[Make-wifi-fast] Re: [Bloat] Re: [Codel] [Rpm] Re: Re: [Cake] "Fi-Wi is a new forwarding plane for wireless" - Bob McMahon

[David Lang <david@lang.hm>]

Sebastian Moeller via Bloat wrote:

>> On May 25, 2026, at 14:45, Frantisek Borsik <frantisek.borsik@gmail.com> wrote:
>>
> and still under WiFi-limited scenarios h++ps://test.libreqos.com shows latency 
> excursions into thte 500-900ms range. For my taste that certainly leaves ample 
> room for improvements. Sure, when I get closer to the AP so WiFi throughput 
> exceeds WAN throughput my WAN shaper kicks in and I get decent test results 
> (with no noticeable increase in latency-under-load), but that is a work-around 
> only.

this tells me that the bandwidth estimates on wifi are fixed, when they need to 
be dynamic. (or set much lower)

When you are too far from the wifi, you aren't getting enough throughput to 
trigger the shaping, it's sending it all out to the wifi chip thinking that it's 
operating at the advertised speed.

so what we are needing is a feedback loop from the wifi layer to the shaping 
layer.

Fi-Wi attempts to move all of the protocol to the main cpu to give it the 
visibility, but I'd bet that we can do better with the stats that we have.

If we can (for now) ignore the problem of a rapidly moving station, and look at 
what I will argue is a more typical problem of someone sitting at a desk a good 
way away from the AP, then later moving to a different desk closer to the AP and 
wanting to get the best performance in both cases, we could have a script look 
at the wifi stats/queues/throughput numbers periodically and tweak the cake 
bandwidth limits to stay within range (say sample every minute, with older info 
remaining valid when idle, but decaying fairly quickly as new samples arrive so 
that one bad sample doesn't mess things up too badly, but you adjust to new 
conditions within a relatively small number of samples

it would take a bit of time to get to the right number when conditions change, 
but it would get there.

I recognize that you cannot know the throughput without trying to use it, but 
you may be able to make a guess as to the amount of idle airtime (subject to 
hidden transmitters) and make an estimate as to the available headroom based on 
the throughput in each direction of your recent speeds and airtime used.

David Lang