Re: impulsive noise
You can modelate it with a Poisson's process. This process have an input with this form:
z(t)=∑δ(t-ti)
because the process is activeds by a secuence of impulses ocurring at random times ti.
You need the set of Poisson points ti, with average λi and a real function h(t), then you have that the process is:
w(t)=∑h(t-ti).
If you use Campbell's theorem, the mean is η=λ∫h(t)dt and variance σ²=λ∫h²(t), where the limits of the integral are from -∞ to +∞
Regards