So one thing that you need to understand is that running a link test pushes data to all clients thus producing optimal MU-MIMO results... but in the real world, not every person is going to be using their radio at the same time. You might have specific groups with specific clients that use a lot more data. That group might be reaching 100% frame utilization, but others have lower utilization, so your 450m still could have plenty of capacity overall.
There's a great post HERE where someone basically asks your question and there's some great responses.
Another thing to try to understand is why you have some SMs that are not able to group... you may want to check these guys first. If you can eliminate those as problems, it may help your overall efficiency.
Hi Mario, there are specific angular separation requirements for SM's requesting data simultaneously. Minimum separation of group members is ~6.5 degrees. The PMP450m works optimally when the SM's are spread across the coverage area and there is a large amount of data flowing through the network. The best way to determine which devices are not groupable is to use our LINKPlanner design tool. Once the SM's are loaded into the tool, there is an option to highlight SM's that are ungroupable.
Also, your number of contention slots is very high. Something closer to 4 would be a better value.
The SM will optimize their own Tx Power to maximize their throughput. That is the ATPC feature at work. They will do this on their own, and there is no user intervention required. Do you have any specific questions on a specific SM or deployment scenario? Maybe we can dive into an example more closely.
If you're looking to set the ATPC target, then I would agree with Eric's assessment... if you set it to -60, and you have relatively low noise in the band, you'll be able to achieve the highest modulation modes. If the noise floor is causing issues with getting to or staying in high modulation, then you can adjust that ATPC target to -57 or -55 to overcome the noise.