Sunday, January 08, 2017

You say you want a spectrum revolution

I’ll be attending the PTC’17 conference in a couple of weeks. Our panel is tackling the question The Spectrum Revolution: Where, What, Why, How?, which got me thinking…

You say you want a revolution... (lyrics)

Revolutions are dreadful things. They typically kill thousands of people (sometimes millions), and leave societies crippled for decades: France, Russia, Mexico, China, Iran, etc. And those are the few that "succeed". Most revolts end in a bloody return to the status quo: Spartacus, Wat Tyler, 1848, the Arab Spring, etc.

Techno-economic revolutions aren't as lethal, but not much better for those left behind; just ask the voters in Michigan how they feel about automation and globalization. Do we really want a spectrum revolution?

It’s curious, at first glance, that a conference where more than 70% of the attendees are service providers or carriers is so interested in revolution. Incumbents hate radical change, i.e. revolutions, so why do they want to hear about it? Daniel Gilbert gives some answers in his 2007 best-seller, Stumbling on Happiness (pp. 20-21). According to him, there two reasons we worry about the future even though it makes us feel bad. First, anticipating unpleasant events can minimize their impact on us when they do occur; second, worry and anxiety can motivate us to do the right thing by dramatizing the unpleasant consequences of not doing so. There’s also a related reason mentioned in a lovely recent podcast about Victorian ghost stories: we might fear that progress is taking us somewhere we don’t want to go.

What would it take to trigger a revolution?

What might count as a spectrum revolution, i.e. change that unseats the companies and industries currently in the driver's seat? Here are some possibilities:

  • Spectrum allocations are abolished, leaving radios to figure out among themselves how to operate without causing interference. This dream was the subtext of the PCAST spectrum report. For contrasting academic approaches, see Hazlett and Benkler

  • Federal spectrum allocations are abolished; federal users have to buy operating rights on the open spectrum market. For one approach, see Tom Lenard's ideas about "GSA for Government Spectrum."

  • Above some frequency (30 GHz, for argument sake) indoor spectrum rights are controlled by property owners, not the FCC. Since wall attenuation is so high, there's no risk of interference with public/outdoor deployments where the regulator would retain control. 

  • All (new?) spectrum allocations are for flexible, auctioned, exclusive use; no more commons.

  • All (new?) spectrum allocations are for unlicensed, commons use; no more auctions and exclusive licenses.

  • The FCC is abolished - not only the legacy telecom/internet and broadcasting/media stuff, but spectrum management as well.

I don't think any of them are in the cards, though. For example,

  • I struggle to see how a world without allocations would work, either technically or politically, even if one focused just on domestic operations and ignored treaty obligations on (say) satellite services. There’s been a lot of talk about cognitive radio and Dynamic Spectrum Allocation over the last 15 years – the current buzzword is Spectrum Access Systems – but regulatory action has been meagre and the implementations so far are rudimentary. The legal questions are harder than the engineering ones: how would one define rights clearly enough that diverse parties (and not just repeat players in the same game, like cellular companies sorting out inter-system interference) could figure out disputes without endless litigation?

  • Assigning indoor rights to property owners is technically entirely feasible, and is the case de facto in private settings for GHz radios. However, the legal and financial status quo (e.g. regulators jealously guarding their prerogatives, and the value of spectrum licenses on cellco balance sheets, respectively) is very strong.

  • Arguing for radio assignments to be either all-commons or all-licensed is a tired polemic from the early 2000s; both approaches work, and neither is going away.

  • Perhaps we magically return to the pre-FCC spectrum situation (priority-in-use rules under common law) before the Red Lion decision and the 1927 Radio Act instituted a regime of public interest regulation by a commission. (For this history, see Hazlett's Wireless Craze.) However, in spite of calls to abolish the FCC mostly from right-wing commentators (e.g. Marc Jamison) and even Larry Lessig, spectrum is usually fenced off as an area where a regulator is still required; Marc Jamison noted that “The only FCC activity that would seem to warrant having an independent agency is the licensing of radio spectrum.”

Revolutionary technology?

Given our fondness of talking about industrial "revolutions," one might wonder if there are breakthrough technologies on offer - like, for example, the combination of movable type and cheap paper that led to "Gutenberg's revolution"

I don't see any; moving to higher frequencies doesn’t change the game. Even impossible stuff like the affordable phased array antennas that are needed for mm-wave cellular handsets won't change anything fundamental; it'll just be more of the same old stuff, at a higher frequency.

The closest I can come to potential breakthroughs are:

  • full-duplex transmission actually works (but that would at best double spectral efficiency)

  • very high density, accurate RF sensor networks (that could lead to closed-loop control of distributed RF networks) are cheap enough to deploy ubiquitously


Since I don't see any impending spectrum revolution, let’s see if there were any disruptions on the horizon. Some possibilities:

  • The MNO business model may be crumbling.  ARPU is flat or down, WACC is going to grow (cf. Tren Griffin). The economics of the cellular business (hardware and services, both) are dodgy: revenues are flat or falling, and low cost competitors have entered at least the equipment market. The 5G hype sounds like whistling past the graveyard. Perhaps the king has feet of clay. Mr Schumpeter's ghost is walking.

  • Spectrum is moving from ~2D to ~3D. This creates some new tussles, e.g. cellular is now battling not just the broadcasters but also satellite. The new crop of NGSO players may give GSO satellite companies (and maybe cellular?) a run for their money, though it may be cooperation rather than competition, e.g. Intelsat partnering with OneWeb.

  • There's an emerging consensus that the RF noise floor is rising, and that it's no longer only a problem in the low-frequency band nobody cares about. Cellular companies in the UHF are beginning to talk about it. The FCC TAC had a technical enquiry this past year, and recommended FCC lab testing and NOI/NPRMs. It's not going to change industry structure, though.

  • The FCC enforcement function's been cut severely, and there's no prospect of recovery. That may prompt more delegated/distributed enforcement. I've called for parties to interference disputes to have a right of private action before ALJs, and federal/commercial sharing may have to fall back on contracts not rules. We may also see more "posse enforcement", i.e. private parties taking on regulator-like roles, e.g. in SAS bands. (We have a precedent in frequency coordinators.) I can't see this changing industry structure, but power relationships could be tweaked.

None of these seem particularly earth-shaking, though.

Lessons from history?

I started wondering about what a disruption (or even revolution) might look like, so I started thinking about historical precedents. One can multiply examples indefinitely; here are a few examples of technologies that did (or didn't) restructure industries:

  • Radio broadcast: AM - yes; FM - no; DAB, SDARS - no.
  • Video broadcast: TV - yes; DBS - no in US, yes in UK.
  • moving to new frequency bands: shortwave to medium wave - no; medium wave to VHF - maybe; VHF to UHF - no.
  • satellite: GSO - yes; NGSO - no.
  • spectrum policy: spread spectrum unlicensed - yes; auctions - no.
  • cellular: introduction - yes; GSM - maybe; CDMA - no; iPhone - yes; Android - no.

There are also all the promised disruptions that never happened. The list is endless, but here are a few radio ones I can think of: Northpoint, M2Z, UWB, digital radio (DAB), unlicensed PCS, LightSquared.

The moral of these stories is that disruption is possible – AM, TV, spread spectrum unlicensed, cellular, iPhone – but that it’s rare.


I am neither distinguished nor (quite yet) elderly, but Clarke’s First Law should be quoted at this point:

“When a distinguished but elderly scientist states that something is possible, he is almost certainly right. When he states that something is impossible, he is very probably wrong.”

You tell me that it's evolution..

Overall, then, I’m skeptical about the prospect of earth-shattering changes. There are some incremental but significant things that could shift the ground a little, though:

  • Using risk assessment rather than worst case when deciding allocations (see my paper)

  • Decentralized dynamic access, along the lines of DARPA's SC2

  • RF mirror worlds: closed-loop measurement/simulation/control systems that run spectrum models in parallel with real-world deployments (see my blog)

The wireless business seems pretty resilient to disruption, however, leaving me wondering what one would have to do to actually GET a disruption, let alone a revolution. A tech breakthrough is necessary, but given regulatory sclerosis, political and business change may also be necessary... A collapse of the cellular business model would help shake things up, particularly if it coincided with eight years of deregulatory fervor.

No comments: