[CUWiN-Dev] degradation between hops, antennas, etc.
Sascha Meinrath
sascha at ucimc.org
Sun Mar 6 13:19:05 CST 2005
On Sun, 6 Mar 2005, Quantum Scientific wrote:
> On Sunday 06 March 2005 10:51, Stephen Ronan wrote:
>> They mention approximately 15% degradation and say the access points
>> could support two radios but they're only using one. I wonder whether
>> the 15% is what's theoretically achievable with two radios rather than
>> what they're getting in practice.
>
> Yes, I'd posted and discussed Mesh Dynamics' research here previously.
> They've actually been around for a while.
>
> Throughput degradation depends entirely on node density. In a
> completely linear topology and single radios you should lose 50% per
> hop, because of the half-duplex nature, and with dual radios about 18%
> per hop. Both approaches get considerably worse when nodes can see
> multiple neighbors, again, because of half-duplex packet propagation...
> and this doesn't take into account multipath.
So I think it's important to remind folks that bandwidth degredation rates
really haven't been figured out yet. I strongly suspect (especially after
talking with mesh researchers like P.R. Kumar and Nitin Vaidya) that there
are numerous ways to help alleviate the problem (e.g., decreasing node
density, automatic power control, dual channel usage) -- using multiple
radios is simply _one_ tactic of many.
> I love the idea of capacious nodes, all seeing one another and
> aggregating their bandwidth, but it can't happen when a frequency is
> shared. So other measures must be taken.
>
> I'm sufficiently freaked out about this to plan two-radio nodes, and
> probably directional antennae on designated nodes. I also believe that
> there must be some measure of software distinction for gateways, but I
> think I'm too enthusiastic to be listened to or something.
The CUWiN software will easily handle two radios (and probably more, but I
defer to the developers on that one). Carl, if you're serious about
implementing dual radio support, we'd _definitely_ like to see it happen.
I think that Dave & co. could provide you with information on this, but
please confirm for us. We'd guestimated that it would only take a couple
weeks of work (potentially less if the -dev group can get you regular
feedback on things).
Let us know,
--Sascha
--
Sascha Meinrath
President * Project Coordinator * Policy Analyst
Acorn Worker Collective *** CU Wireless Network *** Free Press
www.acorncollective.com * www.cuwireless.net * www.freepress.net
More information about the CU-Wireless-Dev
mailing list