Friday, April 11, 2014

The Economic Value of Unlicensed Spectrum $228 Billion Annually in the U.S.

According to a new report completed by Telecom Advisory Services, LLC (Raul Katz, Columbia Business School) commissioned by WiFiForward, the economic value of unlicensed spectrum is over $228 Billion per-year in the U.S. alone!
WiFiForward Value of
Unlicensed Spectrum


The Report Overview (1 page) highlights the use-cases and value of each. WiFiForward has also produced an infographic (shown at right) to highlight the various ways in which unlicensed spectrum provides economic value in the U.S.

The report details the value of unlicensed spectrum in the U.S. based on two different economic impacts:
  1. Gross Domestic Product (GDP) - direct sales of technologies, services and applications that run on unlicensed spectrum. This results in $6.7 Billion per-year in value ($4.559 Billion of which is attributed to Wi-Fi).
  2. Economic Surplus - use of technologies that rely on unlicensed spectrum that add value to the economy. This results in $222 Billion per-year in economic value ($91.474 Billion of which is attributed to Wi-Fi).
Missing Data?
I am a bit confused since the value from enterprise Wi-Fi sales and resulting efficiencies appears to be absent from this analysis. Worldwide sales of enterprise Wi-Fi equipment alone is over $4.4 Billion annually according to IDC. A significant portion of that market is in the U.S. but does not appear to be included in the GDP figures from this report, which only include the value from Wi-Fi cellular offloading, WISPs, and wireless PANs (BT, ZigBee, WirelessHART). Additionally, no economic surplus value is attributed to efficiencies that enterprises gain from private Wi-Fi deployments. Surely there is a tremendous amount of cost savings that organizations realize through the use of Wi-Fi that has not been captured in this report!

Additionally, Infonetics forecasts for the carrier Wi-Fi market include exponential growth from $466 Million in 2012 and $388 Million reported in the first half of 2013, to an anticipated total of over $3.9 Billion annual sales in 2016. This would add an additional $3.4 Billion to the annual GDP figure above by 2016.

Comparison to Licensed Cellular Wireless
Just as a quick point of comparison to licensed cellular wireless sales and services, which have been reported as $225 Billion in wireless wide area network (WWAN) sales (including mobile handsets) as of 2006 (Reference The Case for Liberal Spectrum Licenses (Hazlett & Leo, 2010), pg.15). And since the ARPU of mobile network operators is shifting to data-driven pricing, and Wi-Fi offload represents over 50% of mobile data traffic, a case could be made that over half of the mobile handset sales should be attributed to Wi-Fi / unlicensed spectrum instead.

And comparing the data-carriage portion of the WWAN industry to the Wi-Fi offload value, Benkler writes (pg. 98):
Mark Cooper of Consumer Federation of America offers a more expansive approach that includes both imputed value of unlicensed bundled as part of cellular service and savings from Wi-Fi offloading on the supply side and arrives at about $50 billion per year. And in light of efforts to quantify specifically the data-carriage side of Verizon and AT&T’s business that suggest a revenue more on the order of $50 to $55 billion per year for licensed mobile data in the United States, Hazlett and Leo’s claim of a vast disparity in value appears to be inflated. 
So what can we conclude in comparison? The total value of licensed versus unlicensed is very similar, at both an aggregate level and a mobile data traffic handling level, even without reapportioning a significant percentage mobile handsets sales to Wi-Fi.

Comparison to Previous Research
The full report describes the methodology used for arriving at these figures and provides and extensive comparison of the current work with prior research on this topic. It is quite an interesting read! A few of the previous studies and reports on the value of unlicensed spectrum include those listed below.

Note - many of these studies use older statistics for the basis of the figures, which are no longer accurate. These include Wi-Fi penetration in consumer households, the percentage of Wi-Fi traffic offloaded from cellular networks, and mobile handset Wi-Fi penetration.

2011 - FCC Chairman Genachowski estimated that unlicensed spectrum:
- Increased the value of licensed broadband services by $25 Billion per-year
- Reduced cost of cellular service through Wi-Fi offload by another $25 Billion per-year

- $3.8 Billion per-year in direct sales of Wi-Fi equipment (based on 2006 figures, pg. 97) (References The Case for Liberal Spectrum Licenses (Hazlett & Leo, 2010), pg. 14).
- $12 Billion per-year in value from higher speed of Wi-Fi in mobile phones instead of cellular networks (based on 2010 figures, pg. 97) (References The Case for Unlicensed Spectrum (Milgrom, Levin, Eilat 2011), pg. 19).
- $25 Billion per-year in value based on the share of cellular traffic carried over Wi-Fi (based on 2010 figures, pg. 97-98) (References The Case for Unlicensed Spectrum (Milgrom, Levin, Eilat 2011), pg. 18).

- Value of unlicensed spectrum is ~$50 Billion per-year (based on 2010 figures) (pg. 28-29).

2009 - The economic value generated by current and future allocations of unlicensed spectrum (Thanki)
- Value of Wi-Fi in consumer homes in the U.S. is $4.3 - $12.6 Billion per-year based on broadband extension over Wi-Fi, and also increases broadband adoption creating an additional $5.2 - $15 Billion per-year in economic surplus (based on 2006 figures, pg. 27).
- The total value of three unlicensed uses-cases (Wi-Fi in homes, Wi-Fi in hospitals, and RFID) is $16 - $37 Billion per-year in the U.S. (Note: Thanki cautions that these figures are likely significant underestimates on the value of unlicensed spectrum because they only account for 15% of the total unlicensed chipset market.)
- Potential value created by unlicensed / Wi-Fi uses of white spaces could be $3.9 - $7.3 Billion per-year.
- Potential for new unlicensed uses of white spaces for rural broadband and agriculture water savings could be $0.8 - $4.3 Billion per-year in the U.S.


Friday, April 4, 2014

Impact of the FCC 5 GHz U-NII Report & Order on Wi-Fi Networks

Following the news release of the FCC's actions to change some technical rules for the 5 GHz U-NII bands, the official Report and Order was released on Tuesday.

I've read through the R&O, and here are the technical modifications that were approved:
  • U-NII 1 band (5.150 - 5.250 GHz) indoor operation restriction is removed. This allows use of the band for outdoor hotspots, WISPs, and bridge links. The growth of public hotspots will clearly benefit from this change.

  • U-NII 1 band (5.150 - 5.250 GHz) power level restrictions are changed. 
    • AP power levels at the Intentional Radiator may be 1W (previously 50mW) and the EIRP may be 4W using a 6dBi antenna (previously 200mW), and following the 1dB reduction rule in transmitter power for every 1dB of antenna gain above 6dBi. 
    • Client power levels at the IR may be 250mW and the EIRP may be 1W, following the 1:1 dB reduction rule for antenna gain above 6dBi.
    • WISPs may use up to 23dBi antennas on fixed point-to-point links without any corresponding reduction in transmitter power.

      These changes help to unify the U-NII 1 band with the U-NII 2A/2C and U-NII 3 bands so that larger contiguous swaths of spectrum can be combined using 802.11ac, which provides for 160 MHz channels comprised of two 80 MHz channels. In addition, by adopting power levels commensurate with U-NII 3, 80 MHz channels that are not adjacent to one another can easily be combined as well to form a 160 MHz channel. This ultimately provides greater flexibility in combing U-NII 1 with other U-NII 2A/2C/3 channels. The higher power limits also benefit WISPs for use on point-to-point links with higher gain antennas to achieve greater distances and throughput.
    In the [2013] NPRM, the Commission envisioned that harmonizing the power and use conditions across the lower 200 megahertz of U-NII spectrum (U-NII-1 and U-NII-2A) would likely permit the introduction of a wide-range of new broadband products capable of operating at higher data rates than is now possible.
    Globalstar MSS is the only user of this band in the U.S., using it for terrestrial gateway uplink transmissions to the Internet and phone networks (other spectrum bands are used for the satellite spot-beam transmissions, namely the Lower Big LEO and Upper Big LEO bands). Globalstar initially objected to the U-NII 1 outdoor use and higher power level changes. However, the NCTA analysis found little risk of interference if any one of these three conditions are met:
    1.) Outdoor APs do not radiate more than 125mW (21dBm) EIRP at elevation angles above 30 degrees
    2.) The device is used for a Point-to-Point link
    3.) The device operates indoors

    Devices that do not meet one of those three conditions are limited to 250mW conducted power. Additionally, before deploying an aggregate total of 1,000 APs or more outdoors in U-NII 1, companies must report to the FCC, which will facilitate corrective measures if harmful interference does occur. This is due to the fact that all U-NII 1 devices deployed outdoors in the U.S. will contribute to the noise level of the Globalstar MSS.

  • U-NII 3 band (5.725 - 5.825 GHz) is expanded up to 5.850 GHz, adding 25 MHz of bandwidth that now fall under the FCC Part 15.407 rules for U-NII (effectively consolidating FCC Part 15.247 into 15.407). This means that Wi-Fi channel 165 (5825 MHz) now falls under the U-NII 3 band rules instead of the ISM rules. Hopefully this will also mean more consistent support for channel 165 in vendor implementations (it's been spotty thus far). This additional 25 MHz provides for 1 additional 20 MHz channel (165) but no additional 40, 80, or 160 MHz channel capacity. You may want to revisit my post on 802.11ac Channel Planning.

    Other technical changes to the U-NII 3 band include:
    • PSD changes from 17dBm/MHz to 30dBm/500KHz.
    • No power reduction for antennas above 23dBi on fixed point-to-point systems. This should benefit outdoor WISPs and other point-to-point deployments. (Note - all non-PtP systems still require a 1dB reduction in power for every 1dB antenna gain over 6dBi).
  • The FCC also rejected ARRL proposed changes that would have required DFS operation from 5.65 - 5.925 GHz (including U-NII 3 and the proposed U-NII 4 band), citing no demonstrated need and that it would be overly burdensome.
Channel 165 is now part of UNII-3 (FCC Part 15.407), not ISM (FCC Part 15.247)
  • All devices (AP or client) operating in any U-NII band must be secured to prevent unauthorized software modification and to ensure it operates as approved to prevent harmful interference. The exact methods used to secure the software are left to the manufacturer, but must be documented in their application for equipment authorization to the FCC. The FCC is not setting specific technical security requirements since they are likely to change over time, but rather defining the capabilities that should be implemented by manufacturers. They do make note that more detailed security requirements may be necessary later as software-defined radio technology develops. They also declined to implement rules that would force manufacturers to render a device inoperable if unauthorized modifications were made, citing additional complexity and costs resulting in questionable benefits above the software security being mandated.

    All documented instances of harmful interference were found to be by devices certified for operation in the U-NII 3 band which had been manipulated through software controls to operate in the U-NII 2C band and interfered with TDWR (Terminal Doppler Weather Radar) in the 5.60 - 5.65 GHz sub-section.

    To quote from the R&O:
    "The primary operating condition for unlicensed devices is that the operator must accept whatever interference is received and must not cause harmful interference. Should harmful interference occur, the operator is required to immediately correct the interference problem or to cease operation."

    Other uses of the 5 GHz U-NII bands are show below:
5 GHz U-NII Bands with Primary and Secondary Allocations
    These FCC U-NII technical modifications are separate from another proposal currently under study by the FCC and NTIA that would add another 195 MHz of spectrum under U-NII rules in two new bands, U-NII 2B (5.350 - 5.470 GHz) and U-NII 4 (5.850 - 5.925 GHz). For further details read 'Wi-Fi May Get A Capacity Boost, Thanks to the FCC'.
  • DFS rules and compliance measurement procedures have been modified in the two existing U-NII 2 bands (U-NII 2A and U-NII 2C) to prevent harmful interference to TDWR and other radar systems. DFS is already required to be implemented if devices will operate in the U-NII 2 bands, but is being modified as follows:
    • Explicitly prohibits operators from using equipment without operational DFS in the U-NII 2 bands.
    • DFS must be turned on when operating devices in the U-NII 2 bands (it cannot be disabled).
    • Testing of DFS systems will be performed against a new radar waveform that more closely matches current TDWR systems.
    • Devices operating in U-NII 2 bands must now perform DFS radar sensing across 100% of the device emissions bandwidth (instead of 80% as specified in the 2006 DFS Compliance Measurement Procedures, Table 4). This will make DFS detection more stringent, but possibly also more prone to false-positives when radar is adjacent to the device's operating frequency range.
    • The DFS sensing threshold is modified. For devices operating below 200mW EIRP, the EIRP power spectral density (PSD) must now also be below 10mW/MHz in order to use the relaxed sensing threshold of -62 dBm. In practice this shouldn't cause any DFS change for most indoor Wi-Fi operation where AP power output is typically 100mW or less and EIRP is 200mW or less when using a 3dBi antenna. The minimum Wi-Fi channel width is 20 MHz, thus the 10mW/MHz PSD should be the peak limit, and larger channel widths will have even lower PSD. If operating APs at power levels above 100mW or with higher gain antennas above 3dBi then DFS sensing of radar at the lower threshold of -64dBm may come into play which could result in a slightly higher chance of detecting radar.
    • DFS devices no longer have to conform to a "Uniform Channel Spreading" requirement, which was intended to avoid dense clusters of devices operating on the same channel and might increase the risk of interference to radar systems. In fact, the FCC R&O acknowledges that the use of wide channels with an overall reduced number of usable channels can result in more effective spectrum use at a given location. Larger channels also spread the power of signals out uniformly over the frequency band in which the device is operating, rather than concentrated in a narrow bandwidth. This is not news to any Wi-Fi professional familiar with spread spectrum concepts :)

      By removing this requirement, they also acknowledge that dynamic or manual channel selection may be used. Previously, some manufacturers accomplished channel spreading by forcing dynamic channel selection. This should no longer be the case with such products, and manual selection of U-NII 2 channels should be available in all products moving forward.
    • The "Channel Loading" DFS compliance measurement test no longer requires the use of an MPEG video at 30fps, citing the need for greater flexibility to test and certify devices that are not designed or capable of streaming video. Instead, channel-loading tests will be performed with data types representative of the device under test. This should open up the use of U-NII 2 bands to more devices.
    • The TDWR frequency range of 5.60-5.65 GHz is now available for use by U-NII 2C devices again as long as they meet all new and modified rules! For Wi-Fi networks, this means we will gain access to channels 120-128 again. This adds back to the usable inventory: three 20 MHz channels, two 40 MHz channels, one 80 MHz channel, and one 160 MHz channel. This should be of significant benefit in enterprise environments, providing another 80 MHz channel (6 total now) with Wave-1 802.11ac equipment for channel re-use planning!

We will have to wait to see the effect these DFS rule changes have on the usability of the U-NII 2 bands in practice. Many users already shy away from using DFS channels due to the risk of Wi-Fi channel change and instability in the network. I fear this will heighten apprehension and further constrain use of these bands by network operators.

DFS Radar Detection (Sensing) Thresholds
  • Other DFS modifications were not implemented:
    • The FCC declined to adopt a geo-location database requirement to lookup TDWR in the 5.60 - 5.65 GHz band. They believe the updated DFS rules are sufficient.
    • The FCC declined to modify the out-of-band emissions limits for U-NII devices, as their field investigations have not found properly functioning equipment with the current limits to be a problem. Again, the majority of cases were from devices operating in frequency bands which they were not certified to begin with, or devices which had DFS disabled.
  • All of the new rules are subject to the following transition period before they take effect:
    • 12 months after the effective date of the R&O, applications for certification of 5 GHz devices must meet the new and modified rules.
    • Existing devices that do not meet the new and modified rules must cease to be manufactured, marketed or sold in the U.S. 2 years after the effective date of the R&O.
    • Existing devices that operate in U-NII 2 bands must always comply with DFS. If DFS is not implemented or is disabled, the devices may be confiscated.


Wednesday, April 2, 2014

10 Wi-Fi Terms You've Probably Been Using Incorrectly

Sometimes we fall into bad habits. Unfortunately, the improper use of terminology is quite common in the Wi-Fi industry. This can cause a great deal of confusion when people discuss technical topics. Therefore, as a Wi-Fi industry, I think we should start referring to the following terms using more accurate terminology so we are all on the same page.

Here goes:

  1. Over-the-Air Rogue APs - if it's not on your wired network, it's NOT a "Rogue AP" so let's start calling them Neighboring APs so we all know what someone is talking about rather than having to inquire each and every time someone mentions a rogue for clarification. And let's reserve using the term Rogue APs for when unauthorized APs are on the internal wired network.
    Correct Term: Neighboring APs

  2. Co-Channel Interference (CCI) - APs and clients that are operating on the same channel don't cause interference with one another, they contend for the same airtime and backoff if another one is transmitting. This is distinctly different from interference where a transmission cannot be properly decoded because the receiver can't distinguish the valid signal from noise.
    Correct Term: Co-Channel Contention (CCC)

  3. Collision - okay, here is one that most of you may not have really thought deeply about. Collisions don't actually happen on wireless networks (not in the traditional wired network meaning of the term 'collision'). Instead, the receiver simply cannot properly decode a valid signal because it can't distinguish it from the surrounding noise with the precision required by the modulation used.
    Correct Term: Interference

  4. Coverage Area - most Wi-Fi professionals refer to an APs coverage area as the physical area in which they intend for clients to connect to the AP, usually with an associated signal strength (such as -67dBm). However, the RF signal actually keeps going and can cause co-channel contention (see what I did there!) over a much larger area (usually out to a signal strength of around -85dBm). So, to refer to the area in which we expect clients to connect to the AP based on an RF design let's start using a different term such as Association Area and leave the term Coverage Area to refer to the area where CCC occurs.
    Correct Term: Association Area

  5. AES versus TKIP - this one is easy to get wrong, even for Wi-Fi professionals! Many times we interchangeably use AES, TKIP, and WEP to refer to the encryption on the wireless network. However, in so doing we confuse encryption protocols with cipher suites. For accuracy we should always mention like for like. CCMP, TKIP and WEP are all encryption protocols that we configure for a wireless network. Each of those protocols use a cipher suite to accomplish the heavy lifting: CCMP uses AES, TKIP uses RC4, and WEP uses RC4. Thanks to George Stefanick for bringing this up.
    Correct Term(s): Reference protocols (CCMP, TKIP, WEP) or ciphers (AES, RC4) but don't use them interchangeably

  6. 802.1x - I see this all the time in written material to refer to the IEEE 802.1X Port Based Network Access Control. Unfortunately, it should be used with capital letter 'X' since it is a (standalone) standard, whereas lowercase letters refer to amendments to standards (see here). So, whenever you reference it use the correct capitalization (802.1X).
    Correct Term: 802.1X

  7. WAP - many people use this term to refer to an access point and it's just annoying. It's just AP people. Referring to it as wireless AP (WAP) is just redundant.
    Correct Term: AP

  8. Antenna Gain in Decibels (dB) - many people refer to antenna gain in dB, which is incorrect. Decibels (dB) alone is a relative measurement and requires a point of reference. Instead, you should refer to antenna gain referencing either an isotropic radiator (dBi) or less commonly referenced to a standard dipole antenna (dBd). This establishes the absolute reference point for the measurement which actually gives it meaning.
    Correct Term(s): dBi or dBd

  9. 802.11b 1 Mbps and 2 Mbps Data Rates - do you reference all of the lower data rates of 1, 2, 5.5, and 11 Mbps as 802.11b? If you do, you've been using this amendment name incorrectly. The original 802.11 standard (802.11-1997) defined the 1 Mbps and 2 Mbps data rates as part of the DSSS PHY, as is generally referred to as 802.11 Prime. Then in 1999, along came the 802.11b amendment which added the 5.5 Mbps and 11 Mbps data rates as part of the HR-DSSS PHY. So, to be correct, when talking about 1 Mbps and 2 Mbps data rates you should reference 802.11-Prime (not 802.11b).
    Correct Term: 802.11 Prime (or 802.11-1997)

  10. 5 GHz Signals Attenuate Faster than 2.4 GHz Signals - it's common for many Wi-Fi professionals and writers to state that 5 GHz signals attenuates faster than 2.4 GHz signals in order to describe the common symptom that 5 GHz has less effective coverage area. However, this too is incorrect in most circumstances. 5 GHz signals attenuate through free space at the same rate as 2.4 GHz signals according to the FSPL (free space path loss) formula; it is not directly dependent on the frequency of the signal. Instead, the construction of the receiving antenna is a fractional multiple of the frequency to which it is tuned. This makes a standard 1/4 wavelength antenna for 2.4 GHz longer than a 1/4 wavelength antenna for 5 GHz, which causes a difference in antenna aperture. To put it simply, a 2.4 GHz antenna has a larger aperture than a similar 5 GHz antenna and can "capture" more of the signal as it passes by the antenna element.
    Correct Term: 5 GHz Antennas Have Smaller Apertures

Do you have any other terms that are misleading or misused and you think should be corrected? Drop a comment below!