Re: [ns] About pareto.cc

["merlin.jiang" <merlin.jiang@msa.hinet.net>]

From: "Timothy.Bauge" <Tim.Bauge@rrl.co.uk>

> Note that p1_ and rem_ are numbers of packets, p2_ and t are delays.
During
> an ON period, rem_ != 0 and t = interval_ (i.e. packet serialisation time)
> so packets are sent back to back, which is the correct behaviour. I'm not
> sure what you mean by "more reasonable"...

    Personally, I don't think that generating pareto r.v.'s based on
"ontime_" and
"burstlen_," respectively,  have the same meaning. burstlen_ is magnified,
as always is the case, and the time resolution is spoiled. ontime_ should be
used for calculating p1_, and rem_ is determined right after that by
"p1_ / interval_."

>
> If rem_ = 0 at the beginning of next_interval, there is still 1 packet to
be
> sent in the current burst. Therefore the subsequent packet will be sent at
> interval_ + next idle period, i.e. serialisation time for the last packet
+
> next idle time.

    rem_ = 0 would be caused by 2 factors: one is zero-valued r.v., but the
other is due
to "transmission deadline." For the latter, the "interval_" is not included
in "t" calculation,
I think.

    Thanks for ur valuable interpretation! ^_^

All the best,
M. Jiang