[ns] Re: Urgrnt Q

[Lloyd Wood <l.wood@eim.surrey.ac.uk>]

> I am using cmu-trace file to calculate delay. The delay here is
> calculated as the difference between the send_time for the packet
> and the received_time at the destination. Is there any way to
> include the buffering time at the TCP, I mean to calculate it as
> the difference between the generated_time for the packet at the
> application layer and the received_time. Is there any parameter in
> NS to record the packet generated time.

Buffering of data at the TCP receiver (TCPSink) is certainly a problem
- out-of-order delivery can mean that the application doesn't see
arrivals the way the sender does, and I have a hack to see what the
application sees on my ns page.

Buffering at the sender is less of a problem. FullTcp simulates
Nagle's algorithm (recently discussed on this list) - delays in e.g.
when the TCP sender gets CPU time and gets run aren't simulated, and
are probably out of scope for ns.

L.

I'm just going to have to start ignoring my ns-related mail...

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