Discrete Fourier Transform Prof. Siripong Potisuk.

16
Discrete Fourier Transform Prof. Siripong Potisuk

Transcript of Discrete Fourier Transform Prof. Siripong Potisuk.

Page 1: Discrete Fourier Transform Prof. Siripong Potisuk.

Discrete Fourier Transform

Prof. Siripong Potisuk

Page 2: Discrete Fourier Transform Prof. Siripong Potisuk.

Summary of Spectral Representations

Signal type

Transform Frequency Domain

CT, Periodic

Continuous-time Fourier Series

(CTFS)

Discrete Spectrum

CT, Aperiodic

Continuous-time Fourier Transform

(CTFT)

Continuous

Spectrum

DT, Aperiodic

Discrete-time Fourier Transform

(DTFT)

Continuous

Spectrum,

periodic

DT, Periodic

Discrete-time Fourier Series (DTFS)

Discrete Fourier Transform (DFT)

Discrete

Spectrum,

periodic

Page 3: Discrete Fourier Transform Prof. Siripong Potisuk.

Computation of DTFT

Computer implementation can be accomplished by:

1. Truncate the summation so that it ranges over finite limits x[n] is a finite-length sequence.2. Discretize to k

evaluate DTFT at a finite number of discrete frequencies

For an N-point sequence, only N values of frequency samplesof X(ej) at N distinct frequency points, are sufficient to determine x[n] and X(ej) uniquely.

10, Nkk

Page 4: Discrete Fourier Transform Prof. Siripong Potisuk.

Sequence Truncation

Page 5: Discrete Fourier Transform Prof. Siripong Potisuk.

Uniform Frequency Sampling

,)()( jez

j zHeH

10,][

)()(

21

0

/2

Nkenx

eXkX

N

nkjN

n

Nkj

k

N

kjzk

2expRe(z)

Im(z)

z0

z1

z2

z3

z4

z5

z6

z7

1

N = 8

Page 6: Discrete Fourier Transform Prof. Siripong Potisuk.

Discrete Fourier Transform

Let x[n] be an N-point signal, and WN be the Nth root of unity. The N-point discrete Fourier Transform of x[n],denoted X(k) = DFT{x[n]}, is defined as

N

nkjN

n

N

n

knN

enx

NkWnxkX

21

0

1

0

][

10,][)(

Page 7: Discrete Fourier Transform Prof. Siripong Potisuk.

Inverse Discrete Fourier Transform

Let X(k) be an N-point DFT sequence, and WN be theNth root of unity. The N-point inverse discrete FourierTransform of X(k), denoted x[n] = IDFT{X(k)}, isdefined as

N

nkjN

k

N

k

knN

ekXN

NnWkXN

nx

21

0

1

0

)(1

10,)(1

][

Page 8: Discrete Fourier Transform Prof. Siripong Potisuk.

Nth Root of UnitykNW

1 )4

)3

1 )2

1)

4/3

2/

4/

NN

NN

NN

NN

W

jW

W

jW

1*

2/2

)2/(

)8

)7

)6

5)

NN

kN

kN

kN

NkN

kN

NkN

WW

WW

WW

WW

N

jWN

2exp

Page 9: Discrete Fourier Transform Prof. Siripong Potisuk.

Matrix Formulation

xWX

Nx

x

x

x

WWW

WWW

WWW

NX

X

X

X

NNN

NN

NN

NNNN

NNNN

]1[

]2[

]1[

]0[

1

1

1

1111

)1(

)2(

)1(

)0(

)1)(1()1(21

)1(242

121

Page 10: Discrete Fourier Transform Prof. Siripong Potisuk.

Matrix Formulation

xWN

X

xWX

Nx

x

x

x

WWW

WWW

WWW

N

NX

X

X

X

NNN

NN

NN

NNNN

NNNN

*

1

)1)(1()1(2)1(

)1(242

)1(21

1

]1[

]2[

]1[

]0[

1

1

1

1111

1

)1(

)2(

)1(

)0(

Page 11: Discrete Fourier Transform Prof. Siripong Potisuk.

Example 4.2Define a sequence x[n] = 1, 2, 3, 4 when n = 0, 1, 2, 3,respectively. Evaluate its DFT, X(k).

Page 12: Discrete Fourier Transform Prof. Siripong Potisuk.

Example 4.3Using the result from example 4.2, evaluate the IDFT to obtain the time-domain sequence, x(n).

Page 13: Discrete Fourier Transform Prof. Siripong Potisuk.

DFT Computation Using MATLAB

fft(x) - Computes the N-point DFT of a vector x of length N

fft(x, M) - Computes the M-point DFT of a vector x of length N If N < M, x is zero-padded at the end to make it into a vector of length M If N > M, x is truncated to the first M samples

ifft(X) - Computes the N-point IDFT of a vector X of length N

ifft(X, M) - Computes the M-point IDFT of a vector X of length N If N < M, X is zero-padded at the end to make it into a vector of length M If N > M, X is truncated to the first M samples

Page 14: Discrete Fourier Transform Prof. Siripong Potisuk.

DFT Interpretation

)( spectrum Phase

|)(|N

1 spectrum Magnitude

kX

kX

DFT sample X(k) specifies the magnitude and phase angle of the kth spectral component of x[n].

The amount of power that x[n] contains at a normalizedfrequency, fk, can be determined from thepower density spectrum defined as

10,|)(|

)( 2

2

NkN

kXkPN

Page 15: Discrete Fourier Transform Prof. Siripong Potisuk.

Example 4.3Consider the sequence given below. Compute andsketch the magnitude, phase, and power densityspectra.

Page 16: Discrete Fourier Transform Prof. Siripong Potisuk.