450m Throughput and Frame Utilization

Hi all, just curious as to what I should be expecting for total throughput of a 3GHz 450m on a 20mhz channel? Currently I have 15 subs on the 450m and its almost 1:1 with my throughput and frame utilization… throughput is 73 and utilization is 72. We are using a 20mhz channel, as per cambium the 450m boasts 400mbps on a 20mhz channel. I am surprised that I will most likely be capping out at 100mbps. Here are some screenshots

That’s 400mbps of aggregate throughput in a lab setting and assumes all 4 groups are optimized and being fed data at the same time.

So let’s break it down… @ 20MHz you’d see roughly 130mbps aggregate in sector mode assuming 8X-MIMO-B… but there’s a catch, when transmitting in MU-MIMO mode, modulation is typically 1 tier lower due to self interference… so you’re most likely going to be limited to 6X-MIMO-B… so around 100mbps aggregate… and then we split that by the TDD ratio, so assuming the default of 75/25, that leaves you around 75mbps on the downlink and 25mbps on the uplink. So this means that each group, at best, gives you around 75mbps on the downlink. So that’s where they probably get the 400mbps claim… 4x groups (aka 4x multiplexing gain) @ 100mbps each = 400mbps aggregate.

I have a lot of 450m 3GHz AP’s… some of them with 100+ SM’s on them and the best downlink multiplexing gain I’ve seen is 2.4x on 1 AP… but most of the AP’s are around 1.5-2x.

1 Like

Thank you for your response!! So if I understand correctly you have some APs that are achieving 200mbps (150mbps downlink, 50mbps upload) on a 20mhz channel? What are some of the factors/variables that are associated with these achievements?

Again… key word here is aggregate… across the entire AP.

If you’re unfamiliar with how MU-MIMO works and ways to optimize it, I’d suggest that you take a look at the attached Cambium white paper on the subject.

SP_MU-MIMO_MEDUSA_04042019.pdf (936.7 KB)


Thank you for the literature, I will check it out. I did a flood test on the AP and it showed 319 aggregate for throughput.

The way a flood test works is it sends data to all the SM’s at the same time, and it’s also typically not a huge amount of data. While this tool is handy to get an overall feel for how healthy the AP is, it’s more aligned with what can occur under ideal circumstances and not what you’d actually see in the field.

The problem is twofold in that in the real world, 1. data usage by every SM doesn’t typically happen at the same time, and 2. if there are one or more top talkers not spread out optimally across the AP, and instead they’re in one group, they can quickly monopolize that group, and cause high frame utilization… this is especially prevalent if the top talker(s) have less than stellar modulation.

1 Like

Thank you for the insight! I will continue to look over my stats and reference the user manual so I can better understand how my AP is functioning in light of the literature. When I had the 75% frame utilization I had one SM using around 60mbps… as of right now I have 45 mbps down being used with 39% utilization… the 45 mbps is spead out amongst a few SM.

Keep in mind if you have 1 customer using 45mbps and you then have another user in a completely different area that can be spatially grouped, technically they 2 should not cause the overall utilization to shift, so say 1 customer at 45mbps gives 45% utilization if the other person could be 100% grouped because of the spatial separation, the utilization would still be 45% even though now the sector was transmitting 2x45mbps = 90mbps.

That spatial diversity and getting the customers diverse across the sector so they can be grouped, as well as enough users in each spatial area so that traffic is evenly spread across spatial areas throughout the day is what makes frame utilization in MU-MIMO so weirdly difficult to predict. It’s why even when your panel is at 100% frame utilization, your sectors technically not actually saturated everywhere, because if you add a customer during 100% frame utilization in a spatial area that doesn’t have much usage during that frame saturation period, it’ll likely just get grouped in and get clean bandwidth on the frame.

1 Like