[ns] Clustered errors-Random errors

["S. Stellin" <stellin@interfree.it>]

I must evaluate the Tcp performance under both bursty errors and
random errors conditions (at the same global rate, if possible) . How can I do
for producing  bursty errors? 
I'm using a multistate error model with two states and I have tried by
modifying the random variables or the rates of the two states, but I don't have
obtained the expected results. (clearly!)

Thanks for your help
Stefano