surplus. As I explained in a  2010 ITIF paper:

We free up spectrum for new services by improving the old ones. Almost every American adult is aware of the digital TV transition at some level, although most weren’t affected by it. Before 2009, 500 MHz of RF spectrum was assigned to television broadcasting (not counting satellite TV):

  • 54-72 MHz channels 2-4
  • 76-88 MHz channels 5-6
  • 174-216 MHz channels 7-13
  • 470-896 MHz: channels 14-83

After the transition from analog to digital TV, 200 MHz formerly occupied by UHF channels 52-83 (698-896 MHz) was reassigned to a variety of uses, mainly cellular and public safety networks, and White Spaces networking and wireless microphones were permitted between allocated TV channels in local markets according to a database.

On its face, the transition freed up 50% of the former allocation for analog TV (including the White Spaces allocation.) If we dig a little deeper, however, we find some more interesting implications.

We naively expected that Defense would be willing to play ball, upgrading its legacy systems to make them use spectrum more efficiently and then turn their surplus over to the public just as the TV broadcasters had done. The Pentagon has not responded in a particularly helpful way, however.

The 2012 PCAST Report

The President’s Council of Advisors on Science and Technology issued a fanciful report in 2012 purporting to make good on President Obama’s 2010 order for the government to make 500 MHz of spectrum available “…to be licensed by the FCC for exclusive use or made available for shared access by commercial and Government users in order to enable licensed or unlicensed wireless broadband technologies to be deployed…”

Rather than filling the pipeline with licensable spectrum, PCAST  recommended a complicated system that would enable government users – primarily Defense – to continue using spectrum as they always had. The only concession DoD had to make was to allow non-government users to access DoD spectrum when and where the generals had no immediate use for it.

They reasoned that it would take too long for the government to do what the TV broadcasters had done, because…hmmm…reasons:

[PCAST] concludes that the traditional practice of clearing government-held spectrum of Federal users and auctioning it for commercial use is not sustainable. In light of changes made possible by modern technology, we recommend that [the President] issue a new Memorandum that states it is the policy of the U.S. government to share underutilized spectrum to the maximum extent consistent with the Federal mission, and requires the Secretary of Commerce to identify 1,000 MHz of Federal spectrum in which to implement shared-use spectrum pilot projects.

Rather that allowing the private sector to share 500 MHz of licensed spectrum among its hundreds of millions of users, PCAST offered 1,000 MHz of spectrum. But the catch was that no private party would ever be able to predict when and where it would ever be able to rely on any of it.

CBRS Etcetera

Citizens Broadband Radio Service (CBRS) is the result of the Commerce Department’s NTIA and the FCC trying to make the PCAST vision of one-way spectrum sharing a reality by creating rules and procedures. The swath of spectrum involved is used by naval systems on the coasts and some radar stations inland:

PCAST recommended that shared spectrum be organized into three tiers. 12 The first tier would consist of incumbent federal users. These users would be entitled to full protection for their operations within their deployed areas, consistent with the terms of their assignments. The second tier would consist of users that would receive short-term priority authorizations to operate within designated geographic areas. Secondary users would receive protection from interference from third tier users but would be required to avoid interference with and accept interference from Federal Primary users. Third tier users would be entitled to use the spectrum on an opportunistic basis and would not be entitled to interference protection. Coordination among different tiers would be accomplished through a database-driven SAS. The use of low-power small cells for broadband would facilitate spectral reuse and sharing, increasing overall efficiency. PCAST recommended that the Federal Government identify 1,000 megahertz of federal spectrum for shared use under this system to create the first “shared use spectrum superhighways.”13

The reality of CBRS is much more pragmatic than the grandiose sharing vision: this month’s auction of Priority Access Licenses allowing wireless operators to use the spectrum in the same way they use conventional licenses was a rousing success for both small and large service providers. Small WISPs report they can achieve several miles of coverage from each of their towers at speeds of 100 Mbps.

CBRS is proving to be a boon for rural broadband, but not for the predicted reasons. Because CBRS PALs are auctioned by county, small WISPs can afford them. All along, rural WISPs have simply needed less costly licenses tailored to their service areas.

DoD Operates on Military Time

DoD has been fascinated with magic radios for a long time. In 1997, it launched the 15 year Joint Tactical Radio System project to create one radio to rule them all. Rather than making radios tailored to applications operating with known parameters on a defined set of frequency bands, JTRS was supposed to be a software-based radio that could be all things to all applications in all terrains.

After pouring $6B into this project, the Pentagon purchased 100,000 JTRS-compliant radios that took ten minutes to boot and could only operate on battery power for 30 minutes. The prototype for JTRS, the 1991 SpeakEasy Multiband Multimode Radio (MBMMR) Program, managed to fill the back of a truck with a single radio, but at least it more or less worked.

The generals wanted to build a Software Defined Radio (SDR), which is fine as far as it goes. But putting ever-changing software in the same old hardware year after year after year is fundamentally at odds with Moore’s Law, the thing that ultimately makes software possible.

The Fundamental Problem is Institutional

SpeakEasy and JTRS were attempts as closing the gap between DoD’s expectations about equipment life and the rate of innovation in radio systems.  As the project manager put it in a 1995 IEEE Communications Magazine article:

In the past, a military radio was developed for a 30-year lifetime. It performed a single function, and was optimized for a particular field application. This was primarily caused by the slowly evolving technology and the difficulty of fitting the military user’s needs into the package space available. Today, commercial applications are driving technology so that the half-life of a component is down to 19 months, that is the time from product release to the use of its next generation replacement in new designs.

Essentially, the military was looking for a way to shove current radio software into a 30-year-old box. The fallacy with this goal, of course, is that the electronics in the consumer radio box change just as rapidly as the software (if not faster.)

3GPP radio generations are created on ten year intervals today, so the thing that needs to change is the military’s procurement practice. How about designing equipment with interfaces that can accommodate modular radios that can be replaced without tearing fighter-bombers apart for each new generation?

From SDRs to Spectrum Sharing

The fundamental benefit of SDR is frequency agility, the ability to operate over a wide array of radio bands with a multitude of radio wave formats. Even with SDRs, the problem with using arbitrary frequency bands is the interference between incumbent services and any foreign device wishing to operate on the same frequency at the same time.

So JTRS led to the research programs on the “opportunistic” sharing of radio spectrum. Opportunistic sharing advocates correctly observe that few radio-based data networks fully occupy their spectrum 100% of the time, so there must be available spectrum the rest of the time.

If incumbent radios are sometimes on and sometimes off, the opportunist should be able to operate successfully during the idle intervals. The practical problem with this plan is that this requires insight into what’s going to happen in very small intervals of future time, substantially less than a second in many cases. The only knowledge the radio opportunist really has is what has happened in the past.

The Wi-Fi Example

The biggest fans of opportunistic sharing cite the example of Wi-Fi as illustration of how well dynamic sharing can work. This analysis is sloppy for two reasons:

  1. Wi-Fi is designed from the ground up to share spectrum without coordination, but most commercial radio systems prefer coordination.
  2. Coordination is more efficient than Wi-Fi because Wi-Fi depends on dead air to avoid interference, while commercial systems thrive by eliminating dead air through scheduling.

In practice, Wi-Fi is never able to use more than half of available spectrum while coordinated systems reach more than 95% occupancy. So the opportunistic sharing concept is premised on poor engineering in incumbent systems. Coordination is the way to go.

Silicon Valley’s Wishful Thinking

The PCAST Report was essentially dictated by then-Google CEO Eric Schmidt and Microsoft’s Craig Mundie, both fans of opportunistic sharing with effectively no professional experience in trying to make it work. Schmidt still speaks about it as if it were the Holy Grail of radio engineering.

Systems like Wi-Fi and Bluetooth are nice to have for small networks, but they aren’t general solutions to all of the problems that need to be solved in large scale, pervasive, and secure radio networks. Schmidt took over the newly-formed Defense Innovation Board in 2016 with the apparent goal of breathing new life into SpeakEasy, JTRS, PCAST, and CBRS.

The DoD Request for Information

When the DoD announced in August that it was willing to share 100 MHz of mid-band spectrum with 5G operators, many of us were excited:

White House and DOD officials said Monday (Aug. 10) that they had identified spectrum between 3450 and 3550 GHz currently used by DOD for key radar applications, that could be freed up quickly for sharing with commercial 5G without sacrificing national security or military uses, specifically critical radar for air defense, missile and gun control, battlefield weapon locations, air traffic control, and more. The Administration will work with industry on the sharing regime, which will require ongoing coordination.

Apparently that last sentence was the only real part of the announcement. DoD has now issued a Request for Information on its apparent wish to retain control of the mid-band while allowing commercial users nothing more than occasional access.

DoD wants to be the real operator 5G services over its mid-band spectrum and to lease access to commercial carriers in much the same way that Mobile Virtual Network Operators (MVNO) like Virgin, Cricket, and Boost use networks operated by firms like T-Mobile, Verizon, and AT&T. In the real world, carriers often end up owning the MVNOs.

Read All the Way to the End

The most Eric Schmidt part of the RFI is the last question:

 M. Previously, when federal spectrum has been reallocated, federal operations have been required to share or relocate to other bands and commercial licensees have received exclusive licenses via auction.  The following questions relate to the above statement:

  • Should DoD consider spectrum leasing as an alternative to reallocation? If so, how could it be implemented?
  • What, if any, legal, polucy [sic], statutory and regulatory changes would be required to implement the proposed leasing approach?
  • How could revenue be shared with DoD under a DSS leasing agreement or any type of leasing agreement?

Taking the leasing question first, there are only two ways it can play out: either commercial carriers lease to DoD, FirstNet-style, or DoD becomes an operator and leases to the the private sector, MVNO-style.

DoD doesn’t like the FirstNet model because it has no trust in the private sector. The private sector makes its case by reminding policy makers of DoD’s checkered history in spectrum engineering and lack of experience offering commercial services.

Institutional Inertia Trumps Technology

I have no doubt that DoD means well, but come on, we’ve been down this road before. SDRs, opportunistic sharing, and government-managed networks are all technical work-arounds to institutional problems. You can’t treat epidemics with band-aids.

There are countries where a government-managed network with multiple MVNOs could work: small, cohesive, urbanized states might pull it off, at least for a while. Singapore and Sweden use this sort of model for fiber to the home, and while there are hiccups, the network service quality is generally good to excellent, but these are wired networks.

We see a different pattern in the US when a government entity becomes a network operator. The project starts off with a bang, but the enthusiasm fades over time as the players revert to their traditional roles. Network operators need to be plugged into the marketing side of things to keep up with changing consumer demands and technological possibilities. This is not the defense mind set.

The Long Term Problem

The DoD is not the kind of institution that provides services to others, it’s highly optimized at seeking appropriations and spending them on its own interests.  I have no doubt that DoD has the ability to create a 5G network, but I have serious doubts about its ability to use it to provide services to others and to upgrade it to 6G when the time comes. Its sense of its mission tends to be very narrow.

The two major problems with spectrum engineering are squeezing the greatest number of reliable bits out of a spectrum band and managing the boundaries between bands most efficiently. Because the commercial carriers are rewarded for solving these two problems in tandem, their long term incentives align with those of users.

Government’s role in spectrum policy is that of a user and a facilitator. It best fulfills these roles when it relinquishes the role of network operator and focuses on its actual mission.

It’s best for all of us if the DoD turns its spectrum over to the private sector and becomes a talented and competent user. The traditional approach – upgrade and repack – has more to offer consumers than does the attempt at transforming DoD into a consumer service provider.

Final Word: DoD Doesn’t Understand the Problem

DoD’s argument for taking over the operation of 5G networks is that they, and only they, can ensure that these networks are secure and protected from mischief on the part of the various actors that want to compromise our national security. This argument is nothing more than bluster.

No network can ever be completely secure. The purpose of networks is to convey information, and this makes them fundamentally vulnerable. The smart way to achieve information security is to distrust the network and employ security measures inside the systems that use the network.

We that up in our podcast with Dr. Lisa Porter, until recently the Deputy Under Secretary of Defense for Research and Engineering.