Hi, When I started to use ns-2 three years ago "I heard" that the pseudo random generators of ns-2 PRNG were of high quality (rng.cc). However, after reading "On Shortcomings of the ns-2 Random Number Generator" by Bernhard Hechenleitner et al. from 2002, I wonder if the limitations highlighted by that paper have been fixed in the years after? In short, the paper states that if you select bad seeds, your rng generated data will be based on highly correlated sources (whatever you use the PRNG for). The highly superior Mersenne Twister method is cited and tested for comparison.
When I look into the ns-2 CVS, and open the latest rng.cc, I find the very exact lines of code and comments as indicated by Hechenleiter, so I worry that one can still "select bad seeds"? I hope somebody can throw some light into this issue. best regards Arne Lie