We are seeing about 25-35% of SM's randomly across all of our Medusa's on our network where the modulation is randomly fluctuating from 1X to 8X on the uplink side without any apparent reason. When we see the fluctuations there is no change on the SNR and when running a spectrum analysis we are not seeing any abnormal noise.
We are only seeing these weird modulation bounces on the Medusa's and are not seeing the same issue on the 450i's in our network, even in areas where the i's are in a noisy environment.
I'm on the engineering team for Medusa so let's see if I can help.
I'd like to know if there was a pattern to the modulation drops and whether it occured across several SMs.
For example, they tend to drop if it is noisy or quiet or any particular time of day.
What software version are you using and has the behaviour changed if you have upgraded, or if you haven't upgraded has the behaviour always been there or changed over time.
Finally if you have time to collect some diagnostics files that would be good.
On the Medusa AP go to IP_ADDRESS/field_diags.cgi
This should show a button to download a file via the browser, it may take up to 20 seconds so be sure to let it finish.
If you could email that to me at andrew.rimmer@cambiumnetworks.com
I'll see if there any clues as to the problem and get back to you.
Service may be a bit slower than usual as most of the engineering office (including myself) is out on leave over the Christmas period and not back until the first or second week in January.
If you let the support team know you are in contact with me they will usually include me on the ticket for any updates.
Thanks for the reply Andy. I’ll send you those files tomorrow. I also have a support ticket open with several of the things your asking for in the ticket. I’m also out until the first so no worries…I’m just trying to get a little movement on this issue as my support ticket has been open for about a month with not much progress.
Here is the resolution to this issue for those interested.
Per Cambium Engineer Andy:
“There was another underlying bug where at the end of sending data the SM sent an extra bandwidth request because it thought a packet had gone missing, we had missed 1 packet due to interference probably, but the AP had already allocated a slot in the next frame for this and received it. This spurious bandwidth request then meant the AP ended up allocating a slot for UL and the SM had nothing to send. Normally the SM should be able to tell the AP it’s queue is empty but in this case it didn’t and kept sending packets with zero payload and stale header information. A zero payload packet didn’t satisfy the AP (it was expecting updated header info) so it kept allocating slots and dropped the mod rate as it thought data was going missing.
The AP can now detect this condition (zero payload) and stop allocating slots, this is a fairly simple and robust mechanism for detection and correction while we investigate why the SM didn’t work as expected.”
They have tested this bug fix on a devolpmetal firmware that is currently loaded on one of our Medusa's and we have seen a dramatic improvement and stablized uplink modulation as illustrated below in the before and after of loading the dev firmware. Per Andy this bug fix will be added the the 16.0.0.2 release
It's great to get access to real data to pin down these problems and you were super helpful throughout.
As you mention this fix will be in the 16.0.0.2 release and we hope to have a beta available next week, if anyone is suffering from unexplained mod mode or session drops it may be worth a try.
