Carol and I went down to the Colorado State Fair in Pueblo on Labor Day, looking forward to the animals and the quirky fairgrounds food, and found a little more than we had bargained for: The first commercial implementation of citywide mesh networking that I think has a chance in hell.
We were threading our way through the "none of the above" huckster hall (beef jerky, gimmick mops, folding ladders, stain removers) on our way out of the fair at about 3:30 when we found a very professional-looking booth belonging to Airinet, headquartered in Pueblo. On the surface, Airinet looks like a conventional small-town ISP, which is why I almost walked past their booth. Of course, something in my synapses watches for the word "wireless," and when my Wi-Fi sense twitched I ambled over to read their literature while Carol tried out the Intelligel beds a few booths down. By the time she rejoined me a few minutes later, my jaw was on the floor. What these guys were doing was impossible.
Mesh networking has been with us for some time, though it's generally been a community thing: People put a wireless node up on the chimney and run routing software like Locustworld or MIT Roofnet on a dedicated (often junker) PC to bounce packets across nodes close enough to communicate. It's all very spit'n'baling wire, but with the right routing algorithms, people in the mesh can share files and connections to the Internet pretty seamlessly.
This kind of community networking is, of course, the sneakpath around the "last mile" problem that has bedeviled small-town people hoping for broadband ever since there was broadband. The problem with community networks (the Brisbane Mesh is a good example) is that they grow organically; i.e., as interested people bestir themselves to mount and configure a node. This gives you a network, but it's strangely shaped and full of holes and very difficult to route efficiently. What Airinet has done is created an "engineered" mesh network in which they choose where the distribution nodes are placed. The design is such that no node in the network is more than eight hops from the Airinet backhaul link, which is handled via point-to-point radios on a non-Wi-Fi frequency.
This sort of thing has been done before, usually for parks and college campuses and the occasional municipal Wi-Fi cloud. Existing mesh equipment, however, is hugely expensive, and such networks are almost by definition tax-supported cost centers. To build a mesh network that actually makes money, Airinet has designed a custom Wireless-B mesh node that can be manufactured for well under $100, and created a router system (similar to Locustworld) to work efficiently with the networks that they design. The node box is tiny (about the size of a small wireless router) and they deploy it inobtrusively by negotiating with property owners to mount it under eaves and in other out-of-the-way places in keeping with the overall network design. The node can then serve customers in homes anywhere within about about 800 feet, depending on terrain and intervening obstructions.
The nodes run a captive portal not completely unlike those you see at hotel Wi-Fi hotspots. If you're within range of an Airinet node, you connect to the captive portal, read the terms and conditions, and then sign up for 30 days' service at $25 flat. You don't need any hardware apart from a conventional Wireless-B client adapter. The $25 includes 1.5 MB downlink, 1.0 MB uplink, plus a 100 MB email account and a 100MB Web hosting account. Authentication is done via MAC and a password, and the connection is encrypted with a 2,048-bit key, which makes WEP or even WPA a grease spot by comparison.
Although the project has been in development for some time, the State Fair was their rollout and first real public appearance. They currently cover huge swaths of Pueblo and are adding nodes as fast as they can. They plan to target other smaller Colorado towns like Salida or Cañon City once their Pueblo buildout is complete.
All of this comes from a hurried conversation with Airinet's chief techie, David Bueno, and without any white papers or other hard technical information to draw on, I may have misunderstood some of the details. However, I'm going to stay in touch with them and will report here from time to time on their progress. David claims that networks like his will make WiMax obsolete before it even appears. If Airinet works as well as he claims it does (and of course, I can't test it from here) he may well be right.