[Cu-wireless] 802.11 & multihop networks
David Young
dyoung at pobox.com
Thu Apr 17 17:37:02 CDT 2003
On Wed, Apr 16, 2003 at 09:34:09AM -0500, Zachary C. Miller wrote:
> I'm guessing changing this would mean hacking the firmware on all the
> radios in the network. Not really something that is going to happen
> outside the domain of theory...
Right. And it won't be much longer that there are even firmwares;
makers are putting the 802.11 MAC into silicon. This actually makes the
radios more versatile, provided that the 802.11 MAC is satisfactory.
If not, well....
> so I'd think we'd just have to live
> with it and keep our pods small as we already knew we had to. (Or we
> could change to a heirarchical design with directional antennas... :)
Nothing stops us from using a single directional antenna at each
station, and I would like for us to do that at houses on Green Street
and on Race Street which are too near their nexthop to grow the
network. Those will be well-served by a $30 patch antenna.
Hierarchical routing is fine provided that it is ad hoc. I am convinced
that we have come this far only because when we have turned on two
nodes, they have ordinarily found each other and "just worked." If we
have to start designing the network, we are going to get stuck.
A design with directional antennas is fine, too, provided that it
is ad hoc. Since radios have gotten so cheap ($35 for a decent PCI
radio with SMA connector), I am entertaining this notion that we could
put two, four, or more small patches on a single mast, "bond" their
interfaces together with a software bridge device, and then run OSPF
on the bridge device precisely as we do now. But here we are talking
about more connectors and more cable, and it is likely that antennas on
the same mast will interfere with each other unless they are very nice
(read: very expensive) antennas.
Dave
--
David Young OJC Technologies
dyoung at ojctech.com Urbana, IL * (217) 278-3933
More information about the CU-Wireless
mailing list