[Babel-users] Restarting MeshPoint – seeking advice on routing for crisis/disaster scenarios

[Henning Rogge]

On Fri, Jan 2, 2026 at 3:08 PM Juliusz Chroboczek <jch at irif.fr> wrote:
>
> That's orthogonal to media access.  The community networks I had the
> chance to look at back at the beginning of the century had serious issues
> with bufferbloat and fairness -- a single well-positioned chatty node
> could dramatically increase latency for everyone else.

Not completely... CSMA/CA works really badly if the channel
utilization grows beyond a certain point... even if all nodes locally
see (incoming/outgoing) only reasonable amounts of traffic.

> I'm sure we could do much better nowadays, with all the experience we have
> with fq-codel and CAKE.  (Miss you, Dave.)

The 802.11 stack of Linux provides numbers on how much the channel was
utilized, which should include overheard traffic between two other
nodes. Maybe this could be used with codel/cake ideas to get an idea
how much local traffic can be sent... but even then hidden stations
are a difficult issue. Two senders might overload a receiver hearing
both of them without ever knowing there is an issue.

Henning Rogge