Messing with meshing

What would us geeks do without IRC...
Today it started a discussion about at least attempting to start a meshed wireless network in Gainesville.

Meshed? Ad-Hoc? OLSR? Potatoes?! The terminology involved does take some time to get used to but the underlying principles are not that complicated. What it essentially boils down to is creating a wireless network with standard equipment such as 802.11b that basically everybody uses when not being connected via actually wires. Traditionally you have access points which handle several clients and your wireless network is limited to the reach of your access point. With such a meshed network, everbody puts their wireless network card into ad-hoc mode and with a special routing software acts as a node in a network that can talk to every other node. The point is that when more and more devices get added a cloud develops that moves outward and spans a wider area.
If there is a node on Main Street and you live on 2nd Street and become a node yourself you extend the network outward so that now people on, say, 4th Street can use it too.

Many businesses already offer free wifi for their customer and there is already the campus network, so why bother? Because we can. It's a neat experiment and it wouldn't really hurt us if it failed. In the beginning this will probably not be interesting for the average end user since it does require using the OLSR daemon on every client. It would be possible to rely on a technique that also allows for the classic client/ap approach but it really defeats the whole idea of building a growing cloud since most people would probably not make the effort to even find the daemon. (However later on specific nodes might bridge to a regular AP where many people pass through). I'm currently in favour of OLSR since it is the protocol that has shown in actual working clouds that it can build meshed networks of enormous size.

If I get the consent of everybody involved I plan to take the first step by installing a node at the CMC within the next few weeks. The idea is to use a Linksys WRT54 or WAG54 router and flash it with the freifunk firmware, as well as getting a decent home made antenna on the roof. This should first of all serve as a proof of concept and possibility to try things out and improve on antenna design and the like. This node will however not be connected to the public internet since the ISP doesn't allow the usage for this connection and there is just no way around that. Hopefully, at some point the owner of a node will be willing to share his connection but understandibly this might require the forming of a group of several consenting members to not put the burden on a single person. Some people might argue that this will invite abuse by people who are trying to do bad stuff undiscovered through this network. However, following this logic basically no business could be offering free wireless to their patrons since any evildoer could be wardriving in the parking lot. Furthermore, my apartment is surrounded by other students and my wireless card picks up 21 cells including 3 without encryption. However, the Linksys devices have such limited resources that keeping logs is pretty much physically impossible. So all of that should not be such a big deal, even though one should watch out for noisy people -- there is no protocol encryption.

This is probably the solution that will get the least support from legislators as well as businesses since they can't really control the network, which is a factor in making it the most interesting, in my opinion. Interestingly enough this is probably one of the cheapest ways to getting a large scale network running. What is definitely missing is a good name (preferably usable not only as a domainname but also for the ESSID) and lots of people. Strangely enough, the site for the picopeering agreement doesn't seem to exist anymore, I'll add this when I find it.

OK, if anybody reads this the critique is probably going to be pretty harsh. /me ducks :-)