 
Summary: AMERICAN UNIVERSITY OF BEIRUT
FACULTY OF ENGINEERING AND ARCHITECTURE
DEPARTMENT OF ELECTRICAL AND COMPUTER ENGINEERING
EECE691C Digital Signal Processing Laboratory Week 9
Spectrum Analysis Prelab
1 Overview and Goals
In applications such as speech, modulation, and filter characteristic evaluation it is useful to
perform the analysis of signals in frequency domain, and observe the result real time on a
measurement device. This lab will consist of implementing such a device, called a "spectrum
analyzer", on the DSP kit. The fundamental component of our spectrum analyzer will be the FFT.
In this prelab we will examine some of the techniques in using the FFT for conveniently
performing the spectrum analysis of signals. Two main procedures will be considered:
windowing and zero padding, illustrated through MATLAB exercises.
2 Windowing
A spectrum analyzer is required to track the frequency content of a signal with indefinite length in
time. To resolve this indefinite problem, a common solution is to perform the FFT on signal
segments, a practice otherwise known as "windowing" a signal. To see the variation of signal
characteristics with time, we take consecutive windows that slide forward in the time dimension:
this is the principle of the shorttime Fourier Transform. Unfortunately, windowing, even in the
trivial case of taking the entire signal, introduces undesirable effects, as explored below.
