sanity test: distribution of gap between two noise pulses
Gap between two noise pulses is an interval and IMO follows exponential distribution, not normal. For more exact simulation, this can be reflected (but it's a detail).
On another note, we may use the PRNG implemented in ieee.math_real (uniform) and then transform it to the desired distribution. On the plus side, we will be able to change the seed. It is a question, however, how the performance will be affected.
Normal: N(m,v) = sqrt(-2.0 * log(u1)) * cos(2piu2) * var + mean [Box-Muller] Exponential: E(1/beta) = -log(u)*beta (u from U(0,1))