Thoughts on “Demand” for Unlicensed Spectrum

by on January 23, 2017 · 7 comments

The proposed Mobile Now Act signals that spectrum policy is being prioritized by Congress and there’s some useful reforms in the bill. However, the bill encourages unlicensed spectrum allocations in ways that I believe will create major problems down the road.

Congress and the FCC need to proceed much more carefully before allocating more unlicensed spectrum. The FCC’s 2008 decision, for instance, to allow unlicensed devices in the “TV white spaces” has been disappointing. As some economists recently noted, “[s]imply stated, the FCC’s TV white space policy to date has been a flop.” Unlicensed spectrum policy is also generating costly fights (see WiFi v. LTE-U, Bluetooth v. TLPS, LightSquared v. GPS) as device makers and carriers lobby about who gains regulatory protection and how to divide this valuable resource that the FCC parcels out for free.

The unlicensed spectrum provisions in the Mobile Now Act may force the FCC to referee innumerable fights over who has access to unlicensed spectrum. Section 18 of the Mobile Now bill encourages unlicensed spectrum. It says the FCC must

make available on an unlicensed basis radio frequency bands sufficient to meet demand for unlicensed wireless broadband operations if doing so is…reasonable…and…in the public interest.

Note that we have language about supply and demand here. But unlicensed spectrum is free to all users using an approved device (that is, nearly everyone in the US). Quantity demanded will always outstrip quantity supplied when a valuable asset (like spectrum or real estate) is handed out when price = 0. By removing a valuable asset from the price system, large allocation distortions are likely.

Any policy originating from Congress or the FCC to satisfy “demand” for unlicensed spectrum biases the agency towards parceling out an excessive amount of unlicensed spectrum. 

The problems from unlicensed spectrum allocation could be mitigated if the FCC decided, as part of a “public interest” conclusion, to estimate the opportunity cost of any unlicensed spectrum allocated. That way, the government will have a rough idea of the market value of unlicensed spectrum being given away. There have been several auctions and there is an active secondary market for spectrum so estimates are achievable, and the UK has required the calculation of the opportunity cost of spectrum for over a decade.

With these estimates, it will be more difficult but still possible for the FCC to defend giving away spectrum for free. Economist Coleman Bazelon, for instance, estimates that the incremental value of a nationwide megahertz of licensed spectrum is more than 10x the equivalent unlicensed spectrum allocation. Significantly, unlike licensed spectrum, allocations of unlicensed bands are largely irreversible.

People can quibble with the estimates but it is unclear that unlicensed use is the best use of additional spectrum. In any case, hopefully the FCC will attempt to bring some economic rigor to public interest determinations.

  • Charles Jackson

    You wrote “By removing a valuable asset from the price system, large allocation distortions are likely.”

    Well, yes. But, other things also distort efficiency. We have chosen, by and large, not to use the price system to allocate space on local roadways. We do, however, have parking meters on many of those roads. As technology improved, think E-ZPass etc, the transactions cost of metering roadway use fell substantially and roadway pricing is becoming more widely used.

    Many unlicensed uses are for low-power, short-range applications that individually use relatively little spectrum. The transactions costs associated with licensing and subletting would probably swamp any associated efficiency benefits.

    One issue that concerns me is the language in the bill regarding “unlicensed operations in a range of radio frequencies to meet consumer demand.” Some advocates for unlicensed have promoted the view that unlicensed spectrum should be available in all frequency bands. That was some of the logic behind the arguments for unlicensed use in the TV white space. This position seems to be to be based on weak physics and bad economics. There is a large and vibrant unlicensed ecosystem at 5 GHz. I think it highly likely that adding another 100 MHz of unlicensed spectrum at 5 GHz would create more consumer benefits than would adding 10 or 20 small (10 MHz or so) blocks of unlicensed spectrum here and there.

    I also think that one could make a good case that the FCC’s recent expansion of the unlicensed band at 60 GHz (adding 7 GHz of additional unlicensed spectrum) will generate as much or more consumer benefits than would have been created by licensing that 7 GHz.

    Unlicensed spectrum delivers great value. But, we only have a weak understanding of (1) the marginal value of additional unlicensed spectrum and (2) the efficient margin—which additional unlicensed spectrum would add the most value. As I indicated above, I think that a few large contiguous blocks of unlicensed spectrum are likely to be the most efficient way to structure unlicensed spectrum. Such blocks support very high data speeds and allow the operation of many devices in close proximity.


  • Michael Marcus

    You view unlicensed bands as an “either … or”. While in some cases possible unlicensed spectrum could alternatively be used for licensed applications, many of the unlicensed bands are of little interest to the CMRS gang.

    The original 3 ISM bands that become the home for Wi-Fi, Bluetooth and myriad other lesser known applications were coveted by no commercial interests because of the microwave oven use in the bands. Some bands are restricted to low power due to primary federal users.

    The 60 GHz band was of no interest to licensed operators due to the nature of the quirky radio propagation at that frequency. BUT they ended up as huge beneficiaries of that action when research into practical uses of the this unlicensed band resulted in the surprising discovery that spectrum above 25 GHz could be used for mobile radio applications when MIMO antenna technology is included in the system design. So mmWave 5G can trace its roots to FCC action on unlicensed 60 GHz.

  • bskorup

    Thanks Chuck. I agree with much of what you’ve said, especially with your point that a large contiguous band at 5 GHz would be better than a several, narrower bands. I’m not anti-unlicensed but I do see many problems with the current state of unlicensed policy. I don’t want the FCC to stop unlicensed allocations–just to slow down and study the past 20 years of unlicensed bands before proceeding.

    One, as you noted, scattering unlicensed bands around, each with customized rules to accommodate adjacent and overlapping services, increases the chances the band will fail to attract device makers and users.

    Two, while unlicensed spectrum has (hard-to-measure) innovation value relative to licensed, the increasing resistance from Part 15 incumbents to new, potentially interfering services (like LTE-U) decreases that innovation value. Wi-Fi is great but if first movers are able to populate a band and then throw their weight around at the FCC and Congress to prevent other uses, unlicensed allocations start to resemble beauty contests.

    Three, related to No. 2, I see little concern that an unlicensed allocation is largely irreversible. This decrease in option value for a band must be considered when allocating unlicensed spectrum. (U-PCS is the only instance I know of where an unlicensed band was repurposed and only because it attracted very few devices; TVWS might be able to be rescued for licensed uses for the same reason.) The risk is that in, say, 30 years, when the FCC and the tech industry are looking for licensed spectrum for services that require QoS, hundreds of MHz in 2.4 GHz, 3.5 GHz, 5 GHz and elsewhere will be off limits because they’re populated with 20 year-old APs that people still use.


  • bskorup

    Thanks Michael. I want unlicensed to fulfill its promise as bands where innovators can test and deploy new services. But I see little curiosity at the FCC and Congress as to why some unlicensed bands gain substantial use (2.4 GHz) and why some fail (U-PCS, TVWS). Further, GPS, Bluetooth, and WiFi interests, despite their Part 15 (unprotected) status have sought de facto protection. Before rushing headlong into more unlicensed allocations–as the Mobile Now Act encourages–I’d like to see the FCC slow down and examine licensed-unlicensed tradeoffs and when unlicensed allocations are likely to generate significant use.

  • Charles Jackson

    Well, I don’t understand the opposition to LTE-U. I cannot tell if it arises from:
    A. Well-founded concern about interference issues;
    B. Concern about interference based on flawed analysis; of
    C. A desire to handicap the performance of the CMRS carriers.

    I don’t think that the concern about irreversiblity should be given too much weight. 2.4 GHz is full of weird things like RF stabilized welders, and not so weird stuff like microwave ovens that creates substantial impediment to using it for CMRS-like services. Moreover, use these bands by a CMRS-like service would tend to either (1) reduce the coverage of access point or (2) reduce the data rate that it can support. It would be unlikely to render the bulk of devices useless.

  • Michael Marcus

    Well, I asked the same question during Spectrum Policy Task Force deliberations but my discussion never made it into the final report.

    I did an analysis of ISM band, U-PCS, and U-NII. If you compare the size of each set of rules (measured in word length) with the growth rate of FCC equipment authorizations (public data that was easy to get although sales figures might have been better but FCC did not have ready access to them) you see and inverse correlation between regulatory complexity and growth rate.

    The original ISM band rules were developed by FCC and were pretty simple. The U-PCS and U-NII rules were basically consensus agreements among developer that had specific products in mind. They probably didn’t realize that once the technical details they wanted got codified in FCC Rules they were protected from any change by the “full faith and credit of the APA”. These rules were very detailed and as technology and markets moved ahead they couldn’t be kept up.

    TVWS rules were driven by an uneven balance of TV broadcasters trying to protect their spectrum or at least giving the appearance of protecting their spectrum. They insisted on maintaining the fiction of perfect reception inside the Grade B contour. The proTVWS forces did not have the inside influence at FCC to effectively counter this and were forced into a complex consensus solution with the broadcasters. The original listen-before-talk/cognitive radio proposal was dismissed without even testing that mode in prototypes submitted for FCC testing – apparently at broadcaster behest. The net result was overly complicated rules and then regulatory uncertainty from the incentive auction. Low commercial success to date should then not be a surprise.

    Successful unlicensed systems near clear rules not designed in detail for some use that might never come. (The 60 GHz unlicensed rules have been reasonably successful, but they suffer from details optimized for H-P’s 1990’s market vision and then H-P lost interest in this product line around the time rules were adopted!

    PS – there are no “unlicensed allocations” Unlicensed is a permitted secondary use, not an allocated service.

  • Charles Jackson

    I think there are a few other factors that might contributed to the low sales of U-PCS and TVWS devices.

    Regarding U-PCS:

    A. In the beginning, U-PCS devices had to be coordinated with the incumbent microwave systems. That added cost and complicated deployment. (UTAM process)

    B. The U-PCS band is close to the 2.4 band. For most applications the 2.4 band is as good and devices that operate in that band are a better match to global markets.

    Regarding TVWS

    A. TVWS devices have much less capacity at short range (say less than 100 meters or so) than do unlicensed devices operating in the 2.4 and 5 GHz bands. It’s not “Wi-Fi on steroids” Rather, it is “Wi-Fi on crutches”

    B. The purported range advantage of the lower frequencies in the TV band does not always exist. If you have directional antennas (or MIMO arrays) at each end of the link, then, for equal size antennas, the LOS path loss decreases as the frequency increases. (That’s why the people who build inter satellite links try to use the highest frequencies they can. The LES-9 satellite (1976) supported a link at 36 GHz.)

    C. TVWS devices have to meet fairly stringent limits on out-of-band emissions. This raises the cost of equipment and shrinks the market.

    D. No global market.

    E. Some bad marketing. I heard that one TVWS device did not support VLANs on the Ethernet port. This lack made the product more difficult for some WISPs to use.

    F. As far as I know none of the big guys—firms that sell millions of radio chips per day or per week—bothered to develop chips for TVWS. That lack of entry may reflect the factors listed above that limit the size of the market. But, that lack made designing TVWS products more difficult—restricting the field to fewer players. In contrast, if someone wants to build and sell a Wi-Fi product, they can go buy a few thousand Broadcom or Qualcomm chips and start building. It they want to begin one step farther up the food chain, they can buy Wi-Fi modules

    Bottom line for TVWS—it appears to be the case that there is not a big market for low-bandwidth, somewhat longer range products with no interference protection.

Previous post:

Next post: