tested on 3 identical routers all setup as dumb ap (all have dhcp disabled)
1st i tried on the 2.4ghz then used 5ghz as backhaul.
i did not do any removal or wpad-mini/etc just stock.
did not do any command line.
each router shows 2 mesh ap mac each so afaik they were all meshed.
i swap each router to check if any of them can be the main ap (the one that has the wired link to my lan)
all of them can be used without configuration.
QUESTION:
is this OK? can this be scaled up to say around 8-12 mesh devices?
i do not need any encryption as these are used as public wifi with captive portals.
Throughput will degrade when you add more mesh APs.
The total bandwidth will be divided among local client traffic, rx traffic and tx traffic from other APs.
batman-adv can run on top of a variety of mesh implementations, including 802.11s, ad-hoc (IBSS), and multiple point-to-point links, wired or wireless.
batman-adv is L2 routing protocol, The issue is in L1, as you cannot have multiple stations transmitting at the same time, or there will be a collision. So when a host is transmitting the others around should wait. The more hosts you have the more hosts will have to wait until it is their time to transmit.
It actually scales better than you might at first think.
In a real venue type situation with mesh nodes covering a large area (think 2d or 3d) and each node only seeing its neighbours, the degradation curve tends to become flat after the first few. Latency increases rapidly for those first few nodes then stays pretty much constant. I was pleasantly suprised when I tested with 30 or so nodes in a residential area.
I don't have any actual figures to hand but when things get back to normal more empirical testing is planned.
