Thursday, November 20, 2014

Hallway Design Nightmares Part2: TXPower

One of the effects of the hallway design is that Radio Resource Management (RRM) frequently doesn't work as expected.  It's not that it doesn't work, it's just that hallway designs significantly limit the perspective of how APs see each other, which is primarily how RRM determines what it should be doing.  Given that all the APs in the hall generally see each other at or about the same level, they all have a very similar view of the network.  For illustration purposes, I've mocked up an imaginary residence hall.  A specific hall that I visited recently was build with concrete walls.  We will talk about some of the things I saw, and how they apply to this generic sample floor.



For this example, I've place 3 APs where the black dots are.  Let's also assume that this is the 2nd floor in a 3 story building.





Transmit Power:
In Cisco Unified Wireless, transmit power is set through a function called Transmit Power Control (TPC) and TPC utilizes a threshold to say at what level it should see it's neighbors (minimum of 3). Each AP sends out neighbor discovery messages (NDM) that nearby APs use to measure how well they hear each other. Because of how close the APs are, by default APs will reduce TX power to the lowest value in each band.  This leads to low RSSI and low SNR in the room.


In multi-floor, you may find there is not much inter-floor attenuation.  And since the hallways will likely line up floor to floor you end up with this section of the building where RF propagates easily.  I refer to this as a Low Attenuation Area (LAA).  Because of the LAA, It is pretty common to see APs above and below so strong that they are outside RRMs threshold to adjust for.


For example, a similar building I saw all APs on the current floor and all the APs on the adjacent floors NDM at -50db or stronger with all APs at power level 8.  In the example above, any AP on this floor would see roughly 8 APs at or above -50.


What TAC will say:
So of course, as part of the fallout of any hallway design someone eventually calls TAC.  I'm not trying to call out TAC, they look at the symptoms and make suggestions that can raise the tx power for RRM. But their advise isn't always the most productive. 

"Turn off radios" is the first answer you will get.  TAC looks at the AP neighbor list and sees that RRM has no chance of raising the power.  Turning off radios, effectively reduces the strength of top 3 neighbors the APs sees.  Sounds logical right?  The problem is that because of the LAA, you would have to turn off so many radios you would have large areas of no coverage before TPC will start raising the power.  TAC has the right idea, but we need to fix coverage in the room.

In the example, even if you turned off 4 radios you would still have 3 radios at or above -50, leaving everything at the lowest power level.  If you turned off 5 radios, they would all move to power level 1 as there are not 3 heard neighbors to set the power.


Where you should start for improvement:

Set a minimum TPC value
If you have poor signal in the rooms, I can almost guarantee you will have poor SNR.  The best way to improve the problem of low signal in the room is raising the power by setting a TPC minimum value.  Essentially, we tell TPC to continue setting the power level, but don't go below this threshold.  I prefer this to statically setting the power as you can still have RRM doing the work for you.

I like to start at 8dbm for 2.4Ghz and 11dbm for 5Ghz as a minimum value.  You may have to work up to higher or lower values depending on the wall types and building dimensions.  Don't go below 5dbm or you will be in the same place you are now, low signal and low SNR.  Go test in the rooms to find out where you need to be.


At this point, you will want to circle back to TACs advise on turning off some radios. Why?  Because you just made co-channel worse.  It will be very much like a bell curve.  You will see increased performance until channel contention starts to rise.  We now need to reduce channel contention, probably by turning off some radios.  And the only way to really find out which radios to turn off is to test.  And when you test, testing in the hall will not be anywhere close to what clients see in the room, and the room is where the clients are.  Go. Test. In. The. Room. Measurements in the hall while not worthless don't account for what the client sees at all.


A note on RX-SOP:
So this type of deployment might be a good use case for RX-SOP. In 8.0 we have the ability to adjust the radios threshold where it will demodulate packets.  This can be a Very Dangerous Thing.  It is essentially a wall, 1db too far and your clients will hit the wall.  With the number of APs we likely hear in our LAA, RX-SOP can help improve throughput by tuning out some of the adjacent AP>client traffic.


You need to do some serious testing to make sure you don't hurt the APs ability to hear clients in its cell.  I'm not even kidding.  If you don't feel comfortable tuning RX-SOP, leave it alone or call someone for help.