Kalman Filters and Pink Noise

Started by Tim Wescott August 23, 2012
Oddball question.  Or maybe just a plea for comments.

I'm kind of getting painted into a corner, and the color of the paint is 
pink.  It's looking like I need to estimate pink noise with a Kalman 
filter.

Anyone done anything remotely like this?  Know of anyone who has?  Did 
they survive?

The method that I know of for simulating pink noise involves running 
white noise through a filter with a ton-o-poles (our own RBJ popped up in 
the web search, interestingly enough).  I'm kinda planning on doing just 
that for the kernel in the Kalman, and see if I can find some sane 
balance between the number of states needed and accuracy.

-- 
My liberal friends think I'm a conservative kook.
My conservative friends think I'm a liberal kook.
Why am I not happy that they have found common ground?

Tim Wescott, Communications, Control, Circuits & Software
http://www.wescottdesign.com
perjantai, 24. elokuuta 2012 3.10.27 UTC+3 Tim Wescott kirjoitti:
> Oddball question. Or maybe just a plea for comments. > > > > I'm kind of getting painted into a corner, and the color of the paint is > > pink. It's looking like I need to estimate pink noise with a Kalman > > filter. > > > > Anyone done anything remotely like this? Know of anyone who has? Did > > they survive? > > > > The method that I know of for simulating pink noise involves running > > white noise through a filter with a ton-o-poles (our own RBJ popped up in > > the web search, interestingly enough). I'm kinda planning on doing just > > that for the kernel in the Kalman, and see if I can find some sane > > balance between the number of states needed and accuracy. > > > > -- > > My liberal friends think I'm a conservative kook. > > My conservative friends think I'm a liberal kook. > > Why am I not happy that they have found common ground? > > > > Tim Wescott, Communications, Control, Circuits & Software > > http://www.wescottdesign.com
Here's a nice page about generating pink noise without too much fuzz. http://www.firstpr.com.au/dsp/pink-noise/ See section "Filtering white noise to make it pink" and specially "Paul Kellet's refined method in Allan's analysis". Hope this helps you with getting the correct color out. - Calvin
On Fri, 24 Aug 2012 02:59:47 -0700, kalvin.news wrote:

> perjantai, 24. elokuuta 2012 3.10.27 UTC+3 Tim Wescott kirjoitti: >> Oddball question. Or maybe just a plea for comments. >> >> >> >> I'm kind of getting painted into a corner, and the color of the paint >> is >> >> pink. It's looking like I need to estimate pink noise with a Kalman >> >> filter. >> >> >> >> Anyone done anything remotely like this? Know of anyone who has? Did >> >> they survive? >> >> >> >> The method that I know of for simulating pink noise involves running >> >> white noise through a filter with a ton-o-poles (our own RBJ popped up >> in >> >> the web search, interestingly enough). I'm kinda planning on doing >> just >> >> that for the kernel in the Kalman, and see if I can find some sane >> >> balance between the number of states needed and accuracy. >> >> >> >> -- >> >> My liberal friends think I'm a conservative kook. >> >> My conservative friends think I'm a liberal kook. >> >> Why am I not happy that they have found common ground? >> >> >> >> Tim Wescott, Communications, Control, Circuits & Software >> >> http://www.wescottdesign.com > > Here's a nice page about generating pink noise without too much fuzz. > > http://www.firstpr.com.au/dsp/pink-noise/ > > See section "Filtering white noise to make it pink" and specially "Paul > Kellet's refined method in Allan's analysis". > > Hope this helps you with getting the correct color out.
That page brings back a lot of memories. Allan
On Fri, 24 Aug 2012 10:25:52 +0000, Allan Herriman wrote:

> On Fri, 24 Aug 2012 02:59:47 -0700, kalvin.news wrote: > >> perjantai, 24. elokuuta 2012 3.10.27 UTC+3 Tim Wescott kirjoitti: >>> Oddball question. Or maybe just a plea for comments. >>> >>> >>> >>> I'm kind of getting painted into a corner, and the color of the paint >>> is >>> >>> pink. It's looking like I need to estimate pink noise with a Kalman >>> >>> filter. >>> >>> >>> >>> Anyone done anything remotely like this? Know of anyone who has? Did >>> >>> they survive? >>> >>> >>> >>> The method that I know of for simulating pink noise involves running >>> >>> white noise through a filter with a ton-o-poles (our own RBJ popped up >>> in >>> >>> the web search, interestingly enough). I'm kinda planning on doing >>> just >>> >>> that for the kernel in the Kalman, and see if I can find some sane >>> >>> balance between the number of states needed and accuracy. >>> >>> >>> >>> -- >>> >>> My liberal friends think I'm a conservative kook. >>> >>> My conservative friends think I'm a liberal kook. >>> >>> Why am I not happy that they have found common ground? >>> >>> >>> >>> Tim Wescott, Communications, Control, Circuits & Software >>> >>> http://www.wescottdesign.com >> >> Here's a nice page about generating pink noise without too much fuzz. >> >> http://www.firstpr.com.au/dsp/pink-noise/ >> >> See section "Filtering white noise to make it pink" and specially "Paul >> Kellet's refined method in Allan's analysis". >> >> Hope this helps you with getting the correct color out. > > > That page brings back a lot of memories. > > Allan
That page is where I found the handy algorithm for generating a pinking filter! -- My liberal friends think I'm a conservative kook. My conservative friends think I'm a liberal kook. Why am I not happy that they have found common ground? Tim Wescott, Communications, Control, Circuits & Software http://www.wescottdesign.com
I gave a talk at the 2010 comp.dsp conference that has several pink noise 
filter solutions.

http://compdsp.com/presentations.htm

There are also many other presentations at this site.

Al Clark
www.danvillesignal.com










Tim Wescott <tim@seemywebsite.com> wrote in 
news:JeWdncmTDcEBCqrNnZ2dnUVZ_rSdnZ2d@web-ster.com:

> On Fri, 24 Aug 2012 10:25:52 +0000, Allan Herriman wrote: > >> On Fri, 24 Aug 2012 02:59:47 -0700, kalvin.news wrote: >> >>> perjantai, 24. elokuuta 2012 3.10.27 UTC+3 Tim Wescott kirjoitti: >>>> Oddball question. Or maybe just a plea for comments. >>>> >>>> >>>> >>>> I'm kind of getting painted into a corner, and the color of the paint >>>> is >>>> >>>> pink. It's looking like I need to estimate pink noise with a Kalman >>>> >>>> filter. >>>> >>>> >>>> >>>> Anyone done anything remotely like this? Know of anyone who has? Did >>>> >>>> they survive? >>>> >>>> >>>> >>>> The method that I know of for simulating pink noise involves running >>>> >>>> white noise through a filter with a ton-o-poles (our own RBJ popped up >>>> in >>>> >>>> the web search, interestingly enough). I'm kinda planning on doing >>>> just >>>> >>>> that for the kernel in the Kalman, and see if I can find some sane >>>> >>>> balance between the number of states needed and accuracy. >>>> >>>> >>>> >>>> -- >>>> >>>> My liberal friends think I'm a conservative kook. >>>> >>>> My conservative friends think I'm a liberal kook. >>>> >>>> Why am I not happy that they have found common ground? >>>> >>>> >>>> >>>> Tim Wescott, Communications, Control, Circuits & Software >>>> >>>> http://www.wescottdesign.com >>> >>> Here's a nice page about generating pink noise without too much fuzz. >>> >>> http://www.firstpr.com.au/dsp/pink-noise/ >>> >>> See section "Filtering white noise to make it pink" and specially "Paul >>> Kellet's refined method in Allan's analysis". >>> >>> Hope this helps you with getting the correct color out. >> >> >> That page brings back a lot of memories. >> >> Allan > > That page is where I found the handy algorithm for generating a pinking > filter! >