Digital Signal Processing Reference
In-Depth Information
Fig. 2.6 Block diagram of a
practical signal processing
system
Digital
Signal
Digital
Signal
r(n)
p(n)
x
(
t
),
Analog
Input
Signal
y
(
t
),
Analog
Output
Signal
LPF &
A / D
D / A
DSP
& LPF
2.3 Time-Domain / Frequency-Domain Representations
Like analog signals and systems, digital signals and systems can be analyzed either
in the time-domain or in the frequency domain. The two domains are equivalent,
provided that suitable transformations are used. The most common time-to-fre-
quency transformations for continuous-time signals are the Fourier transform and
the Laplace transform. The counterparts of these transforms for sampled signals
are the discrete-time Fourier transform (DTFT) and the z-transform. The DTFT is
useful for analyzing the frequency content of signals, while the z-transform
is useful for both analyzing the frequency content of signals and analyzing the
stability of systems. The DTFT and z-transforms will be defined and studied in
more detail in
Sects. 2.3.2.2
and
2.3.3
.
2.3.1 Time-Domain Representation of Digital Signals and Systems
This section considers the representation and analysis of digital signals and sys-
tems. Fundamental to time domain analysis of discrete-time signals is discrete-
time convolution, which is defined in what follows.
2.3.1.1 Discrete Linear Convolution
If x(n) and y(n) are two discrete signals, their discrete linear convolution w(n)is
given by:
w
ð
n
Þ¼
x
ð
n
Þ
y
ð
n
Þ¼
X
1
x
ð
k
Þ
y
ð
n
k
Þ :
k
¼1
Note that k is a dummy variable of summation. The above formula is similar to
that of continuous-time linear convolution, except that summation is used instead
of integration.
If both x(n) and y(n) are finite signals of lengths N
1
and N
2
samples, respectively,
the length of w(n) is finite and Is given by L = N
1
? N
2
- 1. Hence, if x(n) and
y(n) are of equal length N, then the result of the convolution is L = 2N - 1
samples long.
Search WWH ::
Custom Search