On Thu, 2005-12-22 at 14:27 -0700, Herb Peyerl wrote:
> On 22-Dec-05, at 2:03 PM, Jim Gettys wrote:
> >> Is it for web browsing, P2P messaging and/or phone?
> >
> > All of the above.  Voice is currently a serious challenge.  There is
> > research here about why voice over more than one 802.11 hops has been
> > losing completely; they've just succeeded in getting it working,  
> > but are
> > still understanding the phenomena of why there is trouble over  
> > multiple
> > (even 2) hops, where horrific packet losses have been observed.
> 
> "Research"?  I thought it was widely acknowledged that $20 pieces of  
> consumer Wifi gear aren't designed to shuffle large quantities of  
> teeny packets....
> 

You underrate the $20 hardware.  And remember, the on the air protocols
found in 802.11abg were not designed for the mesh case.

The packets are getting dropped, but shouldn't be; the different
machines are interfering with each other.  It appears to be the hidden
node problem/and or how carrier sensing should behave. 

The question is exactly why? And what changes to air protocols will fix
it?  At least one change has caused the packet loss to go away, but
exactly what is happening and what the best "fix" is wasn't fully
understood as of last week, when I attended a presentation made by one
of Andy Lippman's students. This is certainly research in my book.

The Athleros 5K chips lets one get at complete data frames and play
whatever games you like, and remember, we don't have to follow the
802.11 standards on the transmission of data on the wire.

In any case, I'm not personally working on this; I'm a very interested
bystander. http://web.media.mit.edu/~lip/ is the group doing the work,
and I'd recommend following up with them.
				- Jim