Forums

gaussian white noise generation

Started by sita January 18, 2006
hello

can any one please suggest me how to generate guassian white noise with
zero mean and variance 1 in matlab?

thanks.

"sita" <swethapriyanka@gmail.com> wrote in message
news:1137566793.060472.127900@f14g2000cwb.googlegroups.com...
> hello > > can any one please suggest me how to generate guassian white noise with > zero mean and variance 1 in matlab? > > thanks. >
If you have uniform distributed noise you can take say 12 samples and average them then subtract the dc level.This should be approx Guassian.Repeat process for every 12 samples giving one Guassian sample. Naebad
Naebad wrote:

> > can any one please suggest me how to generate guassian white noise with > > zero mean and variance 1 in matlab? > > > > thanks. > > > If you have uniform distributed noise you can take say 12 samples ...
There's that number again. There must be something about it that I don't know about :-).
Andor wrote:

>Naebad wrote: > > > >>>can any one please suggest me how to generate guassian white noise with >>>zero mean and variance 1 in matlab? >>> >>>thanks. >>> >>> >>> >>If you have uniform distributed noise you can take say 12 samples ... >> >> > >There's that number again. There must be something about it that I >don't know about :-). > >
There's nothing magic about 12. The more numbers you sum, the closer the result gets to gaussian. Where you stop is a kind of diminishing returns issue - or perhaps I should say a good enough for the application issue. Just summing 4 or 5 numbers achieves perfectly adequate results for many things. 12 is pretty accurate. For some reason I never figured out, some applications requiring only modest quality noise sum over 20 - comfort noise creation in the G.729 reference code is an example of this. Regards, Steve
Hi

Use the function randn.
(It generates white Gaussian samples with unit variance).

Tsachi.

Steve Underwood wrote:

> Andor wrote: > > >Naebad wrote: > > > > > > > >>>can any one please suggest me how to generate guassian white noise with > >>>zero mean and variance 1 in matlab? > >>> > >>>thanks. > >>> > >>> > >>> > >>If you have uniform distributed noise you can take say 12 samples ... > >> > >> > > > >There's that number again. There must be something about it that I > >don't know about :-). > > > > > There's nothing magic about 12.
I know - still, it props up too often to be chosen arbitrary (look for the 12): http://groups.google.ch/group/comp.dsp/msg/80f38cb25f383859?dmode=source http://groups.google.ch/group/comp.dsp/msg/69edad18ddce40aa?dmode=source http://groups.google.ch/group/comp.dsp/msg/85b614f280131be8?dmode=source http://groups.google.ch/group/comp.dsp/msg/e3db95a1ce560b07?dmode=source http://www.pitt.edu/~wpilib/statfaq/gaussfaq.html www.physics.ohio-state.edu/ ~gan/teaching/spring04/Chapter3.pdf http://www.create.ucsb.edu/pipermail/sc-users/2005-August/020193.html www.bipm.fr/utils/common/pdf/nmij-bipm/WS-32.pdf http://www.if.ufrj.br/teaching/compute/fortran/h2mol.for Someone (perhaps Knuth?) must have told them so. Perhaps it originates from bounding the error of the cumulative distribution functions in the Berry-Essen theorem. Regards, Andor
Andor wrote:
> Naebad wrote: > > > > can any one please suggest me how to generate guassian white noise with > > > zero mean and variance 1 in matlab? > > > > > > thanks. > > > > > If you have uniform distributed noise you can take say 12 samples ... > > There's that number again. There must be something about it that I > don't know about :-).
The 12 comes from the formula for the variance of a Uniform random distribution over [a,b] which is (b-a)^2 / 12. It is the same 12 that shows up when you derive the theoretical SNR of an A/D converter. John
Andor wrote:

>Steve Underwood wrote: > > > >>Andor wrote: >> >> >> >>>Naebad wrote: >>> >>> >>> >>> >>> >>>>>can any one please suggest me how to generate guassian white noise with >>>>>zero mean and variance 1 in matlab? >>>>> >>>>>thanks. >>>>> >>>>> >>>>> >>>>> >>>>> >>>>If you have uniform distributed noise you can take say 12 samples ... >>>> >>>> >>>> >>>> >>>There's that number again. There must be something about it that I >>>don't know about :-). >>> >>> >>> >>> >>There's nothing magic about 12. >> >> > >I know - still, it props up too often to be chosen arbitrary (look for >the 12): > >http://groups.google.ch/group/comp.dsp/msg/80f38cb25f383859?dmode=source >http://groups.google.ch/group/comp.dsp/msg/69edad18ddce40aa?dmode=source >http://groups.google.ch/group/comp.dsp/msg/85b614f280131be8?dmode=source >http://groups.google.ch/group/comp.dsp/msg/e3db95a1ce560b07?dmode=source >http://www.pitt.edu/~wpilib/statfaq/gaussfaq.html >www.physics.ohio-state.edu/ ~gan/teaching/spring04/Chapter3.pdf >http://www.create.ucsb.edu/pipermail/sc-users/2005-August/020193.html >www.bipm.fr/utils/common/pdf/nmij-bipm/WS-32.pdf >http://www.if.ufrj.br/teaching/compute/fortran/h2mol.for > >Someone (perhaps Knuth?) must have told them so. Perhaps it originates >from bounding the error of the cumulative distribution functions in the >Berry-Essen theorem. > >Regards, >Andor > >
Or perhaps it the very Anglo-Saxon style engineering approach of "take a dozen random numbers". Of course, if anyone complains you are short changing them on precision, you have the option of baking 13 values instead of 12. :-) Regards, Steve
Naebad wrote:
> "sita" <swethapriyanka@gmail.com> wrote in message > news:1137566793.060472.127900@f14g2000cwb.googlegroups.com... > >>hello >> >>can any one please suggest me how to generate guassian white noise with >>zero mean and variance 1 in matlab? >> >>thanks. >> > > If you have uniform distributed noise you can take say 12 samples and > average them then subtract the dc level.This should be approx > Guassian.Repeat process for every 12 samples giving one Guassian sample.
What is the DC level of a single average? Assuming that the RNG is itself zero mean, you can omit the subtraction step. Jerry -- Engineering is the art of making what you want from things you can get. &#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;
Andor wrote:
> Naebad wrote: > > >>>can any one please suggest me how to generate guassian white noise with >>>zero mean and variance 1 in matlab? >>> >>>thanks. >>> >> >>If you have uniform distributed noise you can take say 12 samples ... > > > There's that number again. There must be something about it that I > don't know about :-).
Six works pretty well too. 13 is better; you have to stop somewhere. Jerry -- Engineering is the art of making what you want from things you can get. &#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;