Roofnet software will extend service from MIT's campus to pass 100,000 Cambridge, Mass., residents: MIT is extending its network as a town-gown gesture, and initially plans for free access and will gauge response. The network will be built using Roofnet software, one of two open-source mesh-routing protocol projects underway. (The other is at CUWiN.)
This article from MIT's The Tech online newspaper has a number of technical errors, but all of them attributed to various sources, who may have provided incorrect details. For instance, "Philadelphia is creating its own utility, [Cambridge CIO Mary P.] Hart said, while San Francisco has partnered with Google to become a wireless city." As readers of this site know, Philadelphia has a non-profit that will handle relations with EarthLink, hardly its own utility, and San Francisco hasn't chosen a winning bidder yet.
Hart is also paraphrased saying, "Mesh technology allows individual computers to propagate the network and act as new access points, making it unnecessary for a user to be within range of the original wireless signal." This is true of mesh networks in which clients and nodes are using mesh routing hardware and software. For normal Wi-Fi, this is untrue: individual computers connect to nodes which mesh over the same or other frequencies with other nodes.
The Associated Press has a very sketchy version of this story out, too, which cites the same technology error: "MIT researchers will help test the "mesh" technology, which allows individual users to both receive and send a Wi-Fi signal, so that a user doesn't need to be within range of the original signal from MIT."
Wrong, wrong, wrong. (Although see Dana Spiegel's comment on this post.)
Mesh Networks, a company acquired by Motorola (and rebranded as Motomesh), is the only firm I know of that has offered commercial products with individual devices having mesh properties, rather than just infrastructure and backhaul. Motomesh abandoned this plan. Their latest offering, which can contain up to four radios (a combination of 2.4 GHz and 4.9 GHz using their proprietary algorithm and Wi-Fi), doesn't mesh across devices.
Update: Per the comment below, Motomesh does still sell adapters that act as mesh nodes--my mistake! But they're not standard Wi-Fi adapters; they're proprietary products. Their new system doesn't require these proprietary adapters, but it's an option.
Glenn,
Just a slight clarification: When the news reports indicate that MIT's Roofnet "allows individual users to both receive and send a Wi-Fi signal", I believe that the confusion isn't about the fact that mesh doesn't work on client devices, and instead about the term "Users".
My understanding of Roofnet is that you put a mesh node on your roof, like CUWiN, and then get local ethernet, while extending the mesh network. The mesh isn't to provide client device access (though places may provide local Wi-Fi for that), but rather entirely for backhaul.
Hi Glenn,
Just another clarification... MOTOMESH does retain the Client Meshing functionality of the original MEA (MeshNetworks) technology. It offers it in both the 2.4 GHz & 4.9 GHz bands.
[Editor's note: Thanks for clarifying -- but my point here is that standard Wi-Fi cards cannot mesh. Only custom wireless cards (which can have Wi-Fi capability among other protocol support) can act as mesh nodes. The MIT project involves end users with Wi-Fi cards or bridges gaining access.--gf]