ePMP 3000: Relationship between channel bandwidth, frame utilization and CPEs

Hi all,

I would like to understand the relationship between channel bandwidth, frame utilization and CPEs, here some assumptions, please confirm if those are correct or wrong, I will appreciate if you add more info.

  1. Does CPEs with higher modulation consumes less frame time? (Ignoring the fact that the AP will be able to talk with two AC compatible CPEs like Force 300-13L at the same time under certain conditions).

  2. Does 40 Mhz deliver more data in less time that 20 Mhz? If that is true, Can I assume that increasing the APs channel bandwidth from 20 Mhz to 40 Mhz will reduce the frame utilization, meaning, it will support more CPEs connected? (Ignoring the packets per second load, which in my case is low, since the plans I offer are 10Mbps and 20Mbps and I have just few customers that use it a lot).


For assumption #1, you are correct. The higher the SNR, the higher the modulation, the better the link quality, and the less time it takes to pass data to and from that particular SM. MU-MIMO makes more demands on the SM group, which is why sometimes an SM may not be in a MU-MIMO group, even though it is near an SM that is in a group.
For assumption #2, 40MHz channel width compared to 20MHz may reduce frame utilization. The thermal noise floor is 3 dB higher on 40MHz compared to 20MHz, and the probability of external interference is higher on 40MHz vs. 20MHz. If we assume the interference level does not change, and the thermal noise thresold increases by 3 dB, then we will lose 3 dB SNR on individual SM’s when moving from 20 MHz to 40MHz. Depending on your SM links, if you don’t have 3 dB of margin to a particular SM, then the SM will reduce it’s modulation level compared to 20 MHz. If your sectors have sufficient margin on SNR, then moving to 40 MHz will reduce frame utilization for the same volume of data passing over the channel. It’s a tricky game…I’ve done audits on networks where reducing the channel bandwidth to 20MHz increased the performance due to thermal noise floor reduction, and reducing the external interference.
Hope this helps.


Probably the cleanest explanation I have seen in a long time on this concept.
The only thing to add is that there is also a 3 to 6db (this depends on many factors) thermal fade every day so just having a 3db margin is not enough to keep the modulation at a steady rate.
Just as an example we have a PTP link with a near clean 40Mhz channel that shifts 5db every day with the worst at about sunset and the best just after sunrise.