Throughput to a single SM in Release 24.0.1/24.1

The throughput of the AP in the Canopy line (450, 450i, 450m, 450v APs) is optimized for sector capacity, not necessarily for a single link. This means that in some configurations the AP cannot completely schedule all the resources if only one SM is active, for example during a speed test. Remaining resources can be allocated to other SMs, therefore increasing the sector capacity, but the single link throughput may not show the full capability of the AP.
This happens if the AP is configured with a 5 ms frame length and a BW of 30 MHz or wider.
At 20 MHz the problem may also be seen if a very large or very low duty cycle is used.
In the 450v AP, where BWs wider than 40 MHz are supported, this is also visible with a 2.5 ms frame and a BW of 60 MHz or wider.

Release 24.2, scheduled for later this year, solves the problem by allowing the AP to schedule all the available resources to a single user regardless of the frame and BW configuration.
Before Release 24.2 is available, users can do the following:

  • To check the capacity of the AP (which will also be the capacity of the SM starting with release 24.2), run a speed test to two SMs simultaneously, instead of a single SM. In this case, all resources are fully utilized, and the full capacity of the link is visible.
  • Operate with a 2.5 ms frame if possible. Consideration needs to be given if there are other co-located deployments operating with a 5 ms frame, as interference may become an issue. Also, this solution works for BWs below 60 MHz.
  • Limit the DL modulation to 6x. This almost eliminates retransmissions for link operating at high modulations and helps with keeping all resources scheduled.

So just to clarify, you’re essentially saying to not use 5ms. It affects 30Mhz or larger and 20Mhz sporadically. I feel like this update should include a bold statement that says, In the MAJORITY of cases, 5ms frame rates have a detrimental effect on performance. Cambium strongly recommends switching AP’s to 2.5ms frame rates until FW24.2 is released

Yes, 2.5 ms is more efficient for the single link throughput until release 24.2 is out. However, some operators may still prefer the higher sector capacity they achieve with the 5 ms frame, even if the invidividual link speed may be lower.

1 Like

Some operators are using the speed test prioritization setting whereas others are leaving it disabled. However, I don’t use the speed test server IP address priority and leave it disabled. It isn’t clear how exactly a situation where one customer is running a speed test if they have this speed test server setting set up correctly would affect Link Test results. I have a major problem with the idea of an ISP setting up a speed test server priority and having customers run speed tests often on their server and not even servers outside their network which would indicate actual “upstream” real-world scenarios. While I cannot speak to how this setting might be beneficial in other scenarios the ISP scenario is troubling to me. I also have a major problem with the idea that someone using for example telehealth or even a phone call using VoIP could have their conversation disrupted because someone is running a speed test over a Cambium Networks network and that is prioritized perhaps even sometimes to convince people their network is faster and will remain faster than it really will be in practice. It seems this article touches on the potential effects of the speed test server setting that many equipment manufacturers do not have could affect a different CPE running a link test depending on what link test settings they are using and it isn’t as simple as if you are trying to create the best scenario for a single user doing a speed test as that may be being documented switch to 2.5 ms if you were using 5 ms.