Monday, July 9, 2018

A Healthcare WLAN build from start to finish - and why it should be built as designed

In March of 2015, I did a blogpost “5GHz WLAN Site Survey AP power settings - What you want, don't want, and don't care about.”  In this post, our goal was to find out the best minimum and maximum transmit power setting of a particular access point’s 5 GHz radio.  That post can be found here: http://justdowifi.blogspot.com/2015/03/5ghz-wlan-site-survey-ap-power-settings_7.html

We needed to determine the Wi-Fi coverage area of the access point at a particular transmit power that was going to be deployed in a building on a large hospital campus.  This coverage area data was used to model the WLAN using Ekahau’s Site Survey software.  In the blog post from 2015, we determined that the min/max transmit power would be 8 dBm and 14 dBm, respectively, and would provide the following -67 dBm cell coverage:



When modeling a WLAN with Ekahau’s ESS software, we need to choose a wall type.  In order to properly determine wall type/attenuation value, you need to measure the attenuation of the walls/doors/floors, etc.  This technique is taught in the Ekahau ECSE class and has probably been explained in a few WLAN Engineer blogs.  In the past I used a battery powered access point, but now I use an Odroid for that job.  More on that Odroid can be found here: http://www.morefrag.com/odroid/Odroid%20WLPC%20Excercises%20%5BFinal%5D.pdf  Other WLAN Designers/Engineers are using a new and improved single board computer, which can be found here: http://www.wlanpi.com/
Basically, you need a signal source – in my case, the battery operated Odroid and something to read the signal strength on the other side of the wall.  My wall measuring kit contains an Odroid single board computer, a Leica laser measuring tool to get accurate distances, a Netscout Aircheck G2 to read the signal strength, and of course, a clipboard with the floor plans on them for me to write on.  I bought the zipper case, the foam, etcetera, and made my own carrying case with places for everything to go so that I would notice when I forgot to put something back in its place.  I find it is the easiest way to not leave some of my tools behind.  Here’s my kit:

Use the signal source and signal strength meter to measure the RF attenuation in free space (about 20-25 feet apart) and then with a wall between the source and meter.  Same goes when measuring floor attenuation – bring the meter to the room beneath you and read the signal strength.  You can do the math to figure out how much attenuation is in the wall/floor and that information can be used when modeling and designing your WLAN.  This is outlined below.

Place your signal source in an office and walk through the open door into the hallway and measure the signal with your meter.  For this example, let’s say your signal meter reads -56 dBm.  Now move your signal source to the other side of the room, away from the door, and walk back to the hallway, closing the door behind you.  While standing in the hallway, opposite from the signal source, read your meter.  If your meter now reads -59 dBm, you know that your wall has 3 decibels of attenuation.  You may find a small tripod works well for consistency when positioning your RF source.  The image shamelessly boosted from one of Devin Akin’s presentations.



You will need to annotate the wall attenuation values however you see fit.  This information will be used later, so standardize how you do it so you’ll remember what your symbols mean.  When you feel you know the attenuation value of the different kinds of walls, you can then use the data to model your WLAN design.  If you have CAD files, you can import them for your project and you can assign attenuation values of the walls at that time and always modify as needed afterwards.


We looked at the construction of the facility and determined the wall types – most all of the walls were 3 dB since this was a wing with patient rooms and construction was similar throughout.

In healthcare, wall measurements can be tricky.  Most every hospital I have designed Wi-Fi for has been under some sort of renovation at that time.  Imaging departments, Operating rooms and other areas must not be overlooked when measuring wall types – assuming the entire hospital has 3 dB walls could be a costly mistake.  Many office areas have lead in the walls because the area used to be an imaging department.  During the renovation, the lead was never removed.  MRI areas are usually enclosed with a metal mesh, and block RF from going through.

Since many hospitals use location based WLANs to triangulate assets, we typically design the facility to meet those requirements.  This usually means we start at perimeter of the facility and work our way in when designing the wireless network.  We typically design to a given signal strength (RSSI) for the RTLS solution, combined with capacity calculations from our voice and data requirements.

Many large organizations with thousands of access points never statically employ the power and channel plan that was is in the original WLAN design.  The design often turns out to be “this is what your WLAN could look like if you follow these directions,” however most organizations simply choose a channel width, channel lineup and min/max transmit power and let RRM take care of the channel plan.  This is the case with this design as well.
This WLAN was designed almost three years ago, and is now built and up and running.  The power and channel plan is being controlled by RRM, and the WLAN Validation survey was completed recently with Ekahau ESS and the Sidekick.

Here is an overview of the Network Issues from our 5 GHz, 20 MHz channel width WLAN design.  The purple areas are the areas highlighting the areas where we have overlapping channels.  Notice that most of the channel overlapping occurs in hallways or other open areas where the RF is not attenuated as much.



Here is the same Network Issues view from within ESS of the Validation Survey.  The Validation Survey is a WLAN survey of what is actually built and up and running.



You may be thinking, “hey, wait a minute! Why are they different?”  The answer is “what is deployed is not actually what was designed.”  If we take the original design and modify the channel plan to match what RRM is doing, look what happens:



Look familiar?  Of course it does!  We have now modified our WLAN design to mimic what RRM is going, and they show similar results.  This is actually good news!  This means that we got it right.  From this point on, we are going to compare our design (using RRM’s channel plan) to the validation survey to see if they are similar.  If so, we know we did our design correctly.

Let’s look at a few other views, comparing what was design to what is actually up and running.
The graphic on the left is from the APoS survey, the one on the right is from the Validation Survey.  They match – now that’s good news!


This design had a requirement for what is known as “Secondary Coverage.”  Secondary coverage is a typical requirement for VoWi-Fi handsets so they can roam properly throughout a facility.

Here is our designed Secondary Coverage:


Here is our Validation Survey’s Secondary Coverage: (They’re pretty close).


For those of you who have access to Cisco Prime Infrastructure, you may have heard WLAN Engineers state that we want to see “mostly 3’s and 4’s” on our Cisco Prime maps.  If you are curious how much power is on each access point, see my previous post on Cisco 3802i series power levels.  The post with the power levels can be found here: http://justdowifi.blogspot.com/2018/07/cisco-3802i-series-power-levels.html


Now that we have shown that our implemented wireless network matches our modified designed network by matching the design to the actual implementation, let’s see how a nicely designed network can be implemented incorrectly.  Keep in mind this is the same design we started out with.  Here’s the design at 11 dBm transmit power, UNII bands 1,2 &3, with 20 MHz channels.


First off, we’ll make it better by including more channels in the lineup.  We will incorporate some of the U-NII-1, 2a, 2c&3 channels.  Notice the channel overlap literally disappears!  The purple color indicates channel overlap.



Using the same channel lineup as the original design, I am going to “virtually” login to the WLAN controller and turn on 40 MHz channels.  Look what happens!


Keep in mind that channel overlap is not a good thing.  Reconfiguring a WLAN without actually going into the original design and running some “what-if’s” could negatively affect your wireless network.

Another thing to mention is the channel overlap doesn’t always come from your access points.  If you are in a metropolitan area with older buildings, your 40 MHz and 80 MHz channels might interfere with your neighbors across the alley and the floors above and below you, depending on the age/construction of the building.  If your access points are interfering with your neighbors that likely means their access points are interfering with yours as well. Using a 20 MHz channel plan in a dense environment may actually increase your throughput!



1 comment:

  1. Hello WLAN Ramblings

    Thank you for this cool blog and the very informative content!

    Do you know, is there a way to get Ekahaus channel planner to plan like cisco RRM will assign the channels?
    I want to prevent the behaviour that you had (Planning doesn't match the validation survey).

    Best regards

    ReplyDelete