Al-Muthanna University

College of Engineering

Department of Electronics & Communication

Engineering

Subject: Communication Systems I

First Semester: 2023-2024

Lecture 6

Signals and Systems

Dr. Abidulkarim K. I. Yasari

Books:
1. Alan V. Oppenheim, Alan S. Willsky, S. Hamid Nawab. Signals and Systems. 2nd Ed.
2. Charles L. Phillips. John M. Parr. Signals, Systems, and Transforms.

Instructional YouTube videos:

https://sites.google.com/a/asu.edu/signals-and-systems/

1
 The Discrete Fourier Transform

6.1- The Family of Fourier Analysis

Fourier analysis is named after Jean Baptiste Joseph Fourier (1768-1830), a French
mathematician and physicist. While many contributed to the field, Fourier is honored
for his mathematical discoveries and insight into the practical usefulness of the
techniques. However, a signal in Fourier can be either continuous or discrete, and it
can be either periodic or aperiodic (non-periodic). The combination of these two
features generates the four categories, described below and illustrated in Fig. 6.1,
these four types generate four types of Fourier analysis, classified as follows:

1- Aperiodic-Continuous

This includes, for example, decaying exponentials and the Gaussian curve. These
signals extend to both positive and negative infinity without repeating in a periodic
pattern. The Fourier Transform for this type of signal is simply called the Fourier
Transform (FT).

2- Periodic-Continuous

Here the examples include sine waves, square waves, and any waveform that repeats
itself in a regular pattern from negative to positive infinity. This version of the
Fourier transform is called the Fourier Series (FS).

3- Aperiodic-Discrete

These signals are only defined at discrete points between positive and negative
infinity, and do not repeat themselves in a periodic fashion. This type of Fourier
transform is called the Discrete Time Fourier Transform (DTFT).

2
4- Periodic-Discrete

These are discrete signals that repeat themselves in a periodic fashion from negative
to positive infinity. This class of Fourier Transform is sometimes called the Discrete
Fourier Series (DFS) but is most often called the Discrete Fourier Transform
(DFT).

For discrete signals we can use the discrete Fourier series (DFS), but it is applicable
only to periodic signals. Or we can use the discrete-time Fourier Transform (DTFT),
which is applicable to non-periodic signals. The spectrum of the DFS is discrete
whereas the spectrum of DTFT is continuous. As a consequence of sampling, the
output spectrums for both of these constructs are periodic.

FIGURE 6.1: Illustration of the four Fourier transforms. A signal may be continuous

or discrete, and it may be periodic or aperiodic. Together these define four possible

3
 combinations, each having its own version of the Fourier transform. The names are
not well organized; simply memorize them.

It applies to discrete time signals and to aperiodic signals. But the rub is that the
DTFT in frequency domain is continuous. Computer storage and calculation mode is
inherently discrete and DTFT is not an algorithm that can be computed using discrete
(which means essentially by computer) math. What we need is a combination of DFS
which can be discrete in frequency domain and the DTFT which is discrete in time
domain. We saw that when we compute a DTFT of a periodic signal, it is also
discrete because this form of the DTFT is same as sampled DFS coefficients. So, we
now move a new transform called the Discrete Fourier Transform (DFT). It
borrows elements from both the Fourier series and the Fourier transform. DFT was
developed after it became clear that our previous transforms fell a little short of what
was needed.

Let’s go through the path we took to get to the DFT. In the beginning there was
continuous time Fourier series. The input signal is periodic with period 𝑇 and
continuous in time, which we don’t like. But CTFS gave us a discrete frequency
spectrum which we do like. Then we let the period go to infinity. This made the
signal aperiodic which is something that we also want, this we called the Continuous-
Time Fourier Transform (CTFT). CTFT gave us a discrete frequency spectrum, also
good. But continuous time is still a problem and we want to avoid dealing with it. See
table (6.1).

4
 Table 6.1: Properties of Fourier transform input and output signals.

Then we sample the signal, get a discrete version of the signal and calculate the
Discrete-Time Fourier Series (DTFS) as in Fig. (6.2). The DTFS gives discrete
frequency spectrum but unfortunately the spectrum replicates.

Figure 6.2: Relationship of various Fourier Transforms.

5
Again, by letting the period go to infinity we get the Discrete-Time Fourier
Transform (DTFT). This gives us a continuous frequency spectrum, which is not
desirable. But we notice that if take the DTFT of a periodic signal, which has the
effect of limiting the signal to a time window, the spectrum becomes discrete. This
gives us the idea for the next development in the series, the Discrete Fourier
Transform (DFT).

6.2- Fourier Series Coefficients of Periodic Digital Signals

The spectrum of a periodic digital signal 𝑥(𝑛) sampled at a rate of 𝑓𝑠 Hz with the
fundamental period 𝑇0 = 𝑁𝑇, as shown in Fig. 6.3, where there are 𝑁 samples within
the duration of the fundamental period and 𝑇 = 1/𝑓𝑠 is the sampling period. For the
time being, we assume that the periodic digital signal is band-limited such that all
harmonic frequencies are less than the folding frequency 𝑓𝑠 = 2 so that aliasing does
not occur.

FIGURE 6.3: Periodic digital signal.

According to Fourier series analysis, the coefficients of the Fourier series expansion
of the periodic signal 𝑥(𝑛) in a complex form are

1
𝐶𝑘 = 𝑇 ∫𝑇 𝑥(𝑡)𝑒 −𝑗𝜔𝑜 𝑡 𝑑𝑡 −∞<𝑘 <∞ (6.1)
𝑜 𝑜

6
where 𝑘 is the number of harmonics corresponding to the harmonic frequency of 𝑘𝑓𝑜
and 𝑤𝑜 = 2𝜋/𝑇𝑜 and 𝑓0 = 1/𝑇0 are the fundamental frequency in radians per second
and the fundamental frequency in Hz, respectively. To apply Equation (6.1), we
substitute 𝑇𝑜 = 𝑁𝑇, 𝑤𝑜 = 2𝜋/𝑇𝑜 and approximate the integration over one period
using a summation by substituting 𝑑𝑡 = 𝑇 and 𝑡 = 𝑛𝑇. We obtain

2𝜋𝑘𝑛
1 −𝑗
𝐶𝑘 = 𝑁 ∑𝑁−1
𝑛=0 𝑥(𝑛)𝑒 𝑁 −∞< 𝑘 <∞ (6.2)

Since the coefficients 𝐶𝑘 are obtained from the Fourier Series expansion in the
complex form, the resultant spectrum 𝐶𝑘 will have two sides. There is an important
feature of Equation (6.2) in which the Fourier Series Coefficient 𝐶𝑘 is periodic of 𝑁.
We can verify this as follows:

2𝜋(𝑘+𝑁)𝑛 2𝜋𝑘𝑛
1 −𝑗 1 −𝑗
𝐶𝑘+𝑁 = 𝑁 ∑𝑁−1
𝑛=0 𝑥(𝑛)𝑒 𝑁 = 𝑁 ∑𝑁−1
𝑛=0 𝑥(𝑛)𝑒 𝑁 𝑒 −𝑗2𝜋𝑛 (6.3)

Since 𝑒 −𝑗2𝜋𝑛 = 𝑐𝑜𝑠(2𝜋𝑛) − 𝑗𝑠𝑖𝑛(2𝜋𝑛) = 1, it follows that

𝐶𝑘+𝑁 = 𝐶𝑘 (6.4)

Therefore, the two-side line amplitude spectrum |𝐶𝑘 | is periodic, as shown in Figure
6.4.

FIGURE 6.4: Amplitude spectrum of the periodic digital signal.

7
Note the following points:

a. As displayed in Figure 6.4, only the line spectral portion between the
frequency – 𝑓𝑠 /2 and frequency 𝑓𝑠 /2 (folding frequency) represents frequency
information of the periodic signal.
b. The spectral portion from 𝑓𝑠 /2 to 𝑓𝑠 is a copy of the spectrum in the negative
frequency range from −𝑓𝑠 /2 to 0 Hz due to the spectrum being periodic for
every 𝑁𝑓𝑜 Hz. Again, the amplitude spectral components indexed from 𝑓𝑠 /2 to
𝑓𝑠 can be folded at the folding frequency 𝑓𝑠 /2 to match the amplitude spectral
components indexed from 0 to 𝑓𝑠 /2 in terms of 𝑓𝑠 − 𝑓 Hz, where 𝑓 is in the
range from 𝑓𝑠 /2 to 𝑓𝑠 . For convenience, we compute the spectrum over the
range from 0 to 𝑓𝑠 Hz with nonnegative indices, that is,
2𝜋𝑘𝑛
1 −𝑗
𝐶𝑘 = 𝑁 ∑𝑁−1
𝑛=0 𝑥(𝑛)𝑒 𝑁 𝑘 = 0,1, … . , 𝑁 − 1 (6.5)

We can apply Equation (6.4) to find the negative indexed spectral values if
they are required.
c. For the 𝑘 𝑡ℎ harmonic, the frequency is
𝑓 = 𝑘𝑓𝑜 𝐻𝑧 (6.6)
The frequency spacing between the consecutive spectral lines, called the
frequency resolution, is 𝑓𝑜 Hz.

Example 6.1

The periodic signal 𝑥(𝑡) = sin (2𝜋𝑡) is sampled using the sampling rate 𝑓𝑠 = 4 Hz.
a. Compute the spectrum 𝐶𝑘 using the samples in one period.
b. Plot the two-sided amplitude spectrum |𝐶𝑘 | over the range from -2 to 2 Hz.
Solution:
a. From analog signal, we can determine the fundamental frequency 𝜔𝑜 = 2𝜋
radians per second and
𝜔𝑜 2𝜋
𝑓𝑜 = = 2𝜋 = 1 𝐻𝑧, and the fundamental period 𝑇𝑜 = 1 second.
2𝜋

8
 1
Since using the sampling interval 𝑇 = 𝑓 = 0.25 second, we get the sampled
𝑠

signal as
𝑋(𝑛) = 𝑋(𝑛𝑇) = sin(2𝜋𝑛𝑇) = sin (0.5 𝜋𝑛)

and plot the first eight samples as shown in Figure 6.5.

FIGURE 6.5: Periodic digital signal.

Choosing the duration of one period, 𝑁 = 4, we have the following sample values:

𝑥(0) = 0; 𝑥(1) = 1; 𝑥(2) = 0; 𝑎𝑛𝑑 𝑥(3) = −1

Using Equation (6.5),

3
1 1 1
𝐶0 = ∑ 𝑥(𝑛) = (𝑥 (0) + 𝑥 (1) + 𝑥(2) + 𝑥(3)) = (0 + 1 + 0 − 1) = 0
4 4 4
𝑛=0

3
1 1𝑛
−𝑗2𝜋( 4 ) 1
𝐶1 = ∑ 𝑥(𝑛) 𝑒 = (𝑥(0) + 𝑥(1) 𝑒 −𝑗𝜋⁄2 + 𝑥(2) 𝑒 −𝑗𝜋 + 𝑥(3) 𝑒 −𝑗3𝜋⁄2 )
4 4
𝑛=0

1 −2𝑗
= (𝑥(0) − 𝑗𝑥(1) − 𝑥(2) + 𝑗𝑥(3)) = 0 − 𝑗(1) − 0 + 𝑗(−1)) =
4 4
= −𝑗0.5

Similarly, we get

9
 2𝑛 3𝑛
1 −𝑗2𝜋( 4 ) 1 −𝑗2𝜋( 4 )
𝐶2 = 4 ∑3𝑘=0 𝑥(𝑛) 𝑒 = 0 , and 𝐶3 = 4 ∑3𝑛=0 𝑥(𝑘) 𝑒 = 𝑗0.5

Using periodicity, it follows that

𝐶−1 = 𝐶3 = 𝑗0.5, and 𝐶−2 = 𝐶2 = 0

b. The amplitude spectrum for the digital signal is sketched in Figure 6.6.

FIGURE 6.6: Two-sided spectrum for the periodic digital signal in Example 6.1.

As we know, the spectrum in the range of -2 to 2 Hz presents the information of the

sinusoid with a frequency of 1 Hz and a peak value of 2|𝐶| = 1, which is obtained
from converting two sides to one side by doubling the two-sided spectral value. Note
that we do not double the direct-current (DC) component, that is, 𝑐0 .

6.3- Discrete Fourier Transform Formulas

Now let us concentrate on development of the DFT. Figure 6.7 shows one way to
obtain the DFT formula.

10
 FIGURE 6.7: Development of DFT formula.

First, we assume that the process acquires data samples from digitizing the relevant
continuous signal for 𝑇0 seconds. Next, assume that a periodic signal 𝑥(𝑛) is obtained
by cascading the acquired 𝑁 data samples with the duration of 𝑇0 repetitively. Note
that we assume continuity between the 𝑁 data sample frames. Finally, we determine
the Fourier series coefficients using one-period 𝑁 data samples and Equ. (6.5). Then
we multiply the Fourier series coefficients by a factor of 𝑁 to obtain

2𝜋𝑘𝑛
−𝑗
𝑋(𝑘) = 𝑁𝐶𝑘 = ∑𝑁−1
𝑛=0 𝑥(𝑛)𝑒 𝑁 𝑘 = 0,1, … . , 𝑁 − 1
11
where 𝑋(𝑘) constitutes the DFT coefficients. Notice that the factor of 𝑁 is a constant
and does not affect the relative magnitudes of the DFT coefficients 𝑋(𝑘). As shown
in the last plot, applying DFT with 𝑁 data samples of 𝑥(𝑛) sampled at a sampling
rate of 𝑓𝑠 (sampling period is 𝑇 = 1⁄𝑓𝑠 produces 𝑁 complex DFT coefficients 𝑋(𝑘).
The index 𝑛 is the time index representing the sample number of the digital sequence,
where 𝑘 is the frequency index indicating each calculated DFT coefficient and can be
further mapped to the corresponding signal frequency in terms of Hz.

Now let us conclude the DFT definition. Given a sequence 𝑥(𝑛), 0 ≤ 𝑛 ≤ 𝑁 − 1,

its DFT is defined as

2𝜋𝑘𝑛
−𝑗
𝑋(𝑘) = ∑𝑁−1
𝑛=0 𝑥(𝑛)𝑒 𝑁 = ∑𝑁−1 𝑘𝑛
𝑛=0 𝑥(𝑛)𝑊𝑁 𝑓𝑜𝑟 𝑘 = 0,1, … . , 𝑁 − 1 (6.7)

Equ. (6.7) can be expanded as

𝑘(𝑁−1)
𝑋(𝑘) = 𝑥(0)𝑊𝑁𝑘0 + 𝑥(1)𝑊𝑁𝑘1 + 𝑥(2)𝑊𝑁𝑘2 + ⋯ + 𝑥(𝑁 − 1)𝑊𝑁 (6.8)

where the factor 𝑊𝑁 (called the twiddle factor in some textbooks) is defined as

2𝜋
2𝜋 2𝜋
𝑊𝑁 = 𝑒 −𝑗 𝑁 = cos ( 𝑁 ) − 𝑗𝑠𝑖𝑛 ( 𝑁 ) (6.9)

The inverse of the DFT is given by

2𝜋𝑘𝑛
𝑗
𝑥(𝑛) = ∑𝑁−1
𝑘=0 𝑋(𝑘)𝑒 𝑁 = ∑𝑁−1 −𝑘𝑛
𝑘=0 𝑋(𝑘)𝑊𝑁 𝑓𝑜𝑟 𝑛 = 0,1, … . , 𝑁 − 1 (6.10)

Similar to Equ. (6.7), the expansion of Equ. (6.10) leads to

1 −(𝑁−1)𝑛
𝑥(𝑛) = (𝑥(0)𝑊𝑁−0𝑛 + 𝑋(1)𝑊𝑁−1𝑛 + 𝑋(2)𝑊𝑁−2𝑛 + ⋯ + 𝑋(𝑁 − 1)𝑊𝑁 )
𝑁
𝑓𝑜𝑟 𝑛 = 0,1, … . , 𝑁 − 1 (6.11)

As shown in Figure 6.7, in time domain we use the sample number or time index 𝑛
for indexing the digital sample sequence 𝑥(𝑛). However, in the frequency domain,

12
use index 𝑘 for indexing 𝑁 calculated DFT coefficients 𝑋(𝑘). We also refer to 𝑘 as
the frequency bin number in Equations (6.7) and (6.8).

The following examples serve to illustrate the application of DFT and the inverse
DFT.

Example 6.2
Given a sequence 𝑥(𝑛) for 0 ≤ 𝑛 ≤ 3, where 𝑥(0) = 1, 𝑥(1) = 2, 𝑥(2) = 3,
𝑎𝑛𝑑 𝑥(3) = 4, evaluate its DFT 𝑋(𝑘).

Solution:
𝜋
Since 𝑁 = 4 and 𝑊4 = 𝑒 −𝑗 2 , using Equ. (6.7) we have a simplified formula,
3 3
𝜋𝑘𝑛
𝑋(𝑘) = ∑ 𝑥(𝑛)𝑊4𝑘𝑛 = ∑ 𝑥(𝑛)𝑒 −𝑗 2
𝑛=0 𝑛=0

Thus, for 𝑘 = 0
3

𝑋(0) = ∑ 𝑥(𝑛)𝑒 −𝑗0 = 𝑥(0)𝑒 −𝑗0 + 𝑥(1)𝑒 −𝑗0 + 𝑥(2)𝑒 −𝑗0 + 𝑥(3)𝑒 −𝑗0
𝑛=0
= 𝑥(0) + 𝑥(1) + 𝑥(2) + 𝑥(3)
= 1 + 2 + 3 + 4 = 10
for 𝑘 = 1
3
𝜋𝑛 𝜋 3𝜋
𝑋(1) = ∑ 𝑥(𝑛)𝑒 −𝑗 2 = 𝑥(0)𝑒 −𝑗0 + 𝑥(1)𝑒 −𝑗 2 + 𝑥(2)𝑒 −𝑗𝜋 + 𝑥(3)𝑒 −𝑗 2
𝑛=0
= 𝑥(0) − 𝑗𝑥(1) − 𝑥(2) + 𝑗𝑥(3)
= 1 − 𝑗2 − 3 + 𝑗4 = −2 + 𝑗2
for 𝑘 = 2
𝑋(2) = ∑3𝑛=0 𝑥(𝑛)𝑒 −𝑗𝜋𝑛 = 𝑥(0)𝑒 −𝑗0 + 𝑥(1)𝑒 −𝑗𝜋 + 𝑥(2)𝑒 −𝑗2𝜋 + 𝑥(3)𝑒 −𝑗3𝜋
= 𝑥(0) − 𝑥(1) + 𝑥(2) − 𝑥(3)
and for 𝑘 = 3
3𝜋𝑛 3𝜋 𝑗𝜋
𝑋(3) = ∑3𝑛=0 𝑥(𝑛)𝑒 −𝑗 2 = 𝑥(0)𝑒 −𝑗0 + 𝑥(1)𝑒 −𝑗 2 + 𝑥(2)𝑒 −𝑗3𝜋 + 𝑥(3)𝑒 −𝑗 2
= 𝑥(0) + 𝑗𝑥(1) − 𝑥(2) − 𝑗𝑥(3)
= 1 + 𝑗2 − 3 − 𝑗4 = −2 − 𝑗2

13
Example 6.3
Using the DFT coefficients 𝑋(𝑘) 𝑓𝑜𝑟 0 ≤ 𝑘 ≤ 3 computed in Example 6.2,
evaluate the inverse DFT to determine the time domain sequence 𝑥(𝑛).

Solution:

𝜋
𝑗2
Since 𝑁 = 4 and 𝑊4−1 = 𝑒 , using Equation (6.10) we achieve a simplified
formula,

𝜋𝑘𝑛
1 1
𝑥(𝑛) = 4 ∑3𝑛=0 𝑋(𝑘)𝑊4−𝑘𝑛 = 4 ∑3𝑛=0 𝑋(𝑘)𝑒 𝑗 2

14
Now we explore the relationship between the frequency 𝑘 and its associated
frequency. Omitting the proof, the calculated 𝑁 DFT coefficients 𝑋(𝑘) represent the
frequency components ranging from 0 Hz (or radians/second) to 𝑓𝑠 Hz (or
radians/second), hence we can map the frequency 𝑘 to its corresponding frequency as
follows:

𝑘𝜔𝑠
𝑤= 𝑟𝑎𝑑𝑖𝑎𝑛𝑠 𝑝𝑒𝑟 𝑠𝑒𝑐𝑜𝑛𝑑 (6.12)
𝑁

or in terms of Hz,

𝑘𝑓𝑠
𝑓= 𝐻𝑧 (6.13)
𝑁

The frequency resolution as the frequency step between two consecutive DFT
coefficients to measure how fine the frequency domain presentation and obtain

𝑤𝑠
∆𝑤 = 𝑟𝑎𝑑𝑖𝑎𝑛𝑠 𝑝𝑒𝑟 𝑠𝑒𝑐𝑜𝑛𝑑 (6.14)
𝑁

or in terms of Hz, it follows that

𝑓𝑠
∆𝑓 = 𝐻𝑧 (6.15)
𝑁

Example 6.4
In Example 6.2, given a sequence 𝑥(𝑛) for 0 ≤ 𝑛 ≤ 3, where 𝑥(0) = 1, 𝑥(1) =
2, 𝑥(2) = 3, 𝑎𝑛𝑑 𝑥(3) = 4, we computed 4 DFT coefficients 𝑋(𝑘) 𝑓𝑜𝑟 0 ≤ 𝑘 ≤
3 as 𝑋(0) = 10, 𝑋(1) = −2 + 𝑗2, 𝑋(2) = −2, and 𝑋(3) = −2 − 𝑗2. If the
sampling rate is 10 Hz,

(a) determine the sampling period, time index, and sampling time instant for a
digital sample 𝑥(3) in the time domain;
(b) determine the frequency resolution, frequency bin, and mapped frequencies
for the DFT coefficients 𝑋(1) and 𝑋(3) in the frequency domain.

15
Solution:

b. In the frequency domain, since the total number of DFT coefficients is four,
the frequency resolution is determined by
𝑓𝑠 10
∆𝑓 = = = 2.5 𝐻𝑧
𝑁 4

The frequency bin for 𝑋(1) should be 𝑘 = 1 and its corresponding frequency is
determined by

𝑘𝑓𝑠 1 x 10
𝑓= = = 2.5 𝐻𝑧
𝑁 4
Similarly, for 𝑋(3) 𝑎𝑛𝑑 𝑘 = 3,
𝑘𝑓𝑠 3 x 10
𝑓= = = 7.5 𝐻𝑧
𝑁 4

Note that from Equ. (6.4), 𝑘 = 3 is equivalent to 𝑘 − 𝑁 = 3 − 4 = −1; and

−1 x 10
𝑓 = 7.5 Hz is also equivalent to the frequency 𝑓 = = −2.5 Hz, which
4

corresponds to the negative side spectrum. The amplitude spectrum at 7.5 Hz after
folding should match the one at 𝑓𝑠 − 𝑓 = 10.0 − 7.5 = 2.5 Hz. We will apply
these developed notations in the next section for amplitude and power spectral
estimation.

16