My WLAN notes: Cisco AP1832 uses non-standard MCS

Problem description

Affected are clients not far from the Access Point Cisco AP1832, which are using time sensitive applications, for example, VoWLAN.

Indirect symptom is very high jitter of IP packets and time latency of a ping up to 100ms

Troubleshooting

In our case the distance between the client and the AP was about 3m without any obstacle.

As I made a troubleshooting, I noticed the retry rate in the downstream is incredibly high. The average value of the retry rate was about 16%! Sometimes it jumps up to 40%.

It is interesting to note that the retry rate is practically independent of number of transmitted frames. Every second about 4-5 frames are repeated.

DOWNLINK

UPLINK

The reason of retries is the Cisco AP1832 use non-standard MCS to transmit frames.

The access Point is permanently trying to send frames with the MCS9. Just to remind this modulation is not allowed for 20 MHz cannel 1SS and 2SS. Considering the fact that the AP1832 is 3x3:2, it should never use this modulation for 20 MHz channel.

From the next picture you can see that every time the access point tries to send a frame with MCS9 modulation it gets no acknowledgment from the client. The client simply cannot decode this frame. So, the access point has to retransmit this frame with the MSC8.

Conclusion

Cisco created the new bug # CSCvv32690. This bug will be fixed on 8.10 version.

Affected is the latest recommended code 8.5.161.0. I also checked old capture files and saw the same issue in the 8.5.131 version. Probably 8.5 train is affected

There are two temporary solutions to solve this problem.

disable the MCS9
double Wi-Fi channels

This issue is not easy to catch. In case a Wi-Fi adapter cannot decode this modulation, it can not capture this frame. The only sign you will see will be a gap of sequence numbers.

To able to see this issue you should capture frames with a Cisco access point in sniffer mode.

I have not been tested any other model of Cisco access point.