Roam, If You Want to (As Long as Your Channels Are 20 Mhz Wide)

Editor's note: In the original publication of this article, the relationship between Wi-Fi channel width and SNR was described imprecisely.  While the point of the article was correct -- that wider channel widths increase the likelihood of Wi-Fi frame failures for mobile client devices -- the mistakes have been corrected.  Thank you to Adrian Granados.

Ahh, roaming. Few things capture the spirit of freedom like the ability to Roam wherever you want to go.



Wi-Fi has its own brand of roaming, and there is one aspect of Wi-Fi roaming that often gets overlooked: 40 MHz and 80 MHz wide channels can make Wi-Fi users feel like they've been bounced from the Love Shack.

Much has been written and spoken about the pros and cons of the three Wi-Fi channel widths: 20 MHz, 40 MHz, and 80 MHz.

20 MHz wide channels allow for the highest number of APs to be deployed -- nice for ultra high-density Wi-Fi -- because each AP takes up less of Wi-Fi's scarce frequency space.  In North America, up to twenty-five APs can be deployed in a given space without causing significant co-channel interference (CCI) issues, because there are twenty-five non-interfering 20 MHz wide channels.  For Wi-Fi deployments using 80 MHz wide channels, only six non-interfering channels exist in North America.

80 MHz wide channels allow for the highest speeds for an individual AP.  A typical Wi-Fi client device supports data rates as high as 867 Mbps when 80 MHz wide channels are used, but maxes out at only a 173 Mbps rate when 20 MHz wide channels are used.

40 MHz wide channels are viewed by some Wi-Fi engineers as the happy medium: higher data rates than a deployment that uses 20 MHz wide channels (up to 400 Mbps, for typical Wi-Fi devices), and more non-interfering channels than what 80 MHz wide channel deployments can support (up to twelve non-interfering channels).

Recommendations I have seen typically involve using 20 MHz wide channels if the budget allows for a dense deployment and/or if a large number of client devices are going to be present.  40 MHz wide channels tend to be recommended for organizations that either don't have the budget for a dense deployment of APs, or organizations that need to squeeze out the highest amount of per-device throughput.  80 MHz wide channels are rarely recommended in the enterprise, because the number of devices per-AP frequently would become overwhelming, due to the limited number of non-interfering channels.

All of the aforementioned recommendations are fine and good when it comes to two important aspects of Wi-Fi performance: speed and user capacity.  But what of another important factor in evaluating Wi-Fi performance: mobility?

Using 40 MHz or 80 MHz wide channels could harm Wi-Fi roaming.  In fact, I have never recommended anything but 20 MHz wide channels when consulting on Wi-Fi deployments that require "seamless roaming".

To understand why wider channels can harm Wi-Fi roaming, one must understand the relationship between signal-to-noise ratio (SNR) and Wi-Fi communication.  SNR is the key factor in determining whether transmitted Wi-Fi frames (commonly called "packets") will be successful.  Low SNR causes Wi-Fi data to fail.  At every Wi-Fi data rate, there is a minimum SNR required for successful frame communication.

In other words, when I look at my iMac and see that it is transmitting frames at a data rate of 878 Mbps...


...that means that my wireless router has to receive those transmitted frames at a high enough SNR to be able to "hear" (a.k.a. "de-modulate", for you techy readers) 878 Mbps frames.  The higher a frame's data rate, the higher the SNR required to "hear" that frame.

An overlooked aspect of SNR is the fact that SNR requirements noise levels -- the "N" in SNR -- increase and decrease with channel width.  If a channel is twice as wide, then the required SNR to be able to "hear" a frame at a given data rate is twice as high the same amount of RF activity on the channel will register as twice the amount of noise.  And if the channel is half as wide, then the required SNR level of noise registered by the receiving Wi-Fi radio is cut in half, as well.  Andrew Von Nagy's Revolution Wi-Fi blog covered the relationship between channel width and SNR noise quite well, in my opinion, a few years ago.

The channel width/SNRnoise relationship can affect Wi-Fi roaming because moving Wi-Fi devices tend to be susceptible to low signal incidents.  Low received signal strength (at either the AP or the client device) can occur in mobile environments due to distance, or "sticky" clients, or the way people hold client devices in their hands, or due to physical environment factors like closing doors, moving forklifts, and more.  When low signal incidents happen in environments where noise levels are naturally two times or four times as high, SNR levels may become perilous for Wi-Fi frame communicaiton.

In a graphical way, it looks like this:



The blue, striped area is the concern.  In that area, SNR may be high enough for the client device to consistently send & receive Wi-Fi data with an AP configured to 20 MHz wide channelsbut because noise levels are "normal".  If noise levels have naturally doubled (due to using 40 MHz wide channels), or quadrupled (due to 80 MHz), then the low signal may result in an SNR that is too low for 40 or 80 MHz wide channels.

As with many Wi-Fi issues, the knee-jerk reaction of Wi-Fi professionals may be to cry, "design!"

Yes, yes.  Design matters.  Ideally, the two APs in the graphic above will be positioned so that the phone never has to use an SNR so low that 40 or 80 MHz wide channels become inoperable.

My experience, on the other hand, is that design cannot overcome some of the issues that arise when 40 or 80 MHz wide channels are used in high mobility environments.  Even with an "optimal" design, an AP or client device could stubbornly transmit using higher-than-expected data rates, thus causing data frame failures even with a "good" SNR.  When the SNR signal has to be at double (for 40 MHz wide channels) or quadruple (for 80 MHz) the level in order for frames to be successful, the problem of stubbornly high data rates is exacerbated.  Or, client devices may be "sticky", and stay connected to a low SNR signal AP, even though the client has moved close to a higher SNR signal AP.

Obviously, I have not seen every Wi-Fi environment.  From the enterprise Wi-Fi environments I have seen, 20 MHz wide channels helps roaming.  The reality is that low SNR instances may occur when people are on the move, and the exclusion of 40 or 80 MHz wide channels can prevent that movement from causing short term Wi-Fi connectivity outages.

***


If you like my blog, you can support it by shopping through my Amazon link, becoming a Patron on Patreon, or by donating bitcoin to 1CcEkG8DJD6VWUr27GwtGzPWRQDFt3qgye



Thank you.

Twitter: @Ben_SniffWiFi

ben at sniffwifi dot com


Comments

  1. I live in a dense community, we have many wifi stations broadcasting their SSID's. What i think would be an answer would be to drop an unwanted wifi packet early in the identification of the packet, with an application. This would be a persistent setting. i would think this would do several things, one in the performance of the system and two cleanup the wifi configuration window. Would also act like a SSID filter for those people that have social issues with naming their SSID stations. is this thinking correct and do you know of anything that does this?

    MacBook Pro

    Webshredder

    ReplyDelete
    Replies
    1. SSIDs are broadcast in beacons, which are transmitted on the primary 20MHz (sub)channel, and at the lowest mandatory rate. This article is not about beacons.
      -- Anders

      Delete
  2. Hi Ben,

    Thanks for a nice article

    What should also be added is that for a comparable modulation and coding rate, the SNR requirements for sustaining a certain PER (0.01) are higher for a higher bandwidth case (80 MHz) as compared to a lower bandwidth (20 MHz). At a given SNR one can use a lower modulation and coding rate (with higher bandwidth) to possibly achieve as good a robustness with higher PHY. rates as compared to a lower bandwidth case. For example, one can consider 80 MHz qpsk 1/2 rate as compared with 20 MHz 16 QAM 3/4. In summary in HD/VHD design where we design with -65 dBm or so RSSI levels at the cell edge, using 80 MHz is not problematic. It is understood that if you want larger cells (coverage dominated design), then 20 MHz might be better.

    Look forward to comments

    ReplyDelete

Post a Comment

Popular posts from this blog

Go To Sleep, Go To Sleep, Go To Sleep Little iPhone

Why You Should Stop Disabling Low Wi-Fi Rates, Illustrated

WPA3 Adds Four Security Enhancements, One of Which Matters