Hello everyone. I've had a single SM linked up since 10/2015 using ePMP2.4 as a proof of concept. We have been monitoring the RSSI since 10/2015 which has been averaging 66 dBm. The link distance is about 1.85 miles NLOS. The fresnel obstruction is just some treetop foliage mostly near the SM end of the link. We are using a 5Mhz channel size. The SM performance has recently degraded to where the upload speed is now virtually non-existent. I can't even get the web GUI to load completely. The wireless link test from the AP to SM shows Downlink of 11.79Mbps but Uplink only 0.138 Mbps. Somehow I managed to issue a SNMP command a few days ago to reboot the SM, which worked but it did not resolve the issue we're experiencing with the slow uplink. I feel like we may be dealing with interference but not sure how to cope with it. Our AP sector antenna is the factory-spec 90deg slant 45 antenna. It is mounted at 200' on our base station, and aimed without any downtilt. The SM side is a connectorized ePMP w/ a UBNT 2' Rocketdish.
Would aiming the sector towards the ground help reduce the noise the AP is being subjected to? Would using a different sector entirely, maybe the RF elements conical sectors help out with this?
I've looked at the upstream router and I'm fairly certain the Uplink trafic is not saturated at all. I am only making an assumption about interference. In fact, last October, the bit of noise we had wtih 10Mhz channel sizes was resulting in a lot of SM de-reg and reg. The SM is not de-registering at all.
One other thing I should add is that when I reboot the SM, it takes almost 20 - 30 minutes for it to re-register to the AP. I'm fairly positive that there is only 1 channel set to scan on the 5Mhz channel size. I don't believe it should take that long at all. Could it be that the CPU or hardware is failing?
Hi. Well, if the packets are generally getting from the AP to the SM, then the conditions at the SM side are generally OK. However, if the packets are generally NOT getting from the SM back to the AP, then that may indicate that the conditions are generally not OK at the AP side of things.
So, if I was looking for interference... based on what I see for your numbers, I'd be lookin at interference at the AP side.
If it is interference on the AP side, you may be able to help the link by limiting the MCS on the SM side of things... the maximum speed it'll try to communicate back to the AP with. Right now, it's trying to send data back to the AP at MCS15, and that only get's through 2.8% of the time, which means 97% of the time, that needs to retry. If you limit your MCS to MCS9 maybe, then 55% of those packets might get through on the first try.
In theory, reducing (or 'CAPping) the MCS does reduce overall capacity - but in practicality, it might help this SM be able to communicate back to the AP more reliably, and that may actually increase performance.
We made a tower climb yesterday and pointed the sector so that it had about -5 degrees of downtilt. I don't believe it had any positive effect on limiting the noise at the AP. As for capping the MCS on the SM, I will need to perform a truck roll visit to the customer to make this change as I still cannot access the GUI of the SM.
We did perform a spectrum analysis when we took this customer live back in October 2015. We selected the cleanest channel we could find at that time. I will do another spectrum analysis after I do the truck roll so that I can also remove the 5Mhz channel scan list from the SM. I'll post it here as soon as I have the chance. I don't believe I can change channels at the AP right now because I am fairly certain the SM is configured with a very specific 5Mhz channel list.
What if I convert this customer to a PTP link by removing the sector from the tower and replacing it with a Force 200? Would that significantly reduce the chance that the noise would affect the link since it will have a very narrow beamwidth? In regards to converting to a PTP link, how will that affect my"back" sector that is currently relying on GPS sync? (I am currently running Front/Back setup).
Whether or not a force200 will be an improvement over the sector depends entirely on where the interference is coming from. The force dishes have a more narrow pattern than the sector so there is a chance you will be able to avoid the interference but if the interference is in that narrower pattern the force dish will not help. If you have other 2.4Ghz ePMP ap's and you aren't using Sync then you may want to consider using sync.
Have you tried to SSH into the radio ? I know the horribly slow/bloated GUI is hardly usable even when the radios are running great so it is completely unusable when the radio is having connection issues but I've found I can often access them via SSH.
What version of firmware is on the SM and AP ? We had issues with some radios taking 20 or 30 minutes to reregister back in v2.5 and the 2.6.1 seems to have fixed it. The cold weather reboot problem in 2.5'ish would also sometimes cause the radios to lose "some" settings which caused all sorts of knots on my head and massive amounts of hair loss. until I figured out what was going on.
I agree with the advice that everyone is giving. But also, not every problem can be fixed at the AP side of things. Especially if the SM is set to only a single channel in its scan list… you may just need to go to the SM side and change that so that you can then try other channel/width options. As Brubble1 (Barney Rubble?) says, you may be able to SSH into the SM so that’s worth a try… but getting it set so that you can at least test other channels would be at least part of the puzzle.
I was finally able to gather a spectrum analysis from the AP. I have attached it here. We were using 2407Mhz as the selected frequency, and there is a front-back sector in play here so the spectrum analyzer is picking that up from the "back" sector.
I have changed the AP frequency to 2467Mhz and the SM has re-registered. I can get to SSH but not the web gui..