Re: congestion window in tcp

[Lloyd Wood <eep1lw@surrey.ac.uk>]

On Mon, 13 Dec 1999, Fabienne LEFEVRE wrote:

> I have a question concerning the tcp.cc file
> 
> According to me, in the tcp standard, when you receive a duplicate ack,
> you only close the window to half of its size. 

that was introduced with Reno.

> But it seems that in the
> code of the function dupack_action() in the tcp.cc file, the window is
> set to 1 as if it was a time out.
> Could someone please tell me if I misunderstood something or if there is
> really a mistake in my script

Well, tcp.cc is Tahoe, not the current tcp standard (whatever that
is - moving target.)

There are a number of tcp variants implemented in ns; these
implement standards at different points. You need to examine
and compare dupack_action() in tcp.cc, in tcp-newreno.cc, in
tcp-sack1.cc...

See section 25.1.5 of the current nsDoc.

cheers,

L.

<L.Wood@surrey.ac.uk>PGP<http://www.ee.surrey.ac.uk/Personal/L.Wood/>