Sunday, October 23, 2011

FCC white paper shows that “spectrum efficiency” is meaningless

The FCC Technical Advisory Council’s (TAC) draft white paper on spectrum efficiency metrics (25 September 2011) is an excellent piece of work. It is authoritative, instructive, and demonstrates decisively that spectrum [1] efficiency metrics are a meaningless concept.

While they don’t say this in so many words, members of the Sharing Working Group perhaps intended this conclusion to be drawn; “spectrum efficiency” is a DC catchphrase that is hard to avoid, and it would probably be unwise to refute it overtly…

The following elements of the paper imply that the “spectrum efficiency” concept is useless:

  1. There is no metric that can be applied across the myriad of different wireless services.
  2. The metrics are incomplete, even within a service.
  3. While the paper suggests metrics for specific services, the taxonomy of services is arbitrary.

  • There is no way to compare the “efficiency” of one radio service (aka one “spectrum use”) to another, denying politicians the pseudo-scientific rationale they dream of for converting a frequency band allocation from one use to another.
  • Even within a given service type, there is no defensible way to rate one deployment’s performance over another; even if one scored much lower using the relevant efficiency metric, its defenders could invoke any of the long list of “additional efficiency considerations” to deny that the comparison was valid.

The paper also misses an opportunity: It hints at the importance of cost effectiveness rather than mere efficiency, but doesn’t address this broader context.

Follow-up posts:

There is No One Metric

The draft white paper makes a key admission in the very first paragraph: “Unfortunately, as discussed in more detail herein, there is no single measure of spectrum efficiency that can be applied across the myriad of different services that rely upon wireless systems.”

 To get a flavor of just how much how the metrics differ, here's what it came up with for its chosen service classes:
Satellite Broadcast: Information bits per second per Hz of allocated (licensed) spectrum within each common program area, e.g. bits / (second – Hz)

Satellite Point-to-Point: Information bits per second per Hz of allocated (licensed) spectrum per square mile of service area, e.g. bits / (second – Hz – sq. mi.)

Terrestrial Broadcast: Information bits per second per Hz of allocated (licensed) spectrum within each common geographic area, e.g. bits / (second – Hz), times the average number of users simultaneously served.

Terrestrial Personal Communication: Information bits per second per Hz of allocated (licensed) spectrum per geographic service area, e.g. bits / (second – Hz – sq. mi.)

Terrestrial Point-to-Point: Information bits per second per Hz of allocated (licensed) spectrum) x (transmitted distance) per square mile of service area

Terrestrial Hybrid: Information bits per second per Hz of allocated (licensed) spectrum over the geographic area served, e.g. bits / (second – Hz - sq. mi.), times the average number of users simultaneously served
An even more profound qualification emerged as the Working Paper declined to offer a metric for radar, concluding that “commonly applied efficiency measures (such as bps/Hz) are not appropriate for radars since the spectrum efficiency of a radar system cannot be directly compared to the spectrum efficiency of a typical communications system.” For this and other reasons, it was “unable to identify or evaluate suitable spectrum efficiency metrics for radar systems at this time.”

It declined to address spectrum efficiency metrics for “passive” (mostly scientific) uses, and short range systems that typically operate on an unlicensed or “licensed by rule” basis.

Bottom line: there is no way to compare different services with each other.  This is a decisive weakness, since the advocates of spectrum efficiency metrics had hoped that that metrics would place arguments over the re-allocation of spectrum from one service to another (because one was supposedly “less efficient”) on a more “scientific” footing.

The Metrics are Incomplete

So, there is no one metric to rule them all – but can’t they be used to choose between operations within the service class for which they can be defined?  No, because they are incomplete, even within a service.

The white paper admits that “additional efficiency considerations need to be evaluated in addition to the spectrum efficiency metric so that a comprehensive determination of […] system efficiency is properly made.” Since it focuses most of its attention on satellite systems, most of these considerations are introduced in that context; however, as one can see from the factors considered, they obviously apply to any system. The considerations include:
Device size: Larger antennas increase the spectrum efficiency metric, but decrease user value. As the paper says, “In terms of overall system optimization and user value, it is not always desirable to use the largest possible antenna sizes to achieve the greatest spectrum efficiency.” This applies more generally to device design; for example, increased spectral efficiency may come at the cost of higher power consumption, which requires a larger battery.
Response time: A more responsive system increases user value, but requires reserving a greater percentage of the spectrum (time or frequency) for the signaling associated with allocating access on an as-needed basis.
Quality of service (QoS): Improving QoS elements like latency/access time, coverage/reliability, information error rates, and peak-to-average loading ratios will reduce spectral efficiency metrics. Mission-critical public safety voice communications have higher QoS requirements in some dimensions compared to, say, consumer-grade mobile broadband internet access, meaning that the one cannot measure both merely using the recommended spectral efficiency metric.
System-level cost: The trade-off between spectrum and capital costs, of backhaul, towers, and user equipment, implies that maximizing spectral efficiency may not be the most cost-effective solution.
Even worse: not only does a comprehensive determination of efficiency require the consideration of “additional efficiency considerations,” but no guidance is given on how to weight them relative to each other. There is thus absolutely no rational basis, within this recommendation, for comparing one operation with another, even within a service class.

Division into Services is Arbitrary

Not only are the metrics are incomplete; the “services” within which they are defined are arbitrary categories, and thus extremely vulnerable legally and politically.

Of course, it goes without saying that any taxonomy is arbitrary. However, the white paper generously highlights and proves it in this case by citing the system classes proposed in an earlier report that it found to be important: “Definitions of Efficiency in Spectrum Use” (2008) by Working Group 1 of the Commerce Spectrum Management Advisory Committee (CSMAC):

  1. Broadcast Systems
  2. Personal Communications Systems
  3. Point-to-Point Systems
  4. Radar Systems
  5. Satellite Systems
  6. Passive Listeners (e.g., radio astronomy)
  7. Short Range Systems

The TAC White Paper refines the CSMAC taxonomy by refining the distinction between satellite and terrestrial systems, and adding a “hybrid” class. While there are similarities, there isn’t a 1-1 mapping between the systems classifications

  1. Satellite Broadcast 
  2. Satellite Point-to-Point
  3. Terrestrial Broadcast
  4. Terrestrial Personal Communication
  5. Terrestrial Point-to-Point
  6. Terrestrial Hybrid (combination of broadcast and point-to-point)
  7. Radar
  8. “Passive” (mostly scientific)
  9. Short range 

The losing operation in a spectrum efficiency shoot-out in a given service will thus not only argue that additional efficiency considerations prove that it’s doing a fine job after all, but also that it has been arbitrarily and unfairly lumped in with the winning operation. The loser will argue that it should be in a different service category, and advocate an even finer distinction that the one the TAC chose; and since the TAC had already refined the CSMAC’s taxonomy, there’s evidently no reason, in principle, to deny this claim.


[1] In the white paper, the term “spectrum” seems to connote the combination of frequency, geography, and time – i.e. not just frequency alone – since it builds its analysis on ITU-R SM.1046-2, which defines Spectrum Utilization Efficiency as a function (undefined!) of the “useful effect” and the “spectrum utilization factor” of a system for that system, where the spectrum utilization factor is the product of the frequency bandwidth, the geometric (geographic) space, and the time denied to other potential users.

1 comment:

Anonymous said...

This is useful because it is antitechnocratic. Spectrum allocations that may be inefficient from a strictly engineering, economic or administrative standpoint may efficiently achieve public policy objectives. And spectrum-efficient services may make little meaningful contribution to the public at large, and should be revisited.