CCID 2 with slowly-responding AIMD parameters?

It would be a simple change to add an option to CCID 2 to
have the 
sender use
different AIMD parameters.  The obvious possibility would be
for the
slowly-responding option to use AIMD (1/5, 1/8) instead of
AIMD(1, 1/2).
This means that the congestion window would be increased by
1/5 of an 
MSS
rather than by one MSS per round-trip time, and after loss
events the 
congestion
window would be decreased to 7/8-ths of its old value, not
to half of 
its old value.
This would differ from the regular CCID 2 by being
slowly-responding, 
and would
be somewhat similar in its time scales to CCID 3.

This is discussed in some detail in the following:

A Comparison of Equation-Based and AIMD Congestion Control.
Floyd, S., Handley, M., and Padhye, J.
May 2000.
URL "http://www.icir.org/t
frc/"

Is this worth doing?  It would be very simple to specify,
and very 
simple to
implement.   And it shouldn't have too many surprises in its
dynamics.

- Sally
http://www.icir.org/floyd/

I think this would be a good idea to add this as an option.
This
option could be an alternative to interactive applications
that use
CCID 3.

-Arjuna

On 11/6/06, Sally Floyd <sallyfloydmac.com> wrote:
> It would be a simple change to add an option to CCID 2
to have the
> sender use
> different AIMD parameters.  The obvious possibility
would be for the
> slowly-responding option to use AIMD (1/5, 1/8) instead
of AIMD(1, 1/2).
> This means that the congestion window would be
increased by 1/5 of an
> MSS
> rather than by one MSS per round-trip time, and after
loss events the
> congestion
> window would be decreased to 7/8-ths of its old value,
not to half of
> its old value.
> This would differ from the regular CCID 2 by being
slowly-responding,
> and would
> be somewhat similar in its time scales to CCID 3.
>
> This is discussed in some detail in the following:
>
> A Comparison of Equation-Based and AIMD Congestion
Control.
> Floyd, S., Handley, M., and Padhye, J.
> May 2000.
> URL "http://www.icir.org/t
frc/"
>
> Is this worth doing?  It would be very simple to
specify, and very
> simple to
> implement.   And it shouldn't have too many surprises
in its dynamics.
>
> - Sally
> http://www.icir.org/floyd/

>
>
>


-- 
Electronics Research Group
University of Aberdeen
Aberdeen AB24 3UE
Web: www.erg.abdn.ac.uk/users/arjuna
Phone : +44-1224-272780
Fax :     +44-1224-272497