Hi,
about WifiMacQueues, please check again the group, I believe there is something.
About the queue sizes, that is the most critical point of your work.
First and foremost, check
Little's Law. It talks about average number of users (or packets in our case). However one should first question: how do I
measure the queue size.
Second point (also crucial). How many packets do you expect to have in the queue ?
The system may be congested (queues growing) or non-congested (queues mostly empty). We don't care about congested state (there is no average queue length if the queue is growing).
In the normal state you have to measure when the queue changes its status (enqueue or dequeue) and average its size considering the time it spent with a given number of items.
If you try to poke the queue and see its size, and then average over the number of observed queue size, then your sampling period should be higher than the one set by the
Nyquist–Shannon sampling theorem.
Since calculating the harmonic principals of the queue / dequeue process is a bit hard (and the queue size changes is a step function anyway), a rule of thumb is: at least 10 samples for each "change" in the queue size.
How much a packet stay in the queue normally? Well, a minimum time is the Tx time for the packet. Do your math, but I'd set the sampling period to at least 1 microsecond. Higher than 20 microsecond, and you risk to not see anything.
You did set your to 10000 microseconds. Enough said.
Now, why I didn't answer you immediately ?
Well, because the answer (this answer) is trivial.
T.