Sex, Drugs & Unix

Home » Archives » August 2004 » more on beamforming for mesh networks

[Previous entry: "running linux, again"] [Next entry: "Mojo mosh networking"]

08/01/2004: "more on beamforming for mesh networks"


Microsoft had a summit on Mesh networking back in June. I didn't attend, but Jim Kajiya presented his ideas and results on "Commondity class Phased Array Antennas". In the video he does pick on Vivato's cost structure some. In fairness, Kujiya admits that the phased array he's designed he's designed isnt' fast enough to packet switch. (Speed is one of the primary reasons that Vivato uses beam switching.)

The PowerPoint slides that go with his talk are mis-linked. The real link is here.
Jim-Bob sez, "check it out".

Jim also thinks you should check out RTM's talk on MIT's Roofnet project. Video here, slides are here.

Click project (used in Roofnet) is one of the slickest bits of networking code I've ever read. Coming soon (watch this space): OLSR with ETX all pre-packaged and pretty.

Mesh will win, though not with 802.11 as a MAC. Kujiya's inexpensive phased array would probably work fine if the transmits were scheduled, and this requires a non-802.11 MAC. Turns out that madwifi makes this possible. And madwifi has recently been forked to work with Click. Mesh with phased arrays (yeilding a highly directional pattern) will be is a marriage made in heaven. Phased array Access Points can only really solve the problems of 802.11g in Enterprise-like settings or other large indoor environments.

Touch with greatness department: RTM once worked at Convex, the summer before I started work there. He left his dresser behind in the house where he lived (and I moved into). I still had it until I left it behind in Spokane.