[e2e] Is a control theoretic approach sound?

[Sig.]

Hi John,

I think that without the self-clocking you need to scale the gain of the 
controller to achieve stability. On the other hand, if you exploit 
the self-clocking, the stability of the system is no more dependent on the 
value of the gain.

Alfredo

 

Scrive "John T. Wen" <wen at cat.rpi.edu>:

> Hi Alfredo:
> 
> Self clocking is modeled as scaling the source controller gain to be
> inversely proportional to the delay.  This was done in the linear context by
> Low, Paganini, and Doyle in their various papers on scalable congestion
> control.   We have also done this in the nonlinear context to achieve global
> asymptotic stability.
> 
> John
>   ----- Original Message ----- 
>   From: Alfredo Grieco 
>   To: end2end-interest at postel.org 
>   Cc: Saverio Mascolo ; shivkuma at ecse.rpi.edu ; wen at ecse.rpi.edu ;
> arcak at ecse.rpi.edu 
>   Sent: Saturday, August 02, 2003 7:04 AM
>   Subject: Re: [e2e] Is a control theoretic approach sound? 
> 
> 
>   Hi all,
> 
>   I would like to add one more point to this discussion,
> 
>   in particular, I noticed that in many papers on TCP modeling, the
> self-clocking principle, which is at the core of the Internet stability, is
> not explicitly taken into account.
> 
> 
> 
>   My question is :
> 
>   can we effectively model the Internet without taking into account one of
> the major key points of TCP congestion control ?
> 
> 
>   Alfredo Grieco
> 
> 
>   ---------------------------------------------------
>   Luigi Alfredo Grieco
>   DEE Politecnico di Bari, 
>   Via Orabona 4 - 70125 Bari, Italy 
>   Tel. +39-080-5963301
>   Fax +39-080-5963410
>   http://www-ictserv.poliba.it/grieco
>