<br><div class="gmail_quote">On Tue, May 13, 2008 at 11:59 AM, John Watlington <<a href="mailto:wad@laptop.org">wad@laptop.org</a>> wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div class="Ih2E3d"><br>
On May 12, 2008, at 3:31 PM, Ricardo Carrano wrote:<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">
<br>
> How does the collision model/scheme change between AP mode and<br>
> ad-hoc/mesh modes?<br>
<br>
As far as I can tell, it doesn't. 802.11s is interoperable with<br>
802.11abg, which means that the same media access algorithms are used.<br>
At least part of our problem might be in the synchronized transmits<br>
occurring in our present 802.11s implementation of broadcast, which<br>
are probably killing whatever CA scheme 802.11abg dictate. See:<br>
<a href="http://wiki.laptop.org/go/" target="_blank">http://wiki.laptop.org/go/</a>Path_Discovery_Mechanism:Sanity#Question_.232_-_Does_PDM_traffic_self-interfere.3F<br>
<br>
which is trying to deal with low-level path discovery requests, which<br>
also use the broadcast mechanism.<br>
--scott<br>
<br>
Yes, it is the same 802.11 DCF for both scenarios (infra and mesh).<br>
<br>
I would like to add to this discussion that sparse and dense mesh are too completely different animals. Most of the problems that we are trying to address now, are associated to the latter.<br>
</blockquote>
<br></div>
Certainly an interesting question. But is your answer really true ?<br>
I'd argue that very little, if any, testing has been done<br>
of the "large, yet sparse" mesh. Certainly none by OLPC.<br>
<br>
Our problems certainly come from "large" meshes (more than<br>
10-15 laptops in the mesh). What is your definition of sparse ?</blockquote><div><br><br>My definition, since I never found o good one in the literature. In a sparse mesh the number of active neighbors is smaller than your retry limit. So, in a sparse mesh your frame will never be discarded without being transmitted at least one time. In a dense mesh, there is a chance that a frame will reach the retry limit without being ever transmitted. So, 10 XOs in a room is a dense mesh (in that definition).<br>
<br>I've done some tests in sparse topologies early in 2007. but I doubt they have any lasting values, after so many changes and fixes. Back there we were particularly interested in the hidden node problem and checked to see if RTS/CTS would help us (by the way it does not seem to help in multihop scenarios <a href="http://www.midiacom.uff.br/~schara/publications/SBrT2007.pdf">http://www.midiacom.uff.br/~schara/publications/SBrT2007.pdf</a>)<br>
<br> </div></div>