Friday, February 10, 2017

Spectrum is not a scarce natural resource

Almost every policy or technology story about radios starts with the litany that Spectrum is a Scarce Natural Resource. I will argue that this claim is false, and that it matters.

In short:

  • Spectrum is no more a scarce natural resource than sound.
  • It is more accurate and productive to talk about radio operation.
  • Rather than saying “spectrum is scarce”, it’s better to say “radio coexistence is hard.”

The pay-off is that this alternative language makes us focus on what matters – the best way to arrange the operation of radios – rather than on ways to manage a resource (spectrum) that may or may exist.


The notion that spectrum is a scarce natural resource is widespread. Here are a couple of recent ones.

The following was attributed to Jim Kurose, head of NSF’s engineering directorate: "The radio frequency spectrum is a finite but exceedingly valuable natural resource that facilitates a variety of applications and services"
A Bloomberg BNA story about Hill staffers vying to head up the NTIA said that it “plays a critical role in managing one of the country’s most valuable, if intangible, natural resources: radio wave spectrum.”

The word “spectrum” has a wide range of meanings, and a lot hinges on which one is being used. I’ll work through a list of possible definitions; in summary, there’s no definition for which all three attributes hold.

Spectrum could mean

  1. A range of radio frequencies
  2. Radio signals
  3. Signals in a region of frequency, space and time 
  4. Uses in a frequency, space, time region
  5. Operating rights in a frequency/ space region

1. If “spectrum” means a range of radio frequencies…

People often think of “the radio spectrum” as the frequencies of wireless signals, measured in oscillations per second, called hertz in SI units and abbreviated Hz. This matches the definition provided by the US federal government’s Glossary of Telecommunication Terms: “electromagnetic spectrum: The range of frequencies of electromagnetic radiation from zero to infinity.”

Spectrum defined as a range of radio frequencies is thus a bounded numerical range; but those numbers are no more a scarce resource than any other physical unit like seconds, meters or kilograms.

Therefore, let’s assume next that we’re talking about radio signals, not radio frequencies.

2. If “spectrum” means radio signals…

OK, so let’s define spectrum as the phenomenon that the units refer to: electromagnetic radiation or, more specifically, radio signals (radiation between 3 kHz and 3,000 GHz, to use a typical definition). Electromagnetic radiation is a physical phenomenon and so is “natural” in some sense.

However, there is no real limit to the amount of electromagnetic signals (short of frying the fabric of the universe), so one cannot say that spectrum in this definition is scarce. Some of these signals can be put to some use (we use the sun’s light to see by), but calling all radio signals a resource stretches credulity.

To underline the point, consider another very similar form of wave radiation: audible sound. Audible frequencies have a finite range (about 20 Hz to 20 kHz) but it’s unlikely anyone would say that sound is scarce, or a resource in the economic sense. Similarly, radio frequencies have some conventional fixed range, but radio signals as such aren’t an economic resource.

The reader might object that we’re never concerned with all the radiation in the entire the universe; we’re talking about radiation in a certain frequency band, at a given time and place. Let’s try that next.

3. If “spectrum” means radio signals with a given range of frequencies at a particular place and time…

Adding boundaries in frequency, space and time means we’re talking about radiation in some container. But just as in the previous case, there’s no real limit to the amount of radiation one can have in a container.

Someone might object that all those signals will interfere with each other. In fact, they won’t: just like water waves on a pond move through each other without affecting each other, electromagnetic signals propagate through one another without any mutual interaction.

Interference can happen, though, in a receiver. This broadens the discussion from radiation and associated transmitters to include receivers, that is devices that convert information carried by electromagnetic radiation to some usable form.

Invoking interference suggests considering spectrum to refer to the uses of radio signals.

4. If “spectrum” means the uses of radio signals in a particular frequency/space/time region…

Since humans can’t decode radio signals, the definition of spectrum as uses of radio entails technology. The term “natural” is therefore inapplicable if one uses this definition.

So is spectrum, in this definition, scarce?  “Spectrum is scarce” can here most generously be taken to mean “the number of concurrent radio uses in given frequency band, time and place is limited.”

This statement is true; for any given level of technology and amount of available money, there is a limit on the number of concurrent uses. However, as technology improves, the number of concurrent uses in a given place, time period and frequency range keeps growing. For example, if each transmitter only emits a narrow beam in the direction of its intended receiver, one can have many more concurrent transmissions than if all the transmitters emitted energy in all directions. For a given technology level, one can also increase the number of concurrent uses by investing in more and better equipment. For example, one can increase the number of concurrent Wi-Fi sessions in sports arenas by replacing access points with large numbers of intensively managed, professional grade routers.

This means that while spectrum-as-uses is not natural, it may be considered scarce. What about the claim that spectrum-as-uses is a resource? Since dictionaries of economics have very little to say about the concept, so we have to fall back on Wikipedia:

“A resource is a source or supply from which benefit is produced. Typically resources are materials, energy, services, staff, knowledge, or other assets that are transformed to produce benefit and in the process may be consumed or made unavailable.” 

Based on this description, what qualifies “spectrum” as a resource is most likely that an asset is made unavailable in the process of its “use.” It seems pretty clear that radio frequencies or radio signals aren’t resources, but that uses of radio frequencies and (especially) permissions to engage in uses are resources.

For spectrum-as-uses, one might say that a resource is in play because one use can block another. That seems to imply scarcity – but it’s not scarcity of uses, but a limitation on the number of concurrent uses. That, finally, takes us to the legal tools used to manage concurrent uses: permissions to operate.

5. If “spectrum” means the rights to operate radios a particular frequency/space/time region…

We’ve arrived at something which is a resource, and is scarce (at least sometimes): radio operating rights. One might think of radio licenses as akin to taxi medallions, i.e. a permit issued by the government that is required to drive a cab in most cities in America. Just like taxi medallions, radio licenses are scarce, which makes them valuable. Unlike taxi medallions, some – but definitely not all – radio licenses offer exclusive use in a particular frequency band in a specific region.

Radio licenses give operators protection against interference from some other operators. A license, and operating permissions generally including unlicensed allocations, entail a long list of parameters that specify service rules: frequency, geography, transmit power mask, allowed uses, single or paired bands, etc. One might thus say that “spectrum” means the collection of such service rules for a given allocation, or for all allocations taken together.

The clearest case is an exclusive license to operate a specific service in a specific frequency range in a specific region, such as the licenses that cellular companies buy for huge sums at auction. These licenses are localized monopolies that are necessary to operate a radio business, and thus qualify as a scarce resource. However, since they’re a legal construct, they don’t qualify as a natural resource.

So-called unlicensed operation is murkier: anyone with equipment that meets the relevant service rules is allowed to operate. There’s no scarcity in these operating authorizations since there’s no limit on the number of manufacturers that can obtain equipment authorization, or the number of devices they can sell. Since unlicensed devices have to accept interference ( from any other device, they also do not have much in the way of rights, at least in terms of Part 15 rules.

To sum up: I’ve shown that there isn’t any definition of “spectrum” for which it is a scarce and natural and a resource. The word has many meanings that vary with context, ranging from frequency ranges to electromagnetic radiation to uses to operating authorizations. At best, radio operating rights are a scarce (though non-natural) resource, at least in the case of licensed services.

In short, it’s the object of radio regulation; just as art is whatever artists do, spectrum is whatever spectrum regulators regulate.

What about congestion?

I don’t think congestion is a particularly useful concept in radio management. If we imagine spectrum as frequencies or signals (definitions 1, 2 and 3), there is no limit on the amount of “spectrum.” Spectrum isn’t like a road, where there’s a limit to the number of vehicles that can be carried; the amount of electromagnetic radiation that can be present in a box is unbounded. If use imagines that spectrum is operating permissions (the least inaccurate interpretation, in my mind), it’s hard for me to even being imagining what congestion would mean: operating permissions are immaterial concepts.

Of course, few people would find the claim “spectrum is congested” to be surprising. What they mean, I think, is that the amount of concurrent usable operation in a box is limited, and as the number of concurrent uses goes up, performance goes down – which is analogized to “congestion”. (In other words, they’re using definition 4.) As I noted above, the problem isn’t the signals –  which don’t affect each other – but the receivers. Again, I end up at the conclusion that we should be talking about radios, not spectrum.

So after all that, I’m left with two questions: Why is this meme so strong? And why does it matter?

Why this "spectrum is a resource" language?

The pervasive invocation of spectrum a scarce natural resource legitimizes government regulation of radio operation. Scarce resources connote the need for management, and invoking the halo word “natural” justifies government ownership by analogy to natural resources like water, oil and minerals where the government’s right to manage the resource is unquestioned.

For example, here’s the ITU in a tutorial about Spectrum Management Fundamentals: “The radio frequency spectrum … is a very precious resource which must be managed to ensure efficient and equitable access for the services which use it.”

The tenor of official statements is that (with apologies to Frank Sinatra and Sammy Cahn) regulation and resource go together like a horse and carriage: you can’t have one without the other.

The need for regulatory legitimacy, combined with famous charts that lay out radio services like plots of land along a frequency axis, leads easily to talking about spectrum as a scarce natural resource.

However, I don’t believe one needs the “scarce natural resource” rhetoric to justify radio regulation. The alternative formulation I will propose not only does that, but also avoids some unfortunate side effects of spectrum-as-resource language.

A more productive (and accurate) alternative to spectrum as a scarce natural resource: radio operation

Avoiding spectrum/resource language doesn’t mean that the government doesn’t have the responsibility or right to regulate. The analogy with audible sound is pertinent: we don’t have property rights in sound, but there are noise ordinances. Noise codes regulate sound sources, for example by setting different limits for exterior sound levels in residential, commercial, and industrial districts, and specifying quiet hours and hours during which construction and maintenance are allowed. No “sound is a thing” assumption is required to get regulatory traction.

Conceiving of spectrum as a resource leads to futile polemics about whether “spectrum” should be treated – and more importantly, assigned for use – as private property or as a commons. As regulatory practice has shown, it’s not an either/or choice: the spectrum regulators around the world have used both property-like licenses and commons-like rules, and both have brought great benefit.

The assumption that spectrum is a thing can lead to the wrong things being regulated. “Spectrum” is at best a conventional term, used unthinkingly without considering its intended meaning; it’s often just a buzzword that’s used to check a box; at worst it smuggles in unjustified conclusions.

A fixation on spectrum, whether as frequency ranges, electromagnetic radiation or permissions to transmit signals, leads to a fixation on transmitters, following the reasoning that since spectrum is signals, the things that create signals (i.e. transmitters) are the object of regulation. However, as has become increasingly well accepted in recent years, receivers play just as important a role as transmitters in deriving maximum possible value from radio operation.

On the other hand, I’ve shown that it’s possible to avoid the term “spectrum” completely. One can talk about what matters – the operation of radios – rather than about something (“spectrum”) that doesn’t have a stable or accepted definition.

This is easy to do: For example, just say “overlapping use is hard”, or “licenses are in great demand”, rather than saying “spectrum is scarce.” (By “use” I mean the use of radios, not the use of “spectrum”. “Overlapping” refers to concurrent productive use of multiple radio systems in the same frequency, time and place.)

A few more translations:

SAY “radio regulation”
   RATHER THAN “spectrum regulation”
SAY “operating radios”
   RATHER THAN “using spectrum”
SAY “overlapping uses”
  RATHER THAN “a chunk of spectrum”
SAY “co-operating radios”
   RATHER THAN “spectrum sharing”
SAY “operating permissions are a resource”
  RATHER THAN “spectrum is a resource.”

The pay-off is that this language makes us focus on what matters – the best way to arrange the operation of radios – not on ways to manage a resource (spectrum) that may or may exist.

It may also need to new ways to manage radios by making it easier to discard conceptual boxes like frequency bands and license areas, leading to more fluid and flexible radio system operations.

In truth I don’t have a good answer for the pay-off question yet. I know that talking about radios rather than spectrum is closer to reality, and that’s likely to yield benefits. It’s like moving from Ptolemy’s description of astronomical bodies moving on epicycles around the earth to Copernicus’s heliocentric model. The Copernican model wasn’t any better as a description than the Ptolemaic one: because it assumed circular orbits, it still had to invoke epicycles, and in fact made worse predictions. However, it opened the path to Kepler’s elliptical orbits (which did provide more accurate predictions than Ptolemy) and ultimately to Newtonian dynamics.

Thanks to Dale Hatfield and Blake Reid for their ideas and pointers. 

No comments: