discussion
[Top] [All Lists]

Re: RES: [Discussion] Alternatives of Auto Adjustment

To: Alessandra Mendonça <alessandra.mg@xxxxxxxxxx>
Subject: Re: RES: [Discussion] Alternatives of Auto Adjustment
From: John Heffner <jheffner@xxxxxxx>
Date: Sun, 02 Apr 2006 16:23:35 -0400
Alessandra Mendonça wrote:


John,

In the DRS the receiver tries to estimate the bandwidthXdelay product of the
network and the congestion-control state of the sender, again using TCP
packet header information and Timestamps.
You it spoke to me on a bounding function and I did not understand what you
want to say with this.
It would like to know more details.
You it could explain me?

In the absence of timestamps, the receiver cannot directly measure the RTT. The best it can do is to bound it by measuring the time between two segments containing sequence numbers that are exactly one window apart. This is described in detail in <http://www.lanl.gov/radiant/pubs/drs/lacsi2001.pdf>.


-John

_______________________________________________
Discussion mailing list
Discussion@xxxxxxxxxx
http://internal.web100.org/mailman/listinfo/discussion

<Prev in Thread] Current Thread [Next in Thread>