[Iccrg] LT-TCP followup

[Wesley Eddy]

On Thu, Aug 09, 2007 at 03:23:53PM +0200, Michael Welzl wrote:
> 
> that's why I suggested using ETEN (and I agree
> that your TCP variant would probably still work
> very well if combined with it!)
> 


Just as another participant, I think this is the right idea for making
this work practical.  Using ETEN's signaling, but LT-TCP's response
might be seen as combining the best features of each.

It also may be good to consider whether the raw rate of losses due to
errors is even easier or more robust to use than trying to signal
individual losses.  CETEN's signaling would provide this and could be
used with LT-TCP rather than the two responses that we explored
(weighted coin-flipping and adapting the decrease factor):

http://www.icir.org/mallman/papers/ceten-ccr-oct2004.ps

-- 
Wesley M. Eddy
Verizon Federal Network Systems