| Magnus on 8 Feb 2004 03:44:02 -0000 |
|
No it isn't. There was discussion, synonym talk, of setting up a wireless mesh among members of this mailing list on this mailing list. I'm a little vague on why it ever left this mailing list. There were others interested in this who had no interest in Linux, so it does make sense to fork the discussion. My end of it, anyway, didn't take any divergent route until I posted a URL for a new mailman list. The exact discussion I recall, especially with Magnus for the brief time that he and I worked at the same place and lived slightly nearer to each other than we do now, involved the Swarthmore area. I think we're actually closer now, and certainly have a much better shot at reaching each other with fewer hops than when I lived by Ridley Creek State Park. This is probably a good place to start because I know personally more than a few people in the area with a broadband connection who'd be happy to share it[1]. Yeah Swarthmore is probably the best bet for an initial pop. Good concentration of tech savvy people (as you mentioned), some tall buildings to mount a repeater node on, etc. Where I'm at now has good topology but lousy concentration of local tech talent. I don't see why it's necessary to fracture this from the existing PLUG mailing list or, at least, why we can't just use the existing mailing list infrastructure that PLUG maintains. Because some of the people who want to do this are Windoze folks. Your own OS preference is on record as being non-Linux (though not entirely off topic here). Really something like this should be OS agnostic. There are compelling reasons to divorce this subject from the PLUG list. But these are just nodes. There's the relaying hurdles (both political and technical) that we still need to cross for those to really be useful the way we've discussed before. (Unless you've actually got these hooked up together, in which case, great! I see you added me, though "laundromat roof" is a bit of a stretch at this point. It's more like "inside a brick building, third floor".) No, these aren't as yet linked in any way. Some of them aren't even online. These just reflect the locations of people who have expressed to me that they would be interested in putting a local repeater at their location even if they had to dump a few hundred into hardware to get linked up. There is no way we're going to get direct point-to-point links with any of those nodes outside of Swarthmore without dropping a few in between. (Magnus knows what I'm talking about, but for the rest of you: we'd like to see a given node have one omnidirectional antenna and a couple of directional, high-gain antennae connecting to other nodes.
Some nodes would have local Internet access--mine does, for instance--some would just be relays. Addressing within the mesh would be some reserved-for-internal IP range, and you'd be able to carry DHCP leases between the cells. Traffic to other nodes would be the highest possible--ideally, 802.11g. Traffic to the Internet would be whatever you could get from the nearest node with Internet service.) I think 802.11g is going to be very likely with the work that has been done on the Linksys 802.11g routers. The Dell TrueMobile 1184 routers turned out to be a dead end. Everyone that I know of that was working on them has dropped them after making a few bricks out of theirs. The WRT54G from Linksys has already been worked out. You can get one from Microcenter for $99 as of yesterday when I was last there. I should also point out that the roaming protocols that would allow you to carry an IP from AP to AP have not worked out too well in real world trials. A lot of the wireless projects out there are falling back to traditional routing protocols like OSPF and BGP. Since the AP's are pretty static, this isn't too big a deal. But your IP will change as you move from AP to AP if you're mobile. But we've *got* a perfectly good Mailman server already running for PLUG!
I've gotten some more hardware together for infrastructure and as the weather warms up I should be popping up some WRT54G AP's (with custom firmware) around my place in Prospect Park for a local proof of concept. I don't think there is any way I'm reach Swarthmore without some more people between here & there putting AP's up. Preferably with good high roofs. The real challenge of course is the tall trees that are just about everywhere in Swarthmore. For those that haven't seen yet, here are some photos of what we're up against: http://trilug.org/~chrish/gallery/swarthmore Those are taken from the top of the elevator shafts (above the roofline) of some of the highest buildings in Swarthmore. There are a few possible ways around this that I can imagine: 1) 802.16 hits the streets and is affordable enough to set up a base station that covers a good sized chunk of DelCo. When I say "affordable", it's almost certainly going to require either a very generous benefactor or a cooperative effort to put up a base station. This would certainly address latency issues, and provide great bandwidth to town hub nodes. 2) Local businesses start to see the benefit, as Gabe mentioned, and here and there along busy commercial roads we could have enough access points to act as stepping stones along the commercial corridors. From Swarthmore to Prospect Park would be tricky. Hops would be necessary to get you from "downtown" Swarthmore through the thick growth to either Baltimore Pike or MacDade Blvd. Rt. 420 intersects both of these, as well as my road (Chester Pike, aka Rt 13) which is also a major commercial blvd. If the mesh topology can reach down 420 and touch all of these Blvds, there is an opportunity for businesses along these major roads to connect. I can't begin to imagine how many hops it would take to get to downtown Swarthmore from Prospect Park / Norwood along such a route. 3) This isn't something I'm willing to take up right now, but if any of the local towns or Chamber of Commerce type organizations see the potential of such a project, it might be possible to see large swaths of consecutive businesses hardwired to each other, and providing wireless access from their respective roofs. The two most geographically distant nodes would only be one hop apart on such a topology and make it much more practical for towns to connect to one another. Think this is farfetched? Something not unlike this is happening now back in my old stomping grounds in North Carolina (though on a somewhat less ambitious scale). The local businesses are driving this to try to get people to come often and linger. Coffee shops, pubs and restaurants are going to get the most benefit. 4) For the time being, VPN across the public Internet between towns. For the time being I'm more interested in the smaller scale aspects, like how to keep the cost of entry down while still getting lots of bang for the buck. The PC in a plastic box route is just going to cost too much. A modded Linksys router in a plastic box with an external antenna, on the other hand, is downright affordable. Let's prove out the various topologies on a small scale first and then figure out how to get our towns connected to one another. We're only like 3 or 4 miles apart as the crow flies, which with LOS could easily be done with the right hardware. The blasted trees and natural topography of the area is going to force us to be a little more clever than that. -- C. Magnus Hedemark http://trilug.org/~chrish "The only way to keep your health is to eat what you don't want, drink what you don't like, and do what you'd rather not." - Mark Twain Attachment:
PGP.sig
|
|