dirceu_cavendish at yahoo.com
Wed Aug 8 04:03:48 BST 2007
Creeping up into other people's topic, if I am allowed...
----- Original Message ----
From: Lachlan Andrew <lachlan.andrew at gmail.com>
To: dpapadimitriou at psg.com; dimitri.papadimitriou at alcatel-lucent.be
Cc: iccrg IRTF list <iccrg at cs.ucl.ac.uk>
Sent: Tuesday, August 7, 2007 6:30:31 PM
Subject: Re: [Iccrg] draft-irtf-iccrg-welzl-congestion-control-open-research-00.txt
[lots of stuff deleted, so as to focus on a specific point...]
>From a control theory point of view, there's recently been a lot of
work on the minimum information-theoretic signal rate required to
"stabilise" a system (e.g,
<http://www.ee.unimelb.edu.au/staff/gnair/NairPIEEE.pdf>). I'm not
sure how much of that relates to congestion control, where
undershooting the rate is much preferable to overshooting.
To decide how much feed back is "needed", we need to decide what
performance measures to use, and what performance level is "needed" --
e.g. do we need unused bandwidth to be taken up again within x
RTTs/seconds/minutes of being released by another flow? Do we need
overload to abate within x RTTs?
<DC> I am assuming that "how much feedback is needed" to mean how often feedback info
should be received from the context above. From a control theoretical point of view, a CC
scheme tries to "follow" the dynamics of the disturbance of the system, which in this case can
be taken as available bandwidth (could be something else, such as queue levels). Suppose that as the result of a disturbance at time To, the response curve of the monitored variable has a maximum frequency component of Fmax. Then, we need to "sample" this variable at least as much as 2 x Fmax, the Nyquist sampling frequency. Obviously that this fact plays a role only on schemes that the "monitored variable" is multibit. For a single bit variable, such as loss, one needs to sample the system as often as the "bit flips". In practical terms, it means that for loss based AIMD CC the controller needs to be informed at every loss event.
Lachlan Andrew Dept of Computer Science, Caltech
1200 E California Blvd, Mail Code 256-80, Pasadena CA 91125, USA
Phone: +1 (626) 395-8820 Fax: +1 (626) 568-3603
Iccrg mailing list
Iccrg at cs.ucl.ac.uk
Be a better Heartthrob. Get better relationship answers from someone who knows. Yahoo! Answers - Check it out.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Iccrg