Re: [ns] self-similar traffic sources in NS

["Peter Rabinovitch" <peter.rabinovitch@alcatel.com>]

Before embarking on such a journey, see
M. Roughan, J. Yates and D. Veitch, "The Mystery of
     the Missing Scales: Pitfalls in the Use of Fractal
     Renewal Processes to Simulate LRD Processes,"

available from http://www.research.att.com/~jyates/publications.html



Tatsuya MORI wrote:
> 
> Hi,
> 
> From: "Peter Rabinovitch" <peter.rabinovitch@alcatel.com>
> 
> > Or, you could use a trace driven simulation. For this you'll need some
> > other software to create the trace. Vern Paxson has some S+ code to do a
> > fractional Brownian motion somewhere on his web site.
> 
> using fBm, you can obtain self-similar 'time series' (and it will be used
> for "rate" soocastic process(serieses of throughput)) , but what we
> want to obtain is 'point　process' (a couple of timestamp and packet size).
> 
> So, if you want do synsetize self-similar traffic, you have to think
> about how to use "self-similar time serieses".
> 
> I think aggregation of Pareto On/Off traffic is good for synthesizing
> self-similar trarric.
> 
> -- Tatsuya Mori <tatsuya@nttlabs.com>