DISCRETE FOURIER

TRANSFORM
(DFT)
Dr. Ajay Singh Raghuvanshi
Electronics & Telecommunication
Engineering, NIT, Raipur
 Recall Discrete-Time Fourier Transform:

1
𝑥[𝑛]= ∫ 𝑋 (𝑒 ) 𝑒 𝑑𝜔(synthesisequation)
𝑗 𝜔 𝑗𝜔 𝑛 IDTFT

2𝜋 2𝜋 DTFT

• Periodic by 2π hence all information lies in one period

• Not practical for (real-time) computation on a digital computer.
• Solution: Limit the extent of the summation to N points for one period (2π)
and evaluate the continuous function of frequency at N equidistant points
(Sampling in Frequency Domain) 𝑁 −1
𝑋 ( 𝑒 𝑗 𝜔 )|𝜔 = 2 𝜋 𝑘 = 𝑋 ( 𝑘)= 𝑋 𝑘= ∑ 𝑥 [ 𝑛] 𝑒 − 𝑗 2 𝜋 𝑘𝑛 / 𝑁
𝑁 𝑛=0
02/26/2025 Dr Ajay Singh Raghuvanshi 2
 𝑁−1
𝑋 (𝑘)= 𝑋 𝑘= ∑ 𝑥 [𝑛]𝑒 − 𝑗 2 𝜋 𝑘𝑛 / 𝑁
N-Point DFT
𝑛=0

The inverse transform follows from the DT Fourier Series

𝑁 −1
1
𝑥 [𝑛]=
𝑁
∑ 𝑋 [𝑘]𝑒 𝑗 2 𝜋 𝑘𝑛/ 𝑁
,𝑛=0 , 1 , ..., 𝑁 −1
N-Point IDFT
𝑘=0
The DFT pair can also be written as

02/26/2025 3
 Important features of the DFT
 A one-to-one correspondence exists between x[n] and X[k].
 The DFT is closely related to the Discrete Time Fourier Transform
sampled in frequency at
 The IDFT is closely related to the Discrete Time Fourier Series
 The DFT is the appropriate Fourier representation for digital computer
realization because it is discrete and of finite length in both the time
and frequency domains.
 There is a high-speed algorithm, called the Fast Fourier transform
(FFT) for its calculation.

02/26/2025 4
02/26/2025 5
 Problem: Sampling in the time domain
The periodic signal x(t) = sin (2πt) is sampled using the rate fs = 4 Hz.
a. Compute the spectrum Xk using the samples in one period.
b. Plot the two-sided amplitude spectrum │Xk│ over the range from −2 to 2 Hz
Here f0 is 1 Hz, as fs = 4 Hz
Choosing one period, N = 4, we have x(0) = 0;
x(1) = 1; x(2) = 0; and x(3) = −1 and x(4)=0

Similarly, c2= 0 and c3 = j0.5. Using

periodicity, it follows that c-1 = c1= - j0.5,
and c-2 = c2 =0

02/26/2025 6
 b. Plot of the two-sided amplitude spectrum │Xk│ over the range from −2 to 2
Hz

02/26/2025 7
Computation of the DFT

DFT. 𝑥[0]=1, 𝑥[1]=2, 𝑥[2]=2, 𝑥[3]=1

Given the signal x[n]= {1,2,2,1} for n=0,1,2,3 and 0 otherwise compute 4 point

3
𝑋 𝑘= ∑ 𝑥[𝑛]𝑒 − 𝑗2𝜋𝑘𝑛/4
,𝑘=0,1,2,3
02/26/2025 8
The inverse transform follows from the DT Fourier Series
𝑁 −1
1 IDFT
𝑥 [𝑛]=
𝑁
∑ 𝑋 𝑘𝑒 𝑗 2 𝜋 𝑘𝑛 / 𝑁
, 𝑛=0 , 1 ,... , 𝑁 −1
𝑘=0

Computation of the Inverse DFT

{
6 𝑘 =0
− 1− 𝑗 𝑘=1
𝑋 𝑘=
0 𝑘=2
−1+ 𝑗 𝑘=3
02/26/2025 9
 {
𝑁 −1
1

𝑘=0
∑
6
𝑥 [𝑛 ]= 𝑋 𝑘 𝑒 𝑗 2 𝜋 𝑘𝑛 / 𝑁 , 𝑛=0 , 1 ,... , 𝑁 −1
𝑁 𝑘=0

− 1− 𝑗 𝑘=1
𝑋 𝑘=
0 𝑘=2
−1+ 𝑗
02/26/2025
𝑥[3]=1
𝑘=3
𝑥[0]=1, 𝑥[1]=2, 𝑥[2]=2,
10
 𝑵 −𝟏 𝒋𝟐 𝝅𝒌𝒏
−
𝑿 [ 𝒌 ]= ∑ 𝒙 [ 𝒏 ] 𝒆 𝑵
,𝒌=𝟎,𝟏,...,𝑵 −𝟏𝑫𝑭𝑻
𝒏=𝟎 − 𝒋𝟐𝝅 /𝑵
To simplify notations, let’s define twiddle factor 𝑾 𝑵 =𝒆

IDFT

02/26/2025 Dr Ajay Singh Raghuvanshi 11

Twiddle factor
Nth root of unity

− 𝑗 2𝜋 /𝑁
𝑊 𝑁 =𝑒 , ¿

02/26/2025 12
Twiddle Factor
For N=4

02/26/2025 13
The DFT as a Linear Transformation

02/26/2025 14
02/26/2025 15
 Example
Compute the DFT of the four-point sequence x[n]={0,1,2,3}

02/26/2025 16

Properties of DFT
Linearity 𝑥1 [ 𝑛 ] 𝐷𝐹𝑇 𝑋 1 [𝑘 ]
↔
𝑥2 [ 𝑛 ] 𝐷𝐹𝑇 𝑋 2 [𝑘 ]
↔
𝑎 𝑥 1 [ 𝑛 ] +𝑏 𝑥 2 [ 𝑛 ] 𝐷𝐹𝑇 𝑎 𝑋 1 [ 𝑘 ] +𝑏 𝑋 2 [ 𝑘 ]
↔

 Duality
𝑥 [𝑛 ] 𝐷𝐹𝑇 𝑋 [ 𝑘]
↔

𝑋 [ 𝑛 ] 𝐷𝐹𝑇 𝑁𝑥 [ ( ( −𝑘 ) ) 𝑁 ]
↔
 Circular Shift of a Sequence
𝑥[𝑛] 𝐷𝐹𝑇 𝑋 [𝑘]
↔

𝑥 [ ( ( 𝑛−𝑚 ) ) 𝑁 ] 0 ≤ n ≤ N −1 𝐷𝐹𝑇 𝑋 [ 𝑘 ] 𝑒− 𝑗 2 𝜋 𝑘 / 𝑁 𝑚
( )

Copyright (C) 2005 Güner Arslan 17

Symmetry Property of DFT
 The magnitude function is even, and the phase function is odd Symmetry.
 Due to periodic nature of 2π it also has circular symmetry

𝑁−1
𝑋 𝑘= ∑ 𝑥[𝑛]𝑒 − 𝑗 2𝜋 𝑘𝑛/ 𝑁
• When N is even, |Xk| is symmetric about N/2.

𝑛=0 • The phase,  Xk, has odd symmetry about N/2.

02/26/2025 18
02/26/2025 19