Re: Congestion vs Channel losses

>
> My first question is: TCP assume channel losses ~zero.
> Is this assumption is correct.

No. Suggest you read RFC 3155 "End-to-end Performance Implications of Links 
with Errors."  You can find it at ftp://ftp.isi.edu/in-notes/rfc3155.txt.

>
> My second question is: Is any working group in IETF or other reseach
> group addresses congestion vs channel losses issue related to TCP.
>

This is part of the PILC working group charter.  You can read it at 
http://www.ietf.org/html.charters/pilc-charter.html.

> My third question is: I think ECN addresses the congestion
> problem not channel losses: Is this assumption is correct.
>

Yes.  Because you can never know that all routers in your path are ECN 
capable and, even then, ECN capable routers throw away packets under high 
levels of congestion.  So, you can never be sure that lost packets are due 
to channel corruption.

The folks at NASA and BBN have been looking at explicit corruption 
notification as a research activity.  I saw an email recently pointing to a 
report at http://wizards.ir.bbn.com/projects/pace/eten/.  Bear in mind that 
this is not an IETF activity.

Hope this helps,

--aaron


> Thanks,
> Amin.