Skip to content

Difference between Minstrel-Ht (linux kernel) and Minstrel-Ht (ns-3) ?

Hi,

I have been reading the Minstrel-Ht source code (as, the algorithm actually implemented in the Linux kernel), and the ns-3 implementation. I noticed a few differences, which I believe it would be good to fix.

The first one is that Minstrel-Ht (Linux) use a 50 ms statistics update duration (and always has), instead of the 100 ms actually implemented in ns-3 since Minstrel-Ht introduction. It's pretty easy to fix, which is why I propose !462 (merged) (but I believe discussion in a real Gitlab Issues is probably best).

The others difference concern some ``recent'' change in the Minstrel-Ht implementation. Since October 2019, the averaging method for the computation of the probability of success for a given transmission rate has also changed (it's not ewma anymore), as stated on https://lore.kernel.org/linux-wireless/20191008171139.96476-2-nbd@nbd.name/

I plan to work on this in a week or so, as it means changing quite a bit of the code for Minstrel-Ht. In the meantime, I'm creating this issue to allow further discussion on these.

Rémy