Chapter 07
Discrete Fourier Transform
1
Discrete Fourier Transform (DFT)
• The Discrete Fourier Transform (DFT) is a computer based frequency
domain analysis.
• To perform frequency analysis on aperiodic discrete-time signal, x(n), we
convert the time domain sequence to an equivalent frequency domain
representation. We know that such a representation is given by the Fourier
transform, the output of which is X(ɷ). However, X(ɷ) is a continuous
function of frequency and therefore, it is not a computationally convenient
representation of the sequence x(n).
2𝜋𝑘
• Thus, we sample X(ɷ), at frequencies, ,where k=0,1,2….N-1
𝑁
• Such a frequency domain representation leads to the Discrete Fourier
transform (DFT ).
2
Frequency Domain Sampling: DFT
Relation between Fourier Transform and DFT:
• The Fourier Transform of an aperiodic signal, x(n) is as follows:
• Suppose that we sample X(ɷ) periodically in frequency at a spacing of δɷ
radians between successive samples. Since X(ɷ) is periodic with period 2π,
only samples in the fundamental frequency range are necessary.
• For convenience, we take N equidistant samples in the interval 0 ≤ 𝜔 ≤
2𝜋 with spacing δɷ=2 π/N , as shown in figure below.
3
Frequency Domain Sampling: DFT
• First, we consider the selection of N, the number o f samples in the
frequency domain.
• If we evaluate, X(ɷ) at kδɷ=2πk/N, we obtain
• This is the DFT of x(N).
2𝜋𝑘
• X[k]=X
𝑁
4
Formula for DFT and IDFT:
5
Inverse Discrete Fourier Transform(IDFT) and
x(n)
IDFT
IDFT
6
Inverse Discrete Fourier Transform(IDFT)
7
Inverse Discrete Fourier Transform(IDFT)
• Since xp(n) is the periodic extension of x(n) ,it is clear that x (n) can be
recovered from xp(n) if there is no aliasing in the time domain, that is, if
x(n) is time limited to less than the period N of xp(n). This situation is
illustrated in the figure on the previous slide where without loss of
generality, we consider a finite duration sequence x(n), which is non-zero
in the interval 0 ≤ 𝑛 ≤ 𝐿 − 1.
• We observe that when N≥L,
so that x(n) can be recovered from xp(n) without ambiguity.
• On the other hand, if N<L, it is not possible to recover x(n) from its
periodic extension due to time domain aliasing.
• It is important to note that there must be (N-L)zeros padded in the
reconstructed signal. It does not provide any additional information about
the spectrum, X(ɷ). The N-point DFT gives a better display.
• This situation is illustrated in the Figure in the previous slide.
8
Frequency Domain Sampling & IDFT:
9
Formula for DFT and IDFT:
10
The DFT as a Linear Transformation
• The formulas for DFT and IDFT can be expressed as:
Twiddle
Factor
11
The DFT as a Linear Transformation
12
The DFT as a Linear Transformation
• The IDFT can be calculated by simply using the formula below:
• Example 01: Compute the DFT of the four point sequence,
𝑥 𝑛 = {0,1,2,3}
13
DFT as a Linear Transformation
n
k
How to determine 𝑊𝑁 ?
𝑗2𝜋
−
We know, 𝑊𝑁 = 𝑒 𝑁
But the transformation matrix is: 𝑊𝑁 𝑘𝑛 Transformation Matrix
Now, if N=4,
14
The DFT as a Linear Transformation
• Example 01: Compute the DFT of the four point sequence,
𝑥 𝑛 = {0,1,2,3}
15
The DFT as a Linear Transformation
16
Multiplication of two DFTs and Circular
Convolution
Property Time Domain Frequency Domain
Steps for Linear Convolution Steps for Circular Convolution
[Link] the two signals in [Link] the two signals in
linear plane circular plane(rotate
anticlockwise)
[Link] one of the signals and [Link] one of the signals and
fold it. fold it (rotate clockwise)
[Link] signal shifted by one [Link] signal shifted by one
unit unit ( rotate anticlockwise)
[Link] [Link]
[Link] [Link]
17
Multiplication of two DFTs and Circular
Convolution
• Example 02:
18
Multiplication of two DFTs and Circular
Convolution
19
Multiplication of two DFTs and Circular
Convolution
20
Multiplication of two DFTs and Circular
Convolution
21
