Assess Yourself

Yo, man.  There's a lot of Wi-Fi installers out there flakin' and perpetratin', but scared to kick reality...

So what you want Wi-Fi engineers and admins to do?

Assess Yourself!  

 

To which you might respond, "Duh!"

Any networking or Wi-Fi person worth a lick knows that a network should be assessed before (and, in many cases, after) deployment.  What is far, far, far less well known is HOW to assess a Wi-Fi installation.  

Properly assessing a Wi-Fi installation involves two parts: assessing configuration settings and assessing the physical location of access points (APs) and antennas.  And if you don't assess both, you'll be making a sucker and you equal

While rolling through the streets of Searcy, AR in my main man GT Hill's 1984 Ferrari convertible, I was given some wise insight.  Not, "I only buy things that make money back," a former favorite quote of GT's prior to the purchase of said Ferrari.  Instead, it was wisdom about churches, which are plentiful in Arkansas.  GT said, "every church you see was started because someone thought that every other church sucked".

That's how I feel about Wi-Fi assessments: everything else sucks.  I see throughput tests and heatmaps and capacity planners and the end result is too often the same: Wi-Fi that passes the assessment, but fails when users show up.

I don't like Wi-Fi that fails when users show up.  I think that Wi-Fi should be all about the user.  If the user wants to use a certain device or run a certain application or access a certain resource, the user should be able to do that.  The user shouldn't be told that he/she can't do that just because there are too many other users around.

My core recommendation for assessing Wi-Fi is exactly the same as my core recommendation for installing or troubleshooting Wi-Fi: focus on the radio frequency (RF).  If the RF is good, then the Wi-Fi will be durable.  If the RF is no good, then the Wi-Fi will fail once the users arrive.

I realize that RF is the last thing that most networking folks want to deal with.  RF usually implies expensive tools and esoteric technologies.  Don't worry.  This ain't that.

This is my recommendation for how to Assess Yourself.  And since you're assessing yourself, I'm going to assume that you want to use inexpensive tools and simple methods.  I'll mention areas where the expensive stuff and complex methods are required.  For those issues, a hiring true Wi-Fi professional ::cough::hire me::cough:: is usually a good idea.

Two Parts

A proper enterprise Wi-Fi assessment requires two parts: a configuration assessment and a walking assessment.

I realize that the latter part -- the walking -- is going to turn a lot of networking people off.  If there's one thing that annoys network engineers and admins, it's having to walk to where the users are.  That's for support people.  If you're an admin, chances are you've done your time in support and you don't want to go back.  Well, sorry.  If you want good Wi-Fi you're going to have to do some walking when you Assess Yourself.

Part 1: Configuration Assessment

Let's save the annoying part -- the walking -- for last.  Let's start with the configuration part of an enterprise Wi-Fi assessment.

It may be impolitic to say it to this blog's audience (Wi-Fi engineers, admins, sales engineers, vendor partners, etc.), but most enterprise Wi-Fi problems are due to the configuration of the controller/wireless network management system (WNMS)/AP.  Sorry.  I know that Wi-Fi people don't like getting blamed for Wi-Fi problems (I sure don't), but that's what I see.  Most enterprise Wi-Fi networks are installed with configuration settings that cause unnecessary problems.

I look for four configuration settings when assessing an enterprise Wi-Fi installation: disabled channels, channel width, data rate/modulation & coding scheme (MCS) settings and auto-RF settings, such as RRM for Cisco, ARM for HPE/Aruba, ACSP for Aerohive, etc.

For most enterprise Wi-Fi installations, you can Assess Yourself by configuring the aforementioned settings as follows:

Disabled channels

Disable channels 2, 3, 4, 5, 7, 8, 9 and 10 in the 2.4 GHz (802.11b/g/n) band, thus leaving only channels 1, 6 and 11 enabled.

Disable channels 52 through 144 in the 5 GHz (802.11a/n/ac) band, thus leaving only channels 36 through 48 and 149 through 165 enabled.  (For some vendors' products, this is as simple as disabling 802.11h/dynamic frequency selection [DFS], which is a required protocol for using channels 52 through 144 in North America.)

Channel width

20 MHz wide channels ONLY, for both 2.4 GHz (802.11b/g/n) and 5 GHz (802.11a/n/ac).

Data rate/MCS settings

Disable all direct sequencing spread spectrum (DSSS) rates.  This includes 1, 2, 5.5 and 11 Mbps.  DSSS rates are only available in the 2.4 GHz (802.11b/g/n) band.

Enable all orthogonal frequency division multiplexing (OFDM) rates, while setting rates as Basic/Mandatory or Supported/Optional according to the 802.11 standard.  The 802.11 standard specified 6, 12 and 24 Mbps as mandatory OFDM data rates and 9, 18, 24, 36, 48 and 54 Mbps as optional OFDM data rates.

Auto-RF settings

Auto-RF is tricky because different vendors have different so-called "features" (I call them "complications") that come with their auto-RF protocols.

The important, overarching concept is to neuter auto-RF as much as you can.  Here are some ways to check that:

Set AP transmit power to a minimum of 14 dBm and a maximum of 17 dBm, on both the 2.4 GHz (802.11b/g/n) and 5 GHz (802.11a/n/ac) radios.

Disallow AP channel changes during working hours.

Set a high interference/congestion threshold for AP channel changes.

Avoid a high receiver-start of packet detection (RX-SOP) threshold.

Part 2: Walking Assessment

Now for the annoying part: the walking.

There are three things to know before doing a walking Wi-Fi assessment: where to walk, what device to walk with and what information to look for.  Let's take the second part first. 

What device to walk with

I'm a big fan of compromise.  It makes the world go 'round.  Anything great in life comes from people working together towards a common goal, which inherently involves compromise. 

But there are two things that I STRONGLY recommend that you do not compromise on: 

1. Use a "real" device during your walking assessment.
2. Stop moving whenever you take a reading.

I am fully aware that Wi-Fi folks RARELY do it this way.  Many Wi-Fi folks use site survey software (AirMagnet Survey or Ekahau Site Survey, typically).  Site survey software uses USB adapters (which are most definitely NOT "real" Wi-Fi devices) to read RSSI and site survey software encourages people to take readings WHILE moving.  And that's great... for publicity.  It's not great for technology.

What I'm saying is, toss aside the site survey software if the purpose of your Wi-Fi assessment is to have good Wi-Fi ("technology").  Break out the Ekahau or AirMagnet if your purpose is to TELL SOMEONE what you've done ("publicity").  

AirMagnet and Ekahau create BEAUTIFUL color-coded "heat maps" over your floor plans.  The heat maps are titillating to tech folks and accessible to non-tech folks.  They are great.  But they are not a real representation of what real Wi-Fi users can expect to really experience.  You create a heat map by reading RSSI while moving.  When they use Wi-Fi, your users are typically stationary.  You create a heat map by reading RSSI from a USB adapter.  Your users do not connect to Wi-Fi using USB adapters.  Heat maps and real Wi-Fi are just two different things.

The way I do it is that I carry four devices:

1. A device that allows me to take notes on floorplans. (An iPad Pro with Apple Pencil, in my case.)
2. An 802.11ac device that supports MIMO. (I use my iPhone 6S.)
3. An 802.11ac device that does NOT support MIMO. (A 6th gen iPod Touch.)
4. An 802.11n device that only supports 2.4 GHz. (I have an old Samsung Galaxy Tab and an old iPhone 4S that I rotate between, depending on the job.)

I realize that looks like a lot.  Having enough pockets can even be a challenge when you have four devices.  

I do, however, find it necessary to have multiple devices.  They key is that I want the Wi-Fi to support EVERYTHING.  I want every user on every device to run every app from every where and absolutely love it.  That's the goal.  If I just survey with one type of device, there's an excellent chance I'll miss something.

What information to look for

Uh oh, it's requirements time.

Everyone hates requirements.  (Unless they're the person who sets them, of course.)  Architects, executives, accountants, managers, network engineers and all sorts of other people who don't specialize in Wi-Fi often have a hand in setting Wi-Fi requirements.  That's frustrating.

You know what, though?  It's fine.  Life is conflict.  When you want great things to happen, you have to work with others and compromise.  Your requirements will most likely be the result of some sort of compromise.  Your job is to offer the Wi-Fi perspective on requirements and then use those requirements during your assessment.

We know what TYPE of requirements you'll have: minimum received signal strength indicator (RSSI), minimum number of overlapping APs and maximum number of overlapping APs.  The first two are obvious, I think.  The last one -- maximum number of overlapping APs -- is less obvious.  "Why set a maximum?" you may ask.

Having a requirement for the maximum number of overlapping APs is essential in order to avoid "co-channel interference (CCI)".  CCI happens when more than one Wi-Fi AP use the same frequencies at the same time in the same place.  The result of CCI can be inconsistent speeds, instability and a host of other things that Wi-Fi users hate.

You (and the people you work with) will have to determine your SPECIFIC requirements.  Here's how I approach each one:

-RSSI: I tend to go low, with a -70 dBm minimum.  Most people go higher.  Some go a lot higher, like to -65 dBm.

I go low because I find that too having many enabled AP radios is a far more common problem than having too few.  Having lots of AP radios enabled gives you the warm & fuzzy feeling of strong RSSI and high throughput readings.  But I don't care about those things.  I just want the Wi-Fi to work.  And I know that when the Wi-Fi gets busy, RSSI and throughput tests don't matter.  In fact, a lot of times the Wi-Fi networks with the best RSSI readings and highest throughput tests are the ones most likely to fail under stress, precisely because of CCI.

No matter what minimum RSSI number you choose, make sure you read it accurately.  That means trying to test in a manner that is as close to real world usage as possible.  If your Wi-Fi users walk around the lot of a car dealership all day on their iPhones, then use an iPhone and keep walking during your assessment.  If your project is like my last project, where most people sit or stand at their desks or in conference rooms while on Wi-Fi, then make sure you stop moving whenever you take a reading.

-Minimum number of APs: I typically look for one-and-a-half 2.4 GHz AP radios and two 5 GHz AP radios.  Keep in mind, we are talking about MINIMUMS.  You can always have more than two AP radios in 5 GHz.

The obvious oddity here is my "one-and-a-half 2.4 GHz AP radios" thing.  "What the heck does 'one-and-a-half' mean?" is a logical question.  One-and-a-half APs is something I came up with because I found that "two APs" just doesn't work for 2.4 GHz Wi-Fi (at least, in the cases I've seen).  There are only three non-overlapping 2.4 GHz channels (1, 6 and 11), and that's just not enough channels to hit a two AP minimum and stay below a three AP maximum.  (A fundamental rule of wireless communication is that you never want more concurrent communications than you have available channels.)  

Assessing a Wi-Fi network for "one-and-a-half" 2.4 GHz AP radios does require sprinkling a little bit of art in with the science.  It means that I am hoping to have at least two 2.4 GHz AP radios covering each area, but I am open to going down to one 2.4 GHz AP radio if I have an intractable CCI situation.  There can definitely be some trial and error involved.

-Maximum number of APs: Three AP radios is the limit in 2.4 GHz for reasons mentioned two paragraphs ago.  The 5 GHz band has more channels, so theoretically you can deploy as many 5 GHz AP radios as your organization can afford.  You may want to have a 5 GHz limit, though.

Way back when I discussed "Disabled channels", I mentioned that I favor disabling the DFS channels (52 through 144) in the 5 GHz band.  I didn't explain why.  I recommend avoiding DFS channels because some older Wi-Fi devices don't support them and because DFS channels must be vacated by Wi-Fi if Doppler RADAR is detected.  I like stability and predictability (which might make you ask, "then why the frick are you working in Wi-Fi?!?"), therefore I dislike using DFS channels unless absolutely necessary.

When DFS channels are disabled, that leaves up to nine non-overlapping 5 GHz channels.  But that's assuming that a 20 MHz channel width is used.  I endorsed the use of 20 MHz wide channels in 5 GHz above, because narrower channels tend to work better in environments that see higher density and/or higher mobility.  Narrow channels help high density because they allow more APs to be deployed without causing CCI.  Mobility is helped by narrow channels because a narrower channel allows for data to be successful at a lower signal-to-noise ratio (SNR), and mobile users tend to find themselves in unpredictable scenarios where SNR ends up being lower than expected.

In some cases, I endorse having a 5 GHz AP radio limit that is less than nine when doing a walking Wi-Fi assessment.  I sometimes use a four AP radio limit in 5 GHz because there are only four non-overlapping 5 GHz channels when DFS is disabled and 40 MHz wide channels are used.  (802.11n, 802.11ac Wave 1 and 802.11ac Wave 2 all support 40 MHz wide channels.)  What I'm saying is, if I am concerned that the networking folks are going to change the Wi-Fi settings at some point and use 40 MHz wide channels in 5 GHz, then I'll use a four AP radio limit for 5 GHz.

One last note on "what to look for": If you have the time and the tools, you may want to add protocol analysis and/or spectrum analysis to your assessment. 

The title of this blog post is 'Assess Yourself', so I'm trying to stick to tools and methods that are relatively simple and relatively inexpensive.  Enterprise class spectrum analyzers & protocol analyzers are neither inexpensive nor simple.  

That said, protocol analyzers and spectrum analyzers can reveal valuable information about your Wi-Fi network.  And if you're going to use them, you should know what to look for.

Spectrum analyzers are useful for identifying and locating RF interference sources.  If you have a spectrum analyzer, you may want to "sweep" your location to see if any problematic RF activity shows up.

Protocol analyzers are useful for analyzing Wi-Fi connections, devices and, most importantly, performance via the capture of Wi-Fi frames (commonly called "packets").  Capturing traffic from a variety of Wi-Fi users' devices can be enlightening.  Sometimes you'll find patterns that can inform your decisions on Wi-Fi configuration settings and assessment requirements.  Data rates and Retries tend to be especially illuminating when using a protocol analyzer.

Where to walk

Alright, we are in the home stretch.  Last thing to talk about is where to walk when assessing Wi-Fi.

I am going to start right off by saying something that might be controversial: you don't need to walk everywhere.

The reality is that you're never going to be able to catch everything during a Wi-Fi assessment.  Wi-Fi is too fluid a technology for that.  The wireless communication environment changes all the time.  Every new device that turns on; every time a door closes; every time someone stands up, sits down or walks, the Wi-Fi environment changes.

At some point walking gets redundant.  You're reading a Wi-Fi signal from the same APs through the same walls over and over.  The goal should be to assess Wi-Fi signal propagation.  That always requires walking, but it rarely requires walking everywhere.

-Edges: I recommend walking around the physical edges of your building(s).  The goal here is to identify neighbors' Wi-Fi networks that could interfere with yours.

-Open spaces: Walking through and around open stairwells, large auditoriums, conference halls and the like is a must.  Wi-Fi can propagate over unexpectedly long distances in some cases, but not all.  It helps to know what type of coverage overlap issues you might be up against.

-High density areas: If a lot of users are going to be somewhere, I want a lot of APs covering those users.

-Walking areas: Walking areas are often the same places as open spaces, but walking areas also include hallways and aisles.  The goal here is to look for coverage overlap, so that users will (hopefully) be able to stay connected to Wi-Fi while they walk.

-Near heavy objects: Concrete and steel (along with other heavy metals) are known to block and/or reflect radio waves.  Glass, tile and other materials can sometimes do the same.  Walking near those heavy objects during a Wi-Fi assessment may allow you to identify reflection, signal loss or other troublesome RF behaviors.

If an area doesn't fit into one of those five categories, ask yourself, "Why am I walking here?"  If you have a good reason, then keep at it.  Usually, areas that don't fit into one of those five categories aren't worth the walk.

Corrections & Amplifications

(There aren't necessarily any corrections here.  I'm a Wall Street Journal fanboy, so I just want to use their famous phrase.)

I've gone through the basics of a Wi-Fi assessment.  I just want to add a few notes.

These recommendations are for optimizing consistency and stability.  These settings are appropriate for the vast majority of enterprise Wi-Fi deployments.

What that implies is that these are NOT recommended settings for blowing out a throughput test.  If you need to blow out a throughput test and you don't mind inconsistent/unstable Wi-Fi, then do the opposite of what I'm suggesting.  Use DFS.  Use wider channels.  Require a higher minimum RSSI.  Require more APs per channel.  Allow for lower AP transmit power (thus allowing more APs to be deployed without being able to tell that there's CCI.)  Remember, my suggestions are for working Wi-Fi.  That's very different than a throughput test.  In fact, they often run opposite to one another.

There are also EXCEPTIONS to typical enterprise Wi-Fi, where my suggested settings may be inappropriate.

How do you know if you're an exception?  Chances are you're an exception if you have an extremely low density of Wi-Fi devices (as in, five devices or fewer) and you need to do large file transfers to/from your Wi-Fi devices.  You're almost certainly an exception if you're setting up Wi-Fi in a large sports and/or concert venue.  (In that case, you just need to use DFS, allow for a large number of overlapping 5 GHz AP radios, and then stick to the rest of the recommendations.)  You may also be an exception if you have an unusual type of Wi-Fi device that is essential for business.  (Those cases are far less common than they once were, as many single-function devices have been replaced by tablets and/or smartphones.)

Now, get out there and Assess Yourself!

***
If you like my blog, you can support it by shopping through my Amazon link.  You can also donate Bitcoin to 19SV1CVi6voRXPVMFrYrSGJzZGCA5cvQ7Q or to my QR code:













Twitter: @Ben_SniffWiFi

ben at sniffwifi dot com

Comments

  1. What a breathe of fresh air! Cutting though the bullshit that I hear all the time, and actually confronting reality and the true world situation that we all face. I don't know how many times I try and talk to "experts" in the field, and hear the same old mantra of scanning, etc etc.. when in reality common sense and a good understand of what the surrounds are made of, and what can interact with WiFi can be of more use than a pretty colour printed chart that makes no sense when you go back, install the gear and things don't quite work the way they should! Well done!

    ReplyDelete
    Replies
    1. Thank you, Gordon. I think we see eye-to-eye.

      Delete
  2. This comment has been removed by a blog administrator.

    ReplyDelete

Post a Comment

Popular posts from this blog

Spectrum Deception

Free Sniffing in Windows! (Kind Of)

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