Forum

Does anyone have any advice for someone experiencing random client disconnects in a dense Cisco CAPWAP/LWAPP AP deployment?

Here's some history:

The site survey for this my hospital environment was very very poor and I have inherited the mess.

The 400+ APs used to all have static channel and power assignments with the 2.4 power set to ~12mW across the board.

I have since moved to RRM using 5.2.178 code with tweaked TPC and DCA sensitivity. I'm not seeing wild DCA changes at all, but I have observed TPC to set many APs to very low power levels. (6mw,3mw,2mW, and even 1mW)

This has led me to believe that there's just too many APs in some areas, which has got me thinking about how clients would handle this. I fee like client are maybe getting confused or are unable to handle the excess of signal bouncing all around.

In some areas I have disabled the 2.4 radios on APs that were set to the abnormally low power levels and then forced RRM to recalculate. After doing so I see more normal values. I'm going to keep trying this but I would sure appreciate any advice.

The 5GHz network is looking very smooth and normal, but I know that is because of the decreased range of 5GHz. Just another reason for me to believe that I have too many 2.4 radios too close together.

The techs I work with will scream about "dead zones" but when I measure the signal everything appears normal. Does this not indicate multipath and co-channel interference?

If you have a latency-sensitive application that is disconnecting, I think you've got a basic co-channel interference and low SNR problem. As everyone else has said, get a spectrum analyzer (or protocol analyzer) and look at your noise floor. Find out if your SNR is anywhere near 20 dB. If not, you probably have a load of interference, causing retries, causing your application to drop the session. I think everyone will agree that you must redesign. Everything rides on proper design. You also need tools.