[CUWiN-Dev] Client/Mesh Domains

Quantum Scientific Info at Quantum-Sci.com
Thu Jan 20 11:44:17 CST 2005


Hello List,

Am I understanding correctly, that each node must be hard-wired to the client's LAN?  That a cheap, generic bridge cannot associate with a node to provide client service across the street from the node?  IOW, there is only a mesh cloud, not a mesh cloud with a client cloud around each node?  If there's not a client cloud, costs would be higher. (below)

Is anyone working on grafting in NoCatSplash?  We may attempt this, although are more hardware-oriented.  Where can the CUWIN kernel conig file be found?

When will the Milestone II images be available?  Will Milestone II have implemented HSLS & ETX?

I cannot see how to boost the power of an Atheros radio to 200mW, as everyone uses Atheros' reference design, and there's not enough space on the PCB.  The existing PA on Netgear's WG511U has a max Pout of +20dBm, and probably actually max's out at +18dBm (although Netgear has not been able to confirm their own Tx power, even at the highest levels of "support").  The Atheros AR5004 chipset is wonderful because Rx sensitivity is as low as -105dBm.  I haven't looked at the miniPCI card with AR5004 yet, but I suspect it's crowded as well.  Maybe someone can throw one on the scanner (RF can lid removed)?  The AR5004 chipset has these major chips: AR5112A, AR5213A (a/b/g) or AR2112A,AR2213A (b/g).  Has anyone overcome the Tx power problem?  Must we resort to Senao and the buggy Prism for 200mW & better?

In the List archive I noticed reference to interference from other sources.  We are planning on using *horizontal* polarization, with waveguide (travelling-wave) antennas.  Even if on the same channel as a noise source, there is a 20dB difference.  Unfortunately these antennae cannot be made for less than $100 each, although this should be a safe method of bandwidth isolation indefinitely, particularly in urban areas.  And they're even quite safe from lightning *inherently*, as lightning is vertically-polarized, and also the enclosure is grounded aluminum.  OTOH, this higher node cost is why it would be best if clients could connect to a node wirelessly.

In solidarity,

Carl Cook


More information about the CU-Wireless-Dev mailing list