Why I Disabled Low Data Rates This One Time

After eight and a half years, one hundred forty posts, and a shade under one-and-a-half million page views, it is clear to me that one topic causes more controversy than any other on this here Sniff Wi-Fi blog: disabling low OFDM rates (such as 6 & 9 Mbps).

Wi-Fi vendors tell you to do it.

CWNP trainers tell you to do it.

Other bloggers tell you to do it.

I (and a very, very small group of other hardcore Wi-Fi professionals) tell you not to do it.

But this blog post isn't about re-litigating that controversial issue (for the most part).   This blog post is about telling you why I **did** disable the 6 & 9 Mbps data rates on a recent Wi-Fi optimization project.

Given the subject matter of this blog post, I feel compelled to begin with some shameless self-promotion: if you have a problematic Wi-Fi deployment (or if you're in the "design" phase and you want to avoid having a problematic Wi-Fi deployment, or if you would like training on how to avoid having a problematic Wi-Fi deployment and/or fix a problematic Wi-Fi deployment), you can hire me.  Here's my business card:


Earlier this year I was hired for a job to fix just such a problematic Wi-Fi deployment.  It was a very large office space over multiple floors (about 1.1 million square feet, in total).  The Wi-Fi network had been up and running for a couple of months, and throughout that time users were complaining about slow speeds and inconsistent connectivity.

The primary thrust of the Wi-Fi optimization ended up being a fix of the Cisco WLC configuration, and the disabling of several AP radios (after a survey of the premises, of course).  Fixing controller configs and disabling AP radios is all that's needed to optimize the majority of problematic Wi-Fi deployments (which is why Wi-Fi "design" is over-rated), and this appeared to be no exception.

As I neared the end of the project, however, there was one outstanding issue: some people who had older iPhones would experience drops when moving through certain areas of the office.

My first move was to re-survey the areas where iPhone users were experiencing drops in connectivity.  Any time I do any surveying or troubleshooting, I have a four item checklist:
  1. Is the controller config appropriate?
  2. Is the signal strength (RSSI) strong enough?
  3. Has interference from nearby APs (CCI) been eliminated (or, depending on the project, minimized)?
  4. Has non-802.11 interference been eliminated or avoided (if we are not allowed to eliminate the non-802.11 interference source)?
Yes, yes, yes, and yes were the answers to the four questions on my checklist.

If my checklist checks out and if there are still Wi-Fi problems, then the issue is almost certain to be one of two unusual things: multipath or device-specific issues.

Multipath happens when radio waves reflect off nearby surfaces and intersect with direct radio waves.  Multipath is impossible to conclusively identify using off-the-rack Wi-Fi surveying & troubleshooting tools.  I was relatively confident that multipath was not the reason for the mobility problems at this office, because the office did not have large spans of exposed metal, smooth concrete, or other known reflective surfaces.

Device-specific issues are problems that occur because of the way that specific devices behave, and they are difficult problems to deal with.  One reason that device-specific issues are difficult is that device makers usually don't publicly share a lot of information about how their devices work.  The other reason is that device behavior can change dramatically.  For example, six years ago iPhones would send Probe Request frames incessantly when unconnected to a Wi-Fi network.  Today they don't.  

In the case of older iPhones, I knew a couple of things about how they behave.  I knew that all iPhones evaluate prospective APs before deciding which AP to connect to when roaming.  I also knew that older iPhones have weaker processors than newer iPhones.  Thirdly, I knew that the performance of portable electronics tends to degrade due to physical wear & tear (jostling, dropping, carrying in pockets & purses, etc.).

I was also dealing with an open office environment.  The company I was doing work for had very few physical barriers throughout the office.

Based on my knowledge of iPhone behavior and the fact that I was working with an open office environment, I surmised that older iPhones may have been having roaming issues because the iPhones were "seeing" too many APs when roaming.  In theory, older iPhones were caught in the following narrative:
  1. iPhone user walks through the open office.
  2. iPhone notices low RSSI from the iPhone's currently associated AP
  3. iPhone begins to scan for new APs to associate with
  4. **iPhone "sees" "too many" APs, because the office environment is so open**
  5. iPhone takes "too much" time processing information about the APs the iPhone can "see"
  6. iPhone fails to make a connection to a new AP fast enough to avoid the iPhone user noticing the "drop" in connectivity
At this point I must re-iterate that the roaming issues in this office were only being reported from iPhone users (not users of Android phones or any model of laptop computer), and were only being reported from Wi-Fi users who had older models of iPhones.  The fact that older iPhones were the problem dovetailed with the narrative described above.  Android phones don't do as much scanning and processing of "seen" APs when roaming (at least, based on the Android phone behavior I've witnessed), and newer iPhones have more processing power and have suffered less performance degradation due to physical wear & tear.

Where low OFDM data rates come in is that I saw the disabling of 6 & 9 Mbps (aka setting a minimum basic rate [MBR] of 12 Mbps) as a possible fix for the 4th step in my iPhone narrative listed above.  Low data rate frames can be "seen" further from the AP than high data rate frames.  If low OFDM data rates were disabled, then Wi-Fi devices would "see" fewer APs, thus allowing older iPhones to spend less time processing information about "seen" APs.

After getting permission from the the owners of the Wi-Fi network, I disabled the 6 & 9 Mbps data rates and, viola!... 

...things got marginally better.

(I know I shouldn't be telling stories like this!  What I should be doing is BS'ing everyone by saying things like "Welp, they brought me in and now the Wi-Fi is PERFECT."  But non-BS artists would know that enterprise Wi-Fi doesn't work that way.  When dozens of different device types need Wi-Fi access, there are always going to be minor issues.)

In all seriousness, disabling the low OFDM rates did help.  A typical walk around one of the floors of the building before the disabling of 6 & 9 Mbps was resulting in 3 or 4 instances (or more) of degraded video conferencing for users of older iPhones, and after the change some users of older iPhones reported no roaming issues, while others reported no more than 1 or 2 instances of degraded video call quality.

The logical follow-up question to all of this is, "why not just disable low OFDM rates all the time, if disabling low OFDM rates can reduce roaming issues for older iPhones?"

The reason is that disabling low OFDM rates makes it impossible to survey or troubleshoot CCI, while causing a non-noticeable increase in channel overhead.  

When low OFDM rates are disabled, the Ghost Frame problem makes it impossible for the true range of APs to be "seen" when using Wi-Fi surveying or troubleshooting tools.  And the amount of channel overhead increases only slightly; from ~1.0% with an MBR of 12 Mbps to ~1.8% with an MBR of 6 Mbps, for APs using 3 SSIDs.

So there, Wi-Fi industry.  Are you happy now?  There are circumstances where low OFDM data rates should be disabled.  They just aren't very common.

***


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. Did you leave them disabled to solve the issue? As the older phones age out and go away, will you re-visit the site to re-enable? Do the clients understand the side effects of this change, or are they focused on the immediate issue?

    Thanks!

    (we discuss this internally about once a year, with many benefits we still have left them enabled to avoid the side effects. We also have a residential pool of users who bring an abundance of device types, so we can't safely predict ripple effects).

    ReplyDelete
    Replies
    1. I left 6 & 9 Mbps disabled.

      It was a consulting job, so if they bring me back I will re-assess whether 6 & 9 Mbps should be re-enabled.

      This company was focused on getting the Wi-Fi working now. My project report included all Wi-Fi related controller settings, including the disabling of 6 & 9 Mbps. The report included why each setting was chosen, and the potential impact of setting adjustments.

      Delete
  2. the link to Revolution Wifi at the bottom of this article you did resolves to Donald Trump website

    https://www.sniffwifi.com/2016/05/go-to-sleep-go-to-sleep-go-to-sleep.html

    ReplyDelete
  3. Could you have left the lower rates enabled, but configured the AP's not to respond to probe requests below a certain threshold? I've seen devices in distribution centers that have the same issues as the older iPhones you described (vocollects mostly) when they 'see' to many AP's. Dropping the probe response threshold of the AP's to only respond to probe requests stronger than -75 or so typically helps these devices 'see' the nearest AP's and roam appropriately.

    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