EE 329

Prof. Rich Kozick

Spring, 1997

**EE 329: Homework Assignment #3 **

**Date Assigned:** Monday, February 10, 1997

**Date Due:** February 12 and 14, 1997

- Continue working on your FFT program as described on the Lab 3 handout. Submit your Matlab programs by Friday, February 14.
- Also for Friday, explain why the computational complexity of the FFT algorithm is when is a power of 2. We showed in class that each butterfly requires 1 complex multiply and 2 complex additions. You should explain why butterflies are needed.
- Please work on the following problem for Wednesday,
February 12.
Bring your plots, ideas, and questions to class on Wednesday for
discussion. Then I will ask you to summarize your results and
submit them on Friday.
- Consider an analog signal that is a rectangular pulse described
by
*x*(*t*) = 1 for and*x*(*t*) = 0 otherwise. Convince yourself that the Fourier transform of*x*(*t*) has magnitude

Sketch versus frequency*f*in hertz, and plot using Matlab over the range Hz. - Use the FFT to compute and plot the magnitude spectrum of
*x*(*t*). Note that in order to use the FFT, the analog signal*x*(*t*) needs to be sampled at some rate and with some number of sampling points*N*. What are the considerations in choosing and*N*? Try various values for and*N*, and plot the FFT magnitude spectra obtained with Matlab. Try to explain the results that you are seeing. Be sure to label the FFT spectra with hertz along the horizontal axis. Can you obtain a spectrum with the FFT that is close to the analytical result in part (a)?

- Consider an analog signal that is a rectangular pulse described
by

Sun Feb 9 18:02:20 EST 1997