New Spectrum

by Rick Murphy

For more than 4 years, the U.S. Wi-Fi community has anticipated the Federal Communications Commission (FCC) would soon allow unlicensed access to the 5.3 GHz band (5350-5475 MHz) commonly referred to as UNII-2B. This expectation was originally sparked by wording in the 2013 FCC Report & Order 14-30, which announced the possible expansion of UNII devices into the UNII-2B and UNII-4 (5.850 – 5.925 GHz) bands, pending additional technical analysis of those bands.

U-NII band usage as published in FCC R&O 14-30 – 2013

However, after a great deal of deliberation the concerns of organizations operating earth observation satellite stations, federal radar installations, and drone communications systems were validated and Wi-Fi coexistence is contraindicated.  Direct documentation of this action is scarce but several indirect comments from high ranking officials have made it clear that UNII-2B access for Wi-Fi is officially “off the table”.  One such statement was made by Assistant Secretary of Commerce for Communications and Information, Lawrence E. Strickling in a speech to the Hudson Institute on December 16, 2016:

“Unfortunately, the methodical analysis we conducted in collaboration with federal agencies, the FCC and industry led us to conclude that there is no feasible path forward today to share the 5.3 GHz band.”

The reason for all the excitement over access to the UNII-2B band was because it would have bridged a gap between UNII-2A and UNII-2C frequencies.  At the time of the announcement, 802.11ac had just been published and for the first time Wi-Fi connection speeds greater than 1 gigabit per second were possible.  The feature most responsible for these higher speeds was the ability to expand the channel bonding capabilities of Wi-Fi, originally defined by 802.11n with 40 MHz-wide channels.  In order to achieve the greatest speeds with 802.11ac, 80 MHz- and 160 MHz-wide channel options are required.  When using OFDM with Wi-Fi, the standard channel width is set at 20 MHz, or ten MHz above and ten MHz below the designated, center frequency.  In order to create channel bonding, multiple 20 MHz channels are reallocated allowing wider band transmissions.  With 802.11n and 40 MHz channels, only two 20 MHz channels are used, but with 80 MHz- and 160 MHz-wide channels, four and eight 20 MHz-wide channels are required, respectively.

The reality is that the restrictions in available frequency space available for unlicensed Wi-Fi usage effectively nullifies the 80 MHz- and 160 MHz-wide channel options for most installations.  The reason is because Wi-Fi’s CSMA/CA-based channel access mechanism assures that all Wi-Fi devices defer to any other transmitters on the channel through a formal back-off and reattempt procedure known as Distributed Coordination Function (DCF).  The result is that all Wi-Fi devices operating within an assigned channel are negatively affected by each other’s transmissions.  The more devices that are present within a channel, or the fewer channels that are available to be shared, the more profound the negative effects on performance.  The solution is to spread transmitters out from each other into separate channel cells, which creates orthogonal, less congested collision domains.  This practice works well for standard-sized, 20 MHz bandwidths, because there are quite a few 20 MHz channels available. But when channel bonding is enabled, the number of unique collision domains available is reduced by the bonding factor.  If using 40 MHz wide channels the number of orthogonal collisions domains available to choose from is reduced by half.  The effect continues downward exponentially for 80 MHz and 160 MHz channel plans.   This sort of results in a “You can have your cake, or you can eat it, but not both,” scenario.

How Wi-Fi Channels Map to  5 GHz According to FCC R&O 14-30 (areas in red are not usable)

The existing frequency ranges allowed for use by Wi-Fi devices in the 5 GHz band allow for 25 non-overlapping, 20 MHz-wide channels.  When channel bonding is enabled, the choices for unique, non-interfering channels are limited to twelve 40 MHz channels, six 80 MHz channels, or just two 160 MHz-wide channels.  Due to the ubiquitous popularity of Wi-Fi, it is impractical in most cases to design an enterprise-grade WLAN that uses more than 40 MHz-wide channels.  Usually we are limited to using only 20 MHz-wide channels to provide enough isolation from Co-Channel Interference (CCI) and allow acceptable performance levels.

In spite of the FCC’s decision, Wi-Fi requires more channel bandwidth be made available for unlicensed usage – and soon.  Wi-Fi is the primary access medium for network devices, enabling two-way conferencing from huddle rooms, life-critical patient monitoring from hospital rooms, and Ultra High Definition streaming video everywhere – even while moving!  With the trend towards even more demanding applications and expectations Wi-Fi network designers will not be able to satisfy these demands much longer given the current frequency limitations.  Access to the UNII-2B band would have not only provided additional channels (eight 20 MHz, four 40 MHz, two 80 MHz, or one 160 MHz), but would also have allowed a contiguous block of channels from UNII-2A through UNII-3.  C’est la vie.
UNII-4

While the FCC is denying expansion of Wi-Fi into UNII-2B, there is still hope for additional frequency expansion at the top end of the 5 GHz band.  The UNII-4 frequency range, 5850 – 5925 MHz, was reserved decades ago by U.S. and European regulatory bodies to allow Wireless Access in the Vehicular Environment (WAVE) communications from vehicle-to-vehicle and vehicle-to-roadway.  This is the realm of 802.11p and the band is designated as Dedicated Short-Range Communications (DSRC).  The auto industry is seeing significant innovation towards self-driving automobiles and enhanced safety features such as blind-side monitoring.  It would seem that these technologies would rely heavily on the UNII-4 band.

In reality, at least to date, there has been little usage of this band by the automotive industry.  Thus there is thinking by the FCC that this band can be shared by traditional Wi-Fi users in places where DSRC is not being used.  Although the UNII-4 band offers only 75 MHz of frequency space, it is perfectly positioned to bridge UNII-3 – making it possible to enable four 20 MHz-, two 40 MHz-, one 80 MHz-, or one 160 MHz-wide channels.  This is only possible with the addition of the UNII-4 frequencies and the extension to the upper edge of UNII-3 from 5.825 to 5.850 GHz (which came to us by way of FCC Report & Order 14-30 in 2013).  Although the addition of UNII-4 appears to be a logical choice to allow expansion for Wi-Fi, the FCC has ordered additional testing and comment collection on this issue and no timeline has been publicly announced.  We will continue to watch this development with optimism and interest.

6 GHz
Another cause for hope is the recent FCC inquiry into the possibility of Wi-Fi expanding into the 6 GHz band.  As with all potential expansions of Wi-Fi, expansion must not cause harm to existing users of this frequency.  A call for feasibility opinions by any and all interested parties was announced by FCC Notice of Inquiry (NOI) 17-183, titled “Expanding Flexible Use in Mid-Band Spectrum Between 3.7 and 24 GHz” dated August 3, 2017.  Of specific interest to the Wi-Fi community are the solicitation for comments on allowing unlicensed band devices to share bandwidth in the 5.925 – 6.425 and 6.425 – 7.125 GHz ranges.  This will be referred to as the 6 GHz band.  The comment period deadline was originally November 1, 2017 but was extended until November 15 at the request of the Wi-Fi Alliance and several other respondents.  The commenting period has now closed and the FCC, presumably, has gone into deep analysis of this subject.  No timeline for a response has been issued as yet.

The comments to NOI 17-183 are part of the public record but can be difficult to find, especially now that the submittal period has closed.  At last count several hundred organizations and individuals had taken the time to respond.  The submissions appeared to be running about even, for and against the proposal.  Response from Wi-Fi industry manufacturers and standards bodies was weighted heavily in favor of making at least part of this frequency range available as quickly as possible.

As an example, the response from IEEE 802 and 802.11 specifically calls for the creation of four new UNII bands and outlines a timeline for their release as shown in the excerpt below:

“UNLICENSED SHARING OF 5.925 GHz to 7.125 GHz

In accordance with the Commission’s Notice of Inquiry, and to best enable the next generation of Wi-Fi, we believe that the 5.925 GHz to 6.425 GHz and 6.425 GHz to 7.125 GHz bands should be designated as U-NII bands, with limits equal to the current 5 GHz U-NII rules as appropriate. This would allow products based on IEEE 802.11 standards to be adapted to support this band without major redesign.

Based on the segmentation of incumbents operating under different rules (Parts 74, 78 and 101), these bands should be assigned as these separate U-NII bands:

U-NII-5: 5.925 GHz to 6.425 GHz (sharing with FS and FSS incumbents)

U-NII-6: 6.425 GHz to 6.525 GHz (sharing with BAS and CARS)

U-NII-7: 6.525 GHz to 6.875 GHz (same as U-NII-5)

U-NII-8: 6.875 GHz to 7.125 GHz (sharing with BAS and CARS)

In order to expedite the release of this badly needed spectrum, the Commission should consider optimal timing for release of these different bands, based on the complexity of the technology required to mitigate interference to incumbents. Therefore the bands should be released as soon as possible for each band.”

The comment responses from entities who oppose making the 6 GHz band unlicensed state that opening the frequencies would severely degrade current operations of the incumbent license holders.  These comprise satellite communications operators, satellite television services, point-to-point communications, and unlicensed ultra-wideband, next generation positioning applications.  The 802.11 working group points out in its response that 802.11 has a proven track record of being a good spectrum neighbor and has purposefully provided an intelligent PHY and MAC layer capable of detecting existing RF services and mitigating any harmful effects to the incumbents.  802.11 cites the capabilities of DFS/TPC through the 802.11h amendment, beamforming (antenna pointing) as defined by 802.11ac, and geo-location using a national geographical database as specified by 802.11ah (TV White Spaces) plus restrictions on maximum transmit power as some of the currently used techniques that allow Wi-Fi to coexist safely with other services.

There is likely to be a lengthy determination interval on this topic before we see an actual announcement – but in the meantime, we can dream a little.  If all the frequencies in the 6 GHz range were opened up to Wi-Fi (not likely) we would have access to an additional sixty-three non-overlapping 20 MHz channels.  And if UNII-4 were also made available that would mean sixty-seven non-overlapping, contiguous channels.  More importantly we would then have a continuous frequency block from the beginning of UNII-2C (5.490 GHz) all the way through to the top of UNII-8 (7.125 GHz) with only a tiny unusable gap of 5 MHz (5.730 – 5.735 GHz) between UNII-2C and UNII-3.  Imagine how Wi-Fi performance and acceptance would blossom under those conditions!

However, one disturbing comment in the FCC NOI 17-183 may come back to bite us.  In paragraph 30, the FCC writes:

“In considering unlicensed use of the 5.925-6.425 GHz bands, we seek to better understand commenters’ experiences in using existing spectrum that has been made available for unlicensed use. In particular, within the 580 megahertz of spectrum available in the U-NII bands, the most active use appears to have congregated in discrete portions of the bands not subject to dynamic frequency selection (DFS) requirements (such as the 5.25-5.35 GHz and 5.47-5.725 GHz bands). This trend appears to increase the prospect of congestion within the most intensely used portions of the U-NII bands, where DFS is not required. While we do not anticipate modifying our DFS requirements, we are interested in the U-NII industry’s experience in deploying equipment with DFS. Though we are not currently contemplating changes to the DFS rules, we seek to better understand whether DFS is providing meaningful access to spectrum. How are the bands subject to DFS requirements currently being used by U-NII devices? What are the costs of developing and deploying U-NII devices in these bands compared to those bands with no DFS requirement? Is the testing regime used to certify the DFS functionality of devices, required to protect incumbent users as established by the Commission rules, a disincentive to developing and deploying U-NII devices in these bands? Have particular implementations of the DFS requirements been more effective at detecting radar signals or more economical to implement than others?”

I interpret this to state that the FCC is basically asking the Wi-Fi industry why we are not using the frequency ranges they have already provided us.  In essence, they want to know why they should provide us with more spectrum given that we are not taking advantage of that which we already have.

And that is indeed a good question.  Why are we not more effectively using the UNII-2 band (UNII-2A and UNII-2C)?

My initial response would be to blame Wi-Fi hardware manufacturers for being overly conservative with their controller-based RRM mechanisms.  Most current, enterprise-grade Wi-Fi equipment is capable of using the UNII-2 channels properly by initiating DFS and TPC.  Why then, do we see in the field that the majority of WLANs are still avoiding UNII-2 completely?  Why is it so common to see enterprise AP deployments with channels stacked on top of each other in UNII-1 and UNII-3 while few or none are using channels within the DFS-requirement zones?  Is it because the manufacturers are concerned that their algorithms will improperly react to incumbent radar systems using those channels?  Is it because the vendors choose to get their equipment to market faster by applying for equipment certification using less stringent, non-DFS testing rules?

Whatever the reasons is now time to solve these issues and begin proper channel utilization of the UNII-2 band.  All automated RRM routines should include the UNII-2 band in their default channel set.  These algorithms should correctly follow modern established DFS requirements and, in the case of radar pulse detection, they should avoid affected channels and remove them from their available channel set.

The FCC has noticed the absence of Wi-Fi in the UNII-2 band and may use that when determining if we deserve more spectrum.  After all, what good will it do to unlicense more channels if the hardware manufacturers don’t support them?

For those of us who daily deploy and troubleshoot WLANs it feels like the proverbial “rock and a hard place” scenario.  Our end users have huge expectations for Wi-Fi performance, but achieving those expectations requires having enough clear spectrum available to prevent inordinate numbers of collisions – especially when using bonded channels.  For this reason many experienced WLAN engineers still prefer to disable the RRM functions and manage the power and channel plans manually.  I believe that properly implemented RRM mechanisms will continue to be needed in the future; however until they start working more uniformly I maintain that static channel assignments are clearly better.

Although this article has focused on FCC issues and frequency allocations that directly affect the U.S., these same efforts are being explored internationally as well.  With the release of 802.11ax it’s hoped that there will be worldwide support for 6 GHz channel availability.  The WLA is an international association and has members around the globe responsible for making Wi-Fi a satisfactory experience for millions of end users.   I’d appreciate more input from our non-U.S. members on these topics in your home regions.  Ronald van Kleunen has published frequent announcements from the WARC and ITU spectrum management conferences – and that information is valuable.  I would like to invite others who have knowledge of these international spectrum expansion efforts to write-up their perspectives for publication in this newsletter as well.  Or if preferred, you can just send your comments here, addressed to The BROADCAST.

Thanks!  I look forward to hearing your thoughts!