[CUWiN-Dev] New Platform

Quantum Scientific Info at Quantum-Sci.com
Sat Mar 12 21:59:02 CST 2005


On Saturday 12 March 2005 9:37, Andreas Petersson wrote:
> whey does hsls require that much memory?
> 
> we use olsr and are fine with 16 MB ram on our wrt54g.

Actually it's the daemons that require it (Squid, DNS) and maybe packet 
routing, although I don't know that a node would actually need more than 
128MB.  The option to readily add RAM is there, if needed.  64MB of flash is 
needed, for remote updating of the image.  Heck -I- could upgrade the flash 
and memory for prototypes, if given materials.  I've done this before.

Docs say this hardware IPSec moves packets at wire speed, which could mean 
high security end-to-end with no delay, although I don't know whether it can 
do IPV6.  The box could likely be powered with POE, although I'm still 
waiting for current draw ratings, etc.

 
> i think in order to use that mimo to your advantage you will also need 
> three identical antennas, which are mounted just a few centimeters away 
> from each other.

Correct, MIMO needs three antennae.  And I suspect standard rubber duckys 
would be sufficient, given the performance of this thing.  Have a number of 
questions pending at Airgo.  Antennae would mount on the outdoor enclosure 
and connect directly to the mPCI card with U.FL connectors.  Far cheaper than 
a pigtail-to-heliax-to-antenna arrangement.  Antenna spacing is 66mm 
center-to-center, so would require an enclosure at least 150mm wide.  I do 
not know how this performs in forest, and this is a concern.  Pine needles 
are about 1/4 wavelength, and so absorb well.  Also MIMO may not be good for 
point-to-point microwave shots unless 3 directional antennae.  But it would 
do much better than 'b' in urban settings, where there's lots of reflection 
(multipath) and many nodes.

Best,

Carl Cook



More information about the CU-Wireless-Dev mailing list