SM disassociated: Communication lost

Products ePMP
1 
Sep  2 05:37:54 NePMP_9b83c6 kernel: [106683.560000] SM disassociated from AP[00:04:56:MAC] F=5485 11naht20. Reason: 48 (COMMUNICATION LOST)
Sep  2 05:38:44 NePMP_9b83c6 kernel: [106733.280000] SM associated with AP[00:04:56:MAC]
RSSI:-51
SNR: 39
Downlink Packets Per MCS

MCS 15 - 64-QAM 5/6

1 259 388 (35%)


MCS 14 - 64-QAM 3/4

750 187 (20.9%)


MCS 13 - 64-QAM 2/3

654 131 (18.2%)


MCS 12 - 16-QAM 3/4

359 228 (10%)


MCS 11 - 16-QAM 1/2

262 551 (7.3%)


MCS 10 - QPSK 3/4

121 176 (3.4%)


MCS 9 - QPSK 1/2

29 173 (0.8%)


MCS 7 - 64-QAM 5/6

56 988 (1.6%)


MCS 6 - 64-QAM 3/4

41 400 (1.2%)


MCS 5 - 64-QAM 2/3

27 802 (0.8%)


MCS 4 - 16-QAM 3/4

12 040 (0.3%)


MCS 3 - 16-QAM 1/2

7 028 (0.2%)


MCS 2 - QPSK 3/4

9 514 (0.3%)


MCS 1 - QPSK 1/2

7 304 (0.2%)


MCS 0 - BPSK 1/2

0 (0%)


Uplink Packets Per MCS

MCS 15 - 64-QAM 5/6

130 404 (2.8%)


MCS 14 - 64-QAM 3/4

233 516 (5%)


MCS 13 - 64-QAM 2/3

497 724 (10.6%)


MCS 12 - 16-QAM 3/4

2 433 000 (51.9%)


MCS 11 - 16-QAM 1/2

836 584 (17.9%)


MCS 10 - QPSK 3/4

108 428 (2.3%)


MCS 9 - QPSK 1/2

12 123 (0.3%)


MCS 7 - 64-QAM 5/6

83 408 (1.8%)


MCS 6 - 64-QAM 3/4

123 492 (2.6%)


MCS 5 - 64-QAM 2/3

126 937 (2.7%)


MCS 4 - 16-QAM 3/4

85 369 (1.8%)


MCS 3 - 16-QAM 1/2

8 929 (0.2%)


MCS 2 - QPSK 3/4

3 353 (0.1%)


MCS 1 - QPSK 1/2

524 (0%)


MCS 0 - BPSK 1/2

0 (0%)

 Version 4.4.2 on both AP and station. This happens ~each 1 hour. eDetect e-detects nothing. 

So, the question is: why the communication is lost? 

2

You have some type of interference. With a -51dbm RSSI and no interference, you should be at MCS 15 99+% of the tilme for download. Your MCS rates are a little scattered for both dl and ul, which is sign of interference. I would lock your MCS upload down to MCS 12 or even 11 to help with throughput and stability. May not help with disconnects, but will help with overall performance. 

eDetect only shows WiFi signals on same channel and same channel width as your AP/SM. It will not show any other type of interference. The scattered MCS rates shows us the interference that eDetect does not see. 

4 Likes
3
@beeonline wrote:
eDetect e-detects nothing. 

As CWB says; "eDetect only shows WiFi signals on same channel and same
channel width as your AP/SM. It will not show any other type of interference"

Basically, if (for example) your AP is on 5750 at 40Mhz wide, there can be an interferer at 5750 and 20Mhz wide - yet it WILL be interfering - but eDetect won't report it.  If (for example) your AP is on 5750, and there is an interferer at 5745Mhz - yet it WILL be interfering - but eDetect won't report it.

So, in lieu of any way of seeing interferers (such as a simple 802.11 site survey for example) you have to reply on indirect evidence - such as MCS rates bouncing around and lower than expected, and such as higher percentages of retry and dropped packets.

As CWB suggests, it can be helpful (if you can't simply find a cleaner channel) in interference situations like yours, you could try setting the MCS level to a maxium of 12, which will get a majority of packets through on the first try.  That may help the experience.

2 Likes
4

Thank you guys! Well, in my area I am 100% sure there is interference on each bloody freq :)

Will try lowering mcs.

1 Like