We currently have one epmp site with a few customers, and I've noticed something really odd: the MCS jumps all over the place. One unit I was watching today bounced from 3 to 11 in the span of a few seconds, and his uplink performance looks like this.
Uplink Packets Per MCS
MCS 15 - 64-QAM 5/6
608411(2.1%)
MCS 14 - 64-QAM 3/4
619508(2.1%)
MCS 13 - 64-QAM 2/3
642526(2.2%)
MCS 12 - 16-QAM 3/4
1240606(4.2%)
MCS 11 - 16-QAM 1/2
5495879(18.7%)
MCS 10 - QPSK 3/4
5984066(20.4%)
MCS 9 - QPSK 1/2
3634458(12.4%)
MCS 7 - 64-QAM 5/6
591854(2%)
MCS 6 - 64-QAM 3/4
617699(2.1%)
MCS 5 - 64-QAM 2/3
779013(2.7%)
MCS 4 - 16-QAM 3/4
2964606(10.1%)
MCS 3 - 16-QAM 1/2
2428175(8.3%)
MCS 2 - QPSK 3/4
1956590(6.7%)
MCS 1 - QPSK 1/2
1758662(6%)
Is this normal?
This customer is pretty close, using a dish, and running outside of DFS (our epmp DFS experience has been less than ideal, but that's another thread).
Distance from AP
1.956 miles
Operating Frequency
5740 MHz
Operating Channel Bandwidth
20 MHz
Downlink RSSI
-62 dBm
Downlink SNR
35 dB
Transmitter Output Power
6 dBm
Uplink MCS
MCS 11
Downlink MCS
MCS 15
Power Control Mode from AP
Closed Loop
Can anyone explain what might be happening? I don't see anything like this with our ubiquiti gear, and the 450 stuff doesn't appear to have this information.
Typically if your MCS levels are jumping around alot and downshifting to lower modulation levels its because of intermittent interference. If you run ACS and/or spectrum analysis, how does the channel that you're using look?
"The transmitter output power is at 6 dBm. Is this intended?"
The epmp stuff is handling the power output based on reaching a target RSSI. I had been suspicious of the low output, but the 450, and our old 100 gear worked great, so I expect this to perform as well.
"which antenna...?"
We have RF elements 17dbi on this site; the model with back-shielding.
Later today, I will update the settings on one of the problem SMs, and make sure it's output is set higher. I'll update with my findings after letting it run for a day.
At the AP, check the Subscriber Module Target Receive Level set. Default is -60dBm. Adjusting this level will make your AP instruct the SMs to change their TX power accordingly to hit the configured level.
At the SM, check the Max Tx Power mode is set to Auto, not Manual. Setting it to Auto allows the AP to control the SM tx power.
"The transmitter output power is at 6 dBm. Is this intended?"
Well, after a few different trials, I can confirm that the low transmit on this unit is not intended. I have manually set it to 17, but it will not go above 8. This is pretty irritating.
With the last hour, I did a spectrum analysis from this AP, and ensured that there is nothing in 5305-5335, then set a 20mhz channel centered on 5320.
One SM, which is 1.9 miles away, connects with -65, and transmit of 7db (though I manually set it to 17). This one shows ~50% downlink traffic on MCS 1, and almost 60% of uplink in MCS 9 and 10.
The other one on this AP is 6.3 miles away, and connects with -76. Its transmit is only 13db, but I manually set it to 17. This one has ~60% of downlink traffic in MCS 1, and ~70% of the uplink in MCS 11, with nearly 10% in MCS 0 and 1.
Another oddity is when I run link tests, both show large amounts of traffic in the higher MCS levels.
The power control mechanism will instruct the SM to power up or down based on this value. So even if the SM could possibly go higher in power, the AP will instruct it to keep the tx power at a level that meets the receive level at -68dBm at the AP. You can squeeze more power of the SMs by adjusting this value to a higher value than -68dBm.
Are these FCC units? For the low 5G band, FCC units are limited to 13dBm. You can get higher power output in the mid and high 5G bands.
So, I had typed up some griping, but then I re-updated the AP to have both banks populated with 2.6. Now, I have moved the AP down to a quiet channel on the lower band (5.18), and set everything on the SMs to auto. For the sake of testing, I would rather keep this AP running further away from the 5.7 band.
The closer customer has 13 for the transmit power, and the further is up at 20. I don't think I can manage a much better uplink signal, because the further SM is -68/-69. I'm pleased that the firmware re-update fixed the signal strengths.
So far, I have noticed that there is still almost 10% of downlink in MCS 1.
All control traffic (control messages between the AP and SM over the wireless interface) is sent using MCS1 but 10% seems high. It may be possible the SM could be picking up some intermittent interference. Glad things are working better now.
I saw that in the firmware about control traffic and lower MCS values, but I wasn't positive what it meant.
If that's what it designates, I'm guessing it's because I've been doing so many things while there was no real traffic being generated (hence the reason bandwidth tests are in the higher modulations).
Many thanks for the insights. I hope my tests work well, and I can apply these settings to more highly populated units with similar issues.
I saw that in the firmware about control traffic and lower MCS values, but I wasn't positive what it meant.
If that's what it designates, I'm guessing it's because I've been doing so many things while there was no real traffic being generated (hence the reason bandwidth tests are in the higher modulations).
Many thanks for the insights. I hope my tests work well, and I can apply these settings to more highly populated units with similar issues.
if this cpe isn't doing much, you'll see a large % at MCS1, the control messages are sent at that code rate by default. when it gets busy you'll see that % shrink. attached is a screen shot from one of our EPMP backhauls thats running as i expect it to and counting some MCS 1 traffic.
this backhaul is getting major interfences and isn't the greatest RSSI, but its still moving 100 meg downlink, and 20 to 25 uplink.
I can see by the picture that you have locked both the upload and download MCS's to 13 and 11... does it work better when it is not changing modulation schemes?
Most of the time the default modulation scheme does the best job peaking throughput. Sometimes with coband interference locking the modulations to the 2nd most used rate will yield better enduser results and improved voip quality. This particular site had been running side by side with a ubnt rocket only 30mhz away, which isnt enough space so i had to give up some mcs states for quality reasons. This like was installed before we had epmp2000s. A epmp2000 would have been the ideal solution for best results here. I made a story post about using epmp2000s in a ptp situation to fix this exact problem
