Info iconThis preview shows pages 1–2. Sign up to view the full content.

View Full Document Right Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: 43 SPECTRAL ANALYSIS TO FIND A HIDDEN MESSAGE 180 43 Spectral Analysis to Find a Hidden Message As you know, the fast Fourier transform ( fft() in MATLAB), turns time-domain signals x into frequency-domain equivalents X . Here you will analyze a given signal from the site and apply several of the common treatments so as to get clean power spectral densities. 1. First, download the x data file from the site; it is called computeSpectra.dat . This signal x looks like a mess at first because of the noise, but we will tease out some specific information about X . Make a plot of the data versus time, assuming the time step is 0.0001 seconds. What features do you see if you zoom in on x , if any? Its an evidently narrowband signal with content around 1.6kHz (10krad/s), but prob- ably with a lot of noise. Overall it is very hard to see anything beyond this. 2. Use the FFT to make the transformed signal X . Plot the magnitude of X versus frequency; recall that the frequency vector corresponding to X goes from zero to the roughly the sampling rate (use the instructions from your previous homework). In your plot, only show the frequency range near 10000 rad/s where there is significant energy, and remember the 2 /n scaling that is needed to put FFT peaks into the same units as x . What do you see? See the top plot of the psds. There are about fifteen peaks at frequencies slightly above 10 krad/s. 3. Treatment 1: Windowing. The Fourier transform operation on a finite-length signal x assumes periodicity, and so any discontinuity between the last and the first points in x is a feature that will be accounted for in X . In general we dont want this effect, because it muddies up the water and makes it harder to see the details in X that we are interested in. For this and some other reasons, it is common to multiply x by a windowing function, before taking the transform. There are many such windows, developed according to different metrics. If you do no windowing, this is called thedeveloped according to different metrics....
View Full Document

This note was uploaded on 11/29/2011 for the course CIVIL 1.00 taught by Professor Georgekocur during the Spring '05 term at MIT.

Page1 / 6


This preview shows document pages 1 - 2. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online