Dynamic Frequency Selection Part 3: The Channel Dilemma

DFS requirements for protecting mission-critical systems pose channel complications for enterprises deploying 802.11ac.

Jennifer Jabbusch

October 17, 2013

4 Min Read
Network Computing logo

In my last post, I examined the impact of Dynamic Frequency Selection (DFS) on wireless systems operating in 5GHz, particularly 802.11ac devices. In the last article of this three-part series, I look at the channel challenges enterprises face in accommodating DFS requirements in their 801.11ac deployments.

Wireless on 5GHz isn’t new, so you may be wondering why DFS is suddenly becoming an issue. The new challenge stems from one of the mechanisms that enables 802.11ac’s faster speeds: its broader channel widths. Earlier Wi-Fi standards used 20 MHz- and 40 MHz-wide channels. 802.11ac allows 40 MHz-wide channels, and its real speed emerges when using 80 MHz-wide channels. The 11ac specification also outlines 160 MHz-wide channels, but at this point it’s a theoretical implementation and not on the roadmap for most vendors.

Additionally, unlike its 802.11n cousin, 802.11ac operates exclusively in the 5GHz spectrum. While 11n had the full spectrum of 2.4GHz and 5GHz channels at its disposal, 11ac is limited to 5GHz, and only 36% of the 5GHz channels are free and clear of DFS and radar.

802.11ac Channel Allocation (N America)
802.11ac Channel Allocation excluding DFS (N America)

If any part of a channel touches a defined DFS frequency, then that channel is off-limits if a radar is detected, or if DFS channels are not in use. Wider channels mean more opportunity to overlap a DFS frequency. This is why the 802.11ac standard with 80 MHz wide channels has more complications with DFS than previous standards with 20-40 MHz wide channels.

[Read about CSMA/CA and RTS/CTS techniques that let wireless devices share common airspace without disrupting traffic in "Wireless for Beginners Part 2: Avoiding Collisions."]

After reviewing the complications involved in deploying Wi-Fi on DFS-enabled channels, it’s no surprise that most wireless vendors recommend users avoid DFS channels when deploying 5GHz wireless. DFS channels must certainly be avoided for applications and client types that are sensitive to AP re-associations, since changing channels for radar avoidance will drop clients at the time of the avoidance and again during the re-assignment to the original channel. Aside from VoIP applications, wireless-enabled medical equipment and similarly sensitive devices would not tolerate the channel hopping that may accompany deployments on DFS channels.

However, if you follow the very reasonable advice of most wireless vendors and nix the DFS-enabled Wi-Fi channels, a new and equally cumbersome problem emerges. Without the DFS channels, 802.11ac has an extremely limited number of channels to use. This presents a problem when we try to stagger non-overlapping channels in an enterprise environment.

It’s especially a nuisance in areas with high-density requirements, where many APs are in close physical proximity with a lower-power setting and staggered, non-overlapping channels. Some have even remarked that this channel limitation on 5GHz with 11ac puts us back in the same boat we sailed with 2.4GHz and its paltry three non-overlapping channels.

Avoiding DFS channels and sticking to the smaller channel sizes (as recommended by most vendors) isn’t a terrible solution for the immediate future, but it’s something we need to put some thought behind and decide how, as enterprises, we’re going to cope with limited channels and limited bandwidth.

Although use of 160 MHz channels is far down the road, most organizations will want to deploy their 802.11ac with 80 MHz channels, and that gives us only two channels if we exclude DFS bandwidths. Working with only two non-overlapping channels will present a huge challenge in enterprise environments, and it will be impossible for any areas that need to support a high density of devices.

802.11ac Channel Availability (N America)

Organizations using the lesser-known single-channel wireless architecture (SCA) systems are uniquely equipped to handle this channel deficiency presented by DFS. SCA vendors like Meru Networks expect 802.11ac to be a driver for greater adoption of single-channel architectures, and I’m inclined to agree. Deploying on a single channel may sound impossible (I argued this the first time I heard a crazy tale of a one-channel Wi-Fi), but the technology uses 802.11 standards and offers the same services as multiple-channel architectures (MCA) with some additional features.

Even if the FCC opens up more spectrum in the 5GHz space, I expect it will fall under the same restrictions as the current DFS channels. If more DFS space is the industry’s solution to channel limitations and crowding in 5GHz, then we better start thinking of better ways to use DFS channels.

Maybe vendors will come up with a graceful way to move clients without dis-associations, maybe the new enhanced RTS/CTS mechanism in 11ac will offer partial-channel use, or maybe we’ll see more single-channel architectures being deployed. Only time will tell.

Note to readers: I am investigating whether the enhanced RTS/CTS (Request To Send/Clear To Send) mechanisms in 802.11ac will offer a means to split channels into smaller transmit bandwidths as a means for DFS avoidance, instead of eliminating the channel completely. If you have information or thoughts on this, I encourage you to share them in the comment space below.

About the Author(s)

Jennifer Jabbusch


Jennifer Minella is VP of Engineering and consulting CISO at Carolina Advanced Digital, and an author, speaker and consultant for infrastructure security for government, education and Fortune 100 and 500 corporations.

Stay informed! Sign up to get expert advice and insight delivered direct to your inbox

You May Also Like

More Insights