[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [ns] Average One-Way delay computation
Of course we can use an awk script to measure One-Way delays, but what about
measuring them during the simulation?
Previously I had found the script queue_rtt.tcl (just search for the filename
in ns-web page to find it), measuring the Round Trip Time (questioning the
source and not the sink). You could modify it to question the sink, but there
is something which is not clear to me.
Is it possible to ask the sink about the one-way packet delay?
I don't think there is any sink variable recording the trip time for the last
packet. Of course we could modify the recv() procedure at the sink to measure
whatever we want and link the variable with a TCL variable to access it during
the simulation, but I don't want to modify the existing components.
I guess we could also assume that the return path is not congested, so the
return delay is constant and equals:
ret_del=total_link_delay+transmition_time,
the transmition time is the time needed to transmit the 40 bytes ACK and there
are no queueing delays. So we can substract the computed ret_del from the rtt
that we measured and hopefully get the same one-trip delay that we can get with
awk (which can be used only if there are not too many packets, so that the
packet_ids don't wrap).
Lloyd Wood wrote:
> On Thu, 31 May 2001, Sergio Andreozzi wrote:
>
> > I want to calculate the average one-way delay from one source to one
> > destination for a certain flow. Looking at the complete trace file I find
> > the whole information and I could extract using awk.
>
> I wrote and posted a script to do just that. See my ns webpage.
>
> L.
> #
> <[email protected]>PGP<http://www.ee.surrey.ac.uk/Personal/L.Wood/>