0
$\begingroup$

I am trying to create brown noise by integration of white noise (random walk).

I created Gaussian white noise. The histogram shows Gaussian distribution and Welch (and also FFT) shows flat spectrum.

Now I try to integrate it

$x(k+1) = x(k) + q(k)$

where $q$ stands for white noise and $x$ for resulting Brownian noise.

But the resulting spectrum isenter image description here

According Wikipedia I would expect that there should be straight line in the plot.

What I am doing wrong?

$\endgroup$
  • 1
    $\begingroup$ It looks like you are plotting the magnitude. Convert your amplitude scale to dB. $\endgroup$ – Envidia May 11 '17 at 17:32
  • $\begingroup$ @Envidia Do you mean frequency scale? If yes, then you are right! It should be log-log, not semilog. $\endgroup$ – matousc May 11 '17 at 17:44
  • $\begingroup$ I'm referring to the vertical scale. Is that in dB? Also, some code would be good to see. $\endgroup$ – Envidia May 11 '17 at 17:49
  • $\begingroup$ @Envidia The vertical scale is ok. If I make the horizontal scale also logarithmic, I can get the line (decreasing 1 decade per decade). What is corresponding to 20db per decade as is shown here en.wikipedia.org/wiki/Brownian_noise#/media/… $\endgroup$ – matousc May 11 '17 at 17:53
  • $\begingroup$ You got it. I stubbornly assumed your scale was correct. $\endgroup$ – Envidia May 11 '17 at 18:09
2
$\begingroup$

So this question does not remain as unanswered:

Making your horizontal scale into log-scale should give you the decade drop and yield the linear amplitude graph.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.