I wanted to find the best possible channel for my Zigbee network, so I did some research and measured the channel utilisation in ZHA ten times (if you want to do this yourself, the instructions are here). I took the measurements at random over the course of two days. Sometimes there are several hours in between, sometimes about an hour, but never less than 45 minutes, I think.
Here is the raw data. On the left is the channel number, then the average usage, then the sum of all the readings. The top 3 channels (with the lowest utilisation) are marked in green:
I then created this graph from the summed values. Each layer represents a single measurement:
This graph shows the average utilisation from all measurements:
As you can see, there is a “valley” from channel 19 to 24. This was to be expected, as my wifi is currently set to channel 1, which explains the high usage of the lower channels. What I don’t understand is the high usage of the higher channels. I initially chose channel 25 because it was as far away from my wifi channels as possible, but now I know that this was not optimal. I don’t understand why, but that’s the way it is. Based on this evaluation, I will now change my network to channel 22 because, as you can see, it has the lowest utilisation of all the channels.
What I also find interesting here: Some channels apparently have huge jumps in utilisation, namely channels 19, 20, 21, 23. Utilisation here is usually below 10, but sometimes jumps to 40 or more. Channel 18 is constantly alternating between high and low values. This means that if you only take one measurement, you may come to the wrong conclusion. Only after taking several measurements will you have a better understanding of which channels seem to have little utilisation.
I will see what this change actually does and report back here with a conclusion.