« Coming Changes in Broadcast Industry | Main | FCC rules that landlords can't control uses of unlicensed spectrum »

June 25, 2004

Innovative Wireless Architectures

In the near-term, say the next three years, I think wireless network architectures will yield some of the biggest technological surprises. Absence of license will unleash constraints on licentious innovation.

For example, Dewayne Hendricks just got back from Europe raving about this little open-source device called MeshCube, a 5cm x 5cm x 7cm computer with lotsa gosintas and gosoutas, both wired and (802.11) wireless, designed for roll-your-own multiple-access-point networks, for sale for 200 Euros as a kit or 240 Euros assembeled. Dewayne has some reservations about the device, but not about the community that has self-organized around it. Dewayne writes that not only does this community have momentum to build "viable wireless user provided telecommunications infrastructure, and deal with any problems that they encountered along the way."

The 2.4 GHz learning experience -- each one teach one -- will not be lost when <1GHz unlicensed wireless arrives. But the main action is in new network architectures, not spectrum.

Posted by David Isenberg at June 25, 2004 10:41 AM

Comments

I agree that the community-driven wireless networks have encouraged the use of "new" network architectures.

One of the main incentives of building a wireless mesh network infrastructure is to avoid costly wired backhauls to link access points together. It's more about ease of deployment and reducing cost. It's certainly not much of a spectrum issue.

Posted by: Mark Cabiling at June 25, 2004 07:27 PM

so what does this new device actualy do?

Posted by: Pete at June 26, 2004 02:18 PM

Too bad most of these alleged mesh networks don't bother to deal with anything below layer 3 and think they can just do IP routing on top of 802.11 MAC/PHYs.

Almost all of the announced "mesh" networks will not scale and will collapse as soon as there is any real load on them.

There need to be tweaks to 802.11 MAC to allow for coordination of transmission/reception between nodes that are near each other. Otherwise the 802.11 devices will contention mechanisms will cause thruput to drop to very low bandwidths.

Posted by: Robert Berger at June 30, 2004 02:13 AM

802.11s hopes to solve the above issues.

Posted by: Mark Cabiling at July 5, 2004 08:27 PM

I agree with Bob's comments above, which themselves seem inline with David's hypothesis - - the 2.4 GHz learning experience, even with a pretty much fatally flawed MAC (for multi-hop networks), is just that - a learning experience. The real businesses will come out of future, better MACs and PHYs.

Posted by: Venki Iyer at July 6, 2004 05:57 PM