home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.math
- Path: sparky!uunet!haven.umd.edu!darwin.sura.net!cs.ucf.edu!news
- From: clarke@acme.ucf.edu (Thomas Clarke)
- Subject: Re: How to Add White Gaussian Noise to Variables?
- Message-ID: <1993Jan5.220753.18742@cs.ucf.edu>
- Sender: news@cs.ucf.edu (News system)
- Organization: University of Central Florida
- References: <1993Jan5.195505.22186@wuecl.wustl.edu>
- Date: Tue, 5 Jan 1993 22:07:53 GMT
- Lines: 31
-
- In sci.math article <1993Jan5.195505.22186@wuecl.wustl.edu> you wrote:
- > Hi fellows!
- > Could anybody tell me how to add white gaussian noises to the process
- > measurements?
- > Any response is greatly appreciated!
- > Lots of thanks in advance!
- >
- > My address: ma@wuche2.wustl.edu
- Don't have references at fingertips but if you mean to generate
- samples of Gaussian noise in software to add to your data there are
- standard methods to generate Gaussian or normally distributed random
- variables from the uniformly distributed random variables that
- computers usually supply. [try Abramowitz and Stegun, Special
- Functions for a possible reference source]
-
- The generatl idea is to take two uniformly distributed random
- samples, say A and B on [0,1] for definiteness. Then turn one into
- an exponentially distributed variable, R=exp(-S/A), where S is
- standard deviation (my memory is rusty here, be sure to check
- other sources), and another into a uniform [0,2pi]=2pi*B.
- Then two Gaussian RVs are given by X=Rcos(B) and Y=Rsin(B).
- I don't recall the constant relating S and the standard deviation
- of X and Y, but the constant is of order unity.
-
- The trick is that the comples Gaussian RV (X+iY) goes into
- exponential and uniform [0,2pi] in polar notation.
- --
- Thomas Clarke
- Institute for Simulation and Training, University of Central FL
- 12424 Research Parkway, Suite 300, Orlando, FL 32826
- (407)658-5030, FAX: (407)658-5059, clarke@acme.ucf.edu
-