[ns] generating background traffic for p2p evaluation

[Shad Stafford]

Does anyone have any experience generating background traffic to run 
simulations against? I would guess this is a complicated topic without 
easy answers, but I want to evaluate how a peer-to-peer system adapts 
to transient congestion over the network and ideally that congestion 
would bear some resemblance to what happens in the real world. This 
idea addressed briefly in the Floyd/Paxson paper "Difficulties in 
Simulating the Internet" but I'm hoping that someone has developed some 
practical methodologies since that paper was published. Perhaps 
something like the ideas presented in "Workload Generation for ns 
Simulations of Wide Area Networks and the Internet" by Sikdar et al.

Suggestions?

Thanks

-Shad