Despite its common occurrence, 1/f noise is not well-understood. |
Physical explanations are lacking, but the late Per Bak
and coworkers proposed a mechanism he called self-organized criticality. |
Though interesting, this view has not been universally accepted. |
The best mathematical model for 1/f noises is fractional
Brownian motion, fBm, developed by Mandelbrot and
Wallis. |
Fractional Brownian motion is characterized by a parameter H, 0 < H < 1:
fBm is a random process Y(t) having increments Y(t + dt) - Y(t) is normally distributed
with mean = 0 and standard deviation = dtH. |
The power spectrum of fBm is P(f) = 1/f2H + 1, so as H
gets closer to 0, fBm gets closer to 1/f noise. |