> How do I know the difference between a signal which is decreasing in
> power as frequency increases because of its characteristic distribution
> (e.g., pink noise, brown noise), and a signal which is decreasing in
> power as frequency increases because it's band limited (e.g., white
> noise with a continuous falloff)?
The difference between "signal" and "noise" is basically the the
"signal"
contains information you might find "useful" in some sense, while the
"noise" does not. Note that the "useless" noise might contain
information
that might be useful to somebody else than you, or elsewhere than the
current application. The obvious example are two interefering radio
sources, one you want to listen to and the other interfering.
So basically, you have to characterize the useful signal as good as
you possibly can, and then regard evberything else, that does not
fit those characteristics, as noise. You can't generally say that
a signal is noise merely based on a general spectrum shape.
I know there exist radio systems where the emitted signal has
a spectrum resembeling white noise.
Rune
Reply by ●February 24, 20062006-02-24
Thanks for all the help guys - my intuition was that the two were
indistinguishable, but I don't have a lot of experience in this area.
-Trevor
Reply by Martin Eisenberg●February 24, 20062006-02-24
Jerry Avins wrote:
> Odd things happen when words become names. Apropos brown: on the
> UPS website: "UPS, the UPS brandmark and the color brown are
> registered trademarks of United Parcel Service of America, Inc.
> All rights reserved." Software patents are tame by comparison.
The German Telekom tried to have the CMYK Magenta characteristic for
their ads protected some time ago but were denied, as far as I know.
Martin
--
Quidquid latine scriptum sit, altum viditur.
Reply by Jerry Avins●February 23, 20062006-02-23
robert bristow-johnson wrote:
> Jerry Avins wrote:
>
>>Tim Wescott wrote:
>>
>>
>>>Brown noise? That's a new one.
>>
>>Brown noise follows the statistics of a random walk (Think Brownian
>>motion). 1/f^2
>
>
> sometimes it's called "Red noise".
>
> http://en.wikipedia.org/wiki/Colors_of_noise
>
> personally, i think "red noise" is less scatolgical and positions pink
> noise nicely as halfway between red and white noise. if the -3 db/oct
> were called "ochre noise" or "tan noise", then "brown noise" would be
> appropriate for -6 dB/oct.
Brown noise is not named after a color. It is named in honor of a person
(whose name happens to be Brown).
http://en.wikipedia.org/wiki/Brownian_motion
Strangers meet and introduce themselves. The first says that he has a
very hard name; the second says that he's sure his name is harder. The
first is sure he takes the cake: "My name is Stone. Not much harder than
that." The second says, "Whatever your name may be, my name is Harder."
Odd things happen when words become names. Apropos brown: on the UPS
website: "UPS, the UPS brandmark and the color brown are registered
trademarks of United Parcel Service of America, Inc. All rights
reserved." Software patents are tame by comparison.
Jerry
--
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������
Reply by robert bristow-johnson●February 23, 20062006-02-23
Jerry Avins wrote:
> Tim Wescott wrote:
>
> > Brown noise? That's a new one.
>
> Brown noise follows the statistics of a random walk (Think Brownian
> motion). 1/f^2
sometimes it's called "Red noise".
http://en.wikipedia.org/wiki/Colors_of_noise
personally, i think "red noise" is less scatolgical and positions pink
noise nicely as halfway between red and white noise. if the -3 db/oct
were called "ochre noise" or "tan noise", then "brown noise" would be
appropriate for -6 dB/oct.
r b-j
Reply by Tim Wescott●February 23, 20062006-02-23
john wrote:
> Oli Filth wrote:
>
>>Tim Wescott said the following on 23/02/2006 20:13:
>>
>>>goodchild.trevor@gmail.com wrote:
>>>
>>>>How do I know the difference between a signal which is decreasing in
>>>>power as frequency increases because of its characteristic distribution
>>>>(e.g., pink noise, brown noise), and a signal which is decreasing in
>>>>power as frequency increases because it's band limited (e.g., white
>>>>noise with a continuous falloff)?
>>>>
>>>
>>>Brown noise? That's a new one.
>>
>>Brown noise = Brownian noise, i.e. 1/f^2 spectrum, or 20 dB/dec, or 6
>>dB/oct.
>>
>>
>>In answer to the OP's question, I don't think there's a lot you can do
>>to identify the difference. Whether the spectral response of the
>>received signal is due to band-limiting or due to the inherent noise
>>process, they're exactly equivalent. Pink and brown noise can be
>>created by low-pass-filtering white noise appropriately, which is
>>exactly equivalent to (appropriate) band-limiting.
>>
>>
>>--
>>Oli
>
>
> I wonder if a histogram would offer a clue. Is it true that if noise
> has the same PSD, that it also has the same PDF? It has just been too
> long since college for me to remember this stuff.
>
> John
>
Brown noise follows the statistics of a random walk (Think Brownian
motion). 1/f^2
Jerry
--
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������
Reply by john●February 23, 20062006-02-23
Oli Filth wrote:
> Tim Wescott said the following on 23/02/2006 20:13:
> > goodchild.trevor@gmail.com wrote:
> >> How do I know the difference between a signal which is decreasing in
> >> power as frequency increases because of its characteristic distribution
> >> (e.g., pink noise, brown noise), and a signal which is decreasing in
> >> power as frequency increases because it's band limited (e.g., white
> >> noise with a continuous falloff)?
> >>
> > Brown noise? That's a new one.
>
> Brown noise = Brownian noise, i.e. 1/f^2 spectrum, or 20 dB/dec, or 6
> dB/oct.
>
>
> In answer to the OP's question, I don't think there's a lot you can do
> to identify the difference. Whether the spectral response of the
> received signal is due to band-limiting or due to the inherent noise
> process, they're exactly equivalent. Pink and brown noise can be
> created by low-pass-filtering white noise appropriately, which is
> exactly equivalent to (appropriate) band-limiting.
>
>
> --
> Oli
I wonder if a histogram would offer a clue. Is it true that if noise
has the same PSD, that it also has the same PDF? It has just been too
long since college for me to remember this stuff.
John
Reply by Oli Filth●February 23, 20062006-02-23
Tim Wescott said the following on 23/02/2006 20:13:
> goodchild.trevor@gmail.com wrote:
>> How do I know the difference between a signal which is decreasing in
>> power as frequency increases because of its characteristic distribution
>> (e.g., pink noise, brown noise), and a signal which is decreasing in
>> power as frequency increases because it's band limited (e.g., white
>> noise with a continuous falloff)?
>>
> Brown noise? That's a new one.
Brown noise = Brownian noise, i.e. 1/f^2 spectrum, or 20 dB/dec, or 6
dB/oct.
In answer to the OP's question, I don't think there's a lot you can do
to identify the difference. Whether the spectral response of the
received signal is due to band-limiting or due to the inherent noise
process, they're exactly equivalent. Pink and brown noise can be
created by low-pass-filtering white noise appropriately, which is
exactly equivalent to (appropriate) band-limiting.
--
Oli
Reply by Tim Wescott●February 23, 20062006-02-23
goodchild.trevor@gmail.com wrote:
> How do I know the difference between a signal which is decreasing in
> power as frequency increases because of its characteristic distribution
> (e.g., pink noise, brown noise), and a signal which is decreasing in
> power as frequency increases because it's band limited (e.g., white
> noise with a continuous falloff)?
> Thanks,
>
> Trevor
>
Brown noise? That's a new one.
If you take white noise and shape it with a 3dB/octave roll-off filter
then it'll be pink noise.
If it's falling off at 6dB/octave then it's a more 'normal' bandlimited
process.
Assuming (nice word, that) that you can get good data and a good
spectrum, look at the slope of the rolloff.
--
Tim Wescott
Wescott Design Services
http://www.wescottdesign.com
Posting from Google? See http://cfaj.freeshell.org/google/