Analyzing Frame Utilization and Spatial Utilization Statistics in the PMP 450m AP

If you want to learn more about interpreting the frame utilization and spatial utilization statistics in the PMP 450m AP, and if the sector is reaching capacity, here is a white paper on this topic: 

https://support.cambiumnetworks.com/file/cd32944507c14a2c7ee32a6de301a68af2284e57

1 Like

I have read this whitepaper a dozen times at least, how can I monitor this via SNMP or in some automated fashion? The summary uses pretty vague language I would like to know (preferably with OID’s) which stats to watch and what values to be aware of on each of those stats.

Here are the OIDs:

Description Name OID
Average MU-MIMO Group Size Downlink frUtlLowMumimoDownlinkAvgGroupSize .1.3.6.1.4.1.161.19.3.1.12.1.33.0
Average MU-MIMO Group Size Uplink frUtlLowMumimoUplinkAvgGroupSize .1.3.6.1.4.1.161.19.3.1.12.1.46.0
Multiplexing Gain Downlink frUtlLowMumimoDownlinkMultiplexingGain .1.3.6.1.4.1.161.19.3.1.12.1.32.0
Multiplexing Gain Uplink frUtlLowMumimoUplinkMultiplexingGain .1.3.6.1.4.1.161.19.3.1.12.1.43.0
SU-MIMO Downlink frUtlLowMumimoDownlinkSumimoUtilization .1.3.6.1.4.1.161.19.3.1.12.1.31.0
SU-MIMO Uplink frUtlLowMumimoUplinkSumimoUtilization .1.3.6.1.4.1.161.19.3.1.12.1.42.0
MU-MIMO Downlink frUtlLowMumimoDownlinkMumimoUtilization .1.3.6.1.4.1.161.19.3.1.12.1.30.0
MU-MIMO Uplink frUtlLowMumimoUplinkMumimoUtilization .1.3.6.1.4.1.161.19.3.1.12.1.41.0
ACK Downlink frUtlLowDownlinkAckUtilization .1.3.6.1.4.1.161.19.3.1.12.1.27.0
ACK Uplink frUtlLowUplinkAckUtilization .1.3.6.1.4.1.161.19.3.1.12.1.47.0
MU-MIMO ACK Downlink frUtlLowDownlinkMumimoAckUtilization .1.3.6.1.4.1.161.19.3.1.12.1.48.0
MU-MIMO ACK Uplink frUtlLowUplinkMumimoAckUtilization .1.3.6.1.4.1.161.19.3.1.12.1.49.0
Broadcast and Multicast Downlink frUtlLowDownlinkBcastMcastUtilization .1.3.6.1.4.1.161.19.3.1.12.1.28.0
Total Utilization Downlink frUtlLowMumimoDownlinkSectorUtilization .1.3.6.1.4.1.161.19.3.1.12.1.29.0
Total Utilization Uplink frUtlLowMumimoUplinkSectorUtilization .1.3.6.1.4.1.161.19.3.1.12.1.40.0

Also note that more of these statistics will be added to cnMaestro over time.

1 Like

OK, I can see and pole all of those stats, now what values should I watch for or what flow of values should I watch for. I know if either of the total utilization stats are below about 90%, all is well, but it is also ok if that gets to 100% as long as other values are below a threshold, I want to know what that threshold should be for the other values. I think the average group size and gain values are probably the ones that are part of that calculation. For brevities sake, I would like something along the lines of

if (total utilization >= 90% and (multiplexing gain >=2 or average group size >=3)) then sector is approaching maximum utilization

I would like to know what each of the bolded numbers should be or a range to watch out for each of them, or if other stats are more important than the ones listed here. Even better if I could get a caution (ie start planning) and a warning (ie get new radio to supplement now) level to watch for each important stat.

Monitoring these statistics will be easier once they are better incorporated in cnMaestro, but for now your approach can work.

As a reference for the average group size you can use the maximum spatial multiplexing gain provided by LinkPlanner. Note that LinkPlanner calculates this value assuming that SMs are continuously busy, and it is therefore closer to the average group size reported by the radio. The multiplexing gain reported by the radio is expected to be lower because some SMs are always more active than others. Using this value as reference, you can start to flag the sector if you approach 70-80% of it.

So your equation could be if (total utilization >= 90% and average group size >= 80% x LinkPlanner maximum spatial multiplexing gain) then sector is approaching maximum utilization

As the sector gets busier, the multiplexing gain as reported by the radio will get closer to the average group size. If is is much lower, it typically means you are far from congestion.

If you are monitoring congestion to eventually offload some SMs to other sectors if one gets too busy, then this method is ok. If you are monitoring congestion to understand if you can add more SMs to a sector, then you also need to take into account the spatial aspect of the congestion. If you look at the heatmap in the AP GUI, you can see how some spatial frequencies (which correspond to azimuth ranges) may be congested, but others may not be. Your AP may not overall be congested, but it’s possible some specific directions are and you should add SMs only in the other (not congested) directions.

If the distribution of your SMs leads to a calculated maximum multiplexing gain that is low, you can probably add more SMs in other directions even if the directions you are using are congested.

Once the heatmap is incorporated in cnMaestro, it will be possible to monitor it over a longer period of time (the AP currently shows results only for the last hour).

2 Likes

Is there a way to calculate this from stats on the radio or a formula I can enter that pulls what this should be live rather than having to manually do it in link planner? I understand link planner takes the % dist of group size and calculates a weighted average to get these but I am looking more for a way to find this from info provided on the radio. I am guessing it is somewhere on the table listed under the OID .1.3.6.1.4.1.161.19.3.1.12.4.1. I would rather not have to parse the .15 portion of that oid as they come in strings that are either empty, a single number or a string of numbers separated by commas and having a script that processes that info run periodically is less than ideal for the number of AP’s we have deployed.

To get an estimate of the average group size, you can run a flood test on the sector. The test gives more accurate results if you block all user traffic, otherwise the capacity is affected by which SM is active at the time of the test. If you don’t want to distrupt traffic and enable user traffic while running the flood test, you should run multiple tests to make sure the results are consistent. At the end of the test, the results show the average DL/UL group size.

Here are the OIDs:

OID to configure Link/Flood test mode

linkTestModeBox             .1.3.6.1.4.1.161.19.3.3.18.1.5.0 

(set it to 4 for Flood Test)

OID to start Link/Flood test

linkTestActionBox            .1.3.6.1.4.1.161.19.3.3.18.1.3.0 

(set to 1 to start Flood/Link Test)

OIDs for Avg. Group Size and Slot Distribution

linkTestMVCAverageGroupSizeDL              .1.3.6.1.4.1.161.19.3.3.18.5.14.0
linkTestMVCAverageGroupSizeUL              .1.3.6.1.4.1.161.19.3.3.18.5.15.0
linkTestMVCSlotDistributionDL              .1.3.6.1.4.1.161.19.3.3.18.5.6.0
linkTestMVCSlotDistributionUL              .1.3.6.1.4.1.161.19.3.3.18.5.13.0

The detailed Flood Test Results are available in the following tables:

whispBoxLinkTestMVCResult                  .1.3.6.1.4.1.161.19.3.3.18.5
whispBoxLinkTestMVCTable                   .1.3.6.1.4.1.161.19.3.3.18.6
whispBoxLinkTestMVCSlotGroupingTable       .1.3.6.1.4.1.161.19.3.3.18.7

The relevant ones are the AverageGroupSizeDL/UL

1 Like

@LuciaCambium Any update on when this will be trackable through cnMaestro?

cnMaestro already displays the DL/UL SU-MIMO frame utilization, the DL/UL MU-MIMO frame utilization, and the DL/UL mux gain. You can find them under the performance tab.
In an upcoming release it will additionally display the heatmap, similar to what is currently shown in the 450m AP GUI, but cnMaestro will have data for a longer time frame (the AP only shows data for the last hour).

3 Likes