Client RSSI and SNR periodically change every 15 minutes

[Gehrig_W]

Hello Cisco WLAN-experts,

I'm investigating a problem on Android tablets connected to this hospital WLAN via 5GHz-Dual-Band-3802i-AP on 5520-WLC running on 8.10.183.0.

The tablets are used for patient interviews in front of kidney transplant.

The users complain about communication interrupts and missing download of updates.

When I look at the client from Prime and WLC, I can see a good Client Signal quality in 5-Ghz-Band:

(WLC-5520-1) >grep include "Signal" "show client detail 80:86:d9:6d:ff:70"
Press any key to continue..
Radio Signal Strength Indicator............ -60 dBm
Signal to Noise Ratio...................... 35 dB

We are using FRA on this WLAN-AP: No 2,4GHz-antenna active

Slot Id ..................................... 0
802.11-abgn Current Channel ................... (112,108)* 40MHz / Cap 160MHz
Allowed Channel List........................... 36,40,44,48,52,56,60,64,100,
........................... 104,108,112,116,120,124,128,
........................... 132,136,140

Slot Id ..................................... 1
802.11a Current Channel ....................... (52,56)* 40MHz / Cap 160MHz
Allowed Channel List........................... 36,40,44,48,52,56,60,64,100,
........................... 104,108,112,116,120,124,128,
........................... 132,136,140

But to my surprise, when I'm digging a little deeper on Prime into Client RSSI and SNR statistics over time, I can see

a permanent variation in Signal Quality and RSSI in a 15-minutes period.

Please have a look at the attached picture.

I have checked other clients on same AP.  Another Android and IPhone XS shows rather stable, reasonable good lines.

Another IPhone14,2 shows also this 15-minutes-periodic changes.

Interesting to see is also, that I cannot grep the bad values via CLi using this commad at the 15-minutes-interval:

(WLC-5520-1) >grep include "Signal" "show client detail 80:86:d9:6d:ff:70"

It always shows the  good figures:

Radio Signal Strength Indicator............ -60 dBm
Signal to Noise Ratio...................... 35 dB

What might cause such a periodic change ?

Thank You for any good tips.

Greetings

Wini

[Leo Laohoo]

---

@Gehrig_W wrote:
The users complain about communication interrupts and missing download of updates.

---

As a test, disable WMM and see if this improve things.

[mrTeigen]

Hi,

Be careful, if you disable WMM you will turn off .11ac and .11n and end up with .11a

Try do an manual SNMP for RSSI and SNR since Prime only updates these values every 3-5 minutes. I have created a Grafana Dashboard (TIG stack) so I can these values + many others every 5-15 seconds (depends on how many clients you have).

bsnMobileStationStatsEntry - .1.3.6.1.4.1.14179.2.1.6.1.#MAC IN DECIMAL#

80:86:d9:6d:ff:70
128.134.217.109.255.112

RSSI
.1.3.6.1.4.1.14179.2.1.6.1.1.128.134.217.109.255.112

SNR
.1.3.6.1.4.1.14179.2.1.6.1.26.128.134.217.109.255.112

Here you will get values more often

[Leo Laohoo]

---

@mrTeigen wrote:
Be careful, if you disable WMM you will turn off .11ac and .11n and end up with .11a

---

I did start with "As a test" because I am trying to determine if this behaviour is due to the MARVAL chipset issue or not.  I am not saying turning off WMM as a solution.  

[mrTeigen]

Yes, I understand that @Leo Laohoo but I thought it was a good idea to mention it since turning it off will create other issues for thousands of other clients while trying to fix one of them, since I see it is joined to a eduroam SSID. 

Then it is maybe better to make a separate test SSID with WMM disabled.

But, back to my answer on getting RSSI and SNR with SNMP. I asked about that because I have seen the same in Prime before, but it looks more like a client issue since every time Prime poll the data it gets 0 RSSI and 0 SNR. This is not live data, you have to use other monitoring solutions to get more updated data.

I would run a debug of the client for a an hours and then uploading it to the Wireless Debug Analyser.

debug client <MACADDRESS>

https://cway.cisco.com/tools/WirelessDebugAnalyzer/

With this tool you will hopefully get more info, maybe the client is disassociating, roaming to a different access-points for a few seconds so Prime does not have info of it, there is a short session-timeout on the SSID etc. This tool has been updated recently and is a fantastic tool for debugging clients.

[Leo Laohoo]

Do not put too much faith in the WiFi Debug Analyzer:  Sometimes it strips away the cause of the problem.  

[Arshad Safrulla]

Can you check which slot of the AP (Slot 0 or slot1) the client is connected to?  Read more about FRA behaviour and what micro/macro cells mean. For starters I would shut down the slot 0 on that AP, re-run the RRM if possible and then check.

Cisco Aironet Series 2800/3800 Access Point Deployment Guide - Client Roaming in a Micro and Macro Cell [Cisco 5500 Series Wireless Controllers] - Cisco

[Gehrig_W]

Hello Cisco-WLAN-experts,

to me it looks like an SNMP-overload-problem between Prime and WLCs or WLAN-APs.

The value "0 DBm" is of course not realistic and can only be interpretes as missing SNMP-Value from WLC/AP as it is appearing in 15 minutes Poll-Interval.

To my surprise the values from other clients on same WLAN-APs apperar to be realistic.

Has anyone a good idea how we can improve SNMP-polling ?

For example, we also send all WLC-logs to the central Prime, but none has time to investigate them on Prime.

Can we simple ommit this traffic to offload burden from Prime ?

Where can i check Prime SNMP polling and timeouts ?

Kind regards

Wini