Can anyone shed some light on the meaning of some of the sounding statictics on the AP.
The below screengrab is from our first 450m AP.
I'm interested in the highlighted entries that seem to indicate sounding faults.
What do the NULLING_SNR and CHANNEL_DISTORTION entries show?
We currently have just 36SM's on this AP, with modulations all 6x or above.
Link capacity test to all VC's gives 324Mbps at 96% efficiency.
However, we're getting 88% distribution in the one group.
What's everyone else seeing?
Is there a way to further diversify the groupings?
Hi,
The VC's with the soundingFaults logged look good at the point you captured the page. They have 'valid' spatial frequency (SF) measurements and they are 'tracking' in terms of the sounding algorithm which keeps track of the VC sounding state at the AP. Sounding is the process whereby the AP works out which VCs it can group for MUMIMO purposes. The channel distortion and nullingSNR all look good as well. I expect the faults were latched at some point previously, I'm not sure of whether there is a timeout on this flag. If the status had dropped to INVALID/ASSESSING then the fault would be current. Perhaps you clould try clearing the statistics and see if the faults re-occur.
I'm not sure what you mean by 'diversify the groupings'. The table in the link test to multiple VCs refers to group SIZE not a specific group of VCs. The actual VCs in the group will be different every TDD (2.5ms). The grouping algorithm is looking for VCs with an SF spacing of around 146. It will be looking for all possible combinations of the 36 VCs you have connected to build up groups whilst making sure all VCs get an equal chance to pass data during the test. A group size of 5 is good and gives a notional 5 times improvement on throughput as it can talk to 5 VCs in the same downlink slot.
We hope to bring out some white papers and much improved MUMIMO statistics in the near future which should make these things much clearer. At the moment it's more useful to the engineers who were developing the code, sorry.
you may also want to check out the earlier discussion in
community.cambiumnetworks.com/t5/PMP-450/PMP450M-Grouping-Question/m-p/61580
regards,
Andrew.
Hi Andrew
Thanks, the errors did clear on the Sounding Statistics tab so all looks good on that.
I had obviously misunderstood the grouping stats but after reading your post it is clear now.
So achieving group sizes of 4,5 or 6 are what we want as this represents a 4,5 or 6 times improvement on throughput over a non MUMIMO AP?
Look forward to some white papers to go through this in more detail.
Thanks
colin
Hi Colin,
Glad that is clearer now.
The AP may only acheive 6x DL rate when in MUMIMO even if it got 8x in SUMIMO as the 7 beams do raise the effective noise floor. So we don't promise the full 4, 5 or 6 times improvement however it should be at least 6/8ths of that.
There is a radio button in the link test to multiple VCs page that lets you turn off MUMIMO during the test so you can compare SUMIMO vs. MUMIMO performance. We are always interested to hear real world deployment results !
thanks,
Andrew.
