[Iccrg] review of Compound TCP draft

[Lachlan Andrew]

Greetings Doug,

On 21/11/2007, Douglas Leith <Doug.Leith at nuim.ie> wrote:
> How about if the spoofing makes the base delay look higher e.g. by
> tinkering with tcp timestamps.   Would that mean the estimated queue
> delay would be too low and so make a flow more aggressive ?

Good suggestion.  I don't think that *changing* the timestamp would
help (it is normally too coarse to use for queue estimation, and
messes with PAWS), but you could use it to estimate the one-way delay,
and then pace out the ACKs to pretend there was no forward queueing
delay.

You'd have to account for clock drift.  Perhaps you could start by
deliberately having very long delays (like 1s), and gradually reducing
them over time.

There's lots of fun to be had.  I don't think any of it would bring
enough benefit to the user to be a dangerous exploit, but worth
looking into.

Cheers,
Lachlan

-- 
Lachlan Andrew  Dept of Computer Science, Caltech
1200 E California Blvd, Mail Code 256-80, Pasadena CA 91125, USA
Ph: +1 (626) 395-8820    Fax: +1 (626) 568-3603
http://netlab.caltech.edu/~lachlan