+ All Categories
Home > Documents > LN5 Frequency Domain Representation Part2 FourierTransform v2

LN5 Frequency Domain Representation Part2 FourierTransform v2

Date post: 26-Feb-2018
Category:
Upload: amino-file
View: 218 times
Download: 0 times
Share this document with a friend

of 33

Transcript
  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    1/33

    ilolu 1

    FREQUENCY DOMAIN REPRESENTATION OF LTI SYSTEMS

    DICRETE TIME FOURIER TRANSFORM (DTFT)

    EXISTENCE

    MEAN SQUARE CONVERGENCE

    FOURIER TRANSFORM OF A CONSTANT SEQUENCEFOURIER TRANSFORM OF A COMPLEX EXPONENTIAL SEQUENCE

    FOURIER TRANSFORM OF A SINUSOIDAL SEQUENCE

    FOURIER TRANSFORM OF UNIT STEP SEQUENCE

    SYMMETRY PROPERTIES OF FOURIER TRANSFORM

    REAL SEQUENCES

    FOURIER TRANSFORM THEOREMS

    FOURIER TRANSFORM PAIRS

    IDEAL LOWPASS FILTER (IMPULSE RESPONSE)MOVING AVERAGE FILTER

    EXAMPLES

    LCCDES AND FREQUENCY RESPONSE

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    2/33

    ilolu 2

    DICRETE TIME FOURIER TRANSFORM (DTFT)

    The Fourier transform of a sequence x n is defined as

    j j nn

    X e x n e

    If the FT exists (summation converges) the sequence can be obtained from its FT

    as

    1

    2

    j j n

    x n X e e d

    Fourier Transform is periodic with 2.

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    3/33

    ilolu 3

    LTI SYSTEMS

    The frequency response function, ()is the FT of the impulse response

    []

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    4/33

    ilolu 4

    EXISTENCE

    FT of a sequence []exists, i.e., []=

    converges to a continuous function of ,

    if []is absolutely summable.(sufficient condition)

    Proof: Exercise

    []= []cos sin

    =

    [] cos

    =

    [] sin

    =

    Both sums have to converge

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    5/33

    ilolu 5

    All stable LTI systems have frequency response functions.

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    6/33

    ilolu 6

    Ex: n

    x n a u n

    0 01

    if 1 or 11

    nj n j n j j

    jn n

    X e a e ae ae aae

    cos21

    1

    sincos1

    1

    1

    1

    2

    2

    2

    2

    aa

    ja

    aeeX

    j

    j

    cos1

    sintan

    sincos10

    11

    1

    a

    a

    jaa

    aeeX jj

    -4 -3 -2 -1 0 1 2 3 40.5

    1

    1.5

    2

    2.5

    3

    3.5

    4

    4.5

    5

    jX e 8.0a

    -4 -3 -2 -1 0 1 2 3 4-1

    -0.8

    -0.6

    -0.4

    -0.2

    0

    0.2

    0.4

    0.6

    0.8

    1

    4

    4

    jeX

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    7/33

    ilolu 7

    MEAN SQUARE CONVERGENCE

    Some sequences, which are not absolutely summable but square summable

    2

    n

    x n

    can still be represented by Fourier Transform, but

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    8/33

    ilolu 8

    Ex: Ideal lowpass filter.

    1

    0

    cj

    c

    H e

    Lets find h n !

    [] 1

    2

    12 ( ) sin

    Note that,

    [] sin is not absolutely summable!

    Then, one may question the Fourier transform of [], sin

    = ?

    c

    c

    jH e

    1

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    9/33

    ilolu 9

    Define =

    Even if you take M oscillations do not die to zero.

    However 2

    lim 0j j

    MM

    H e H e d

    . This is called mean square convergence.

    The oscillatory behavior aroundc

    is called the Gibbs phenomenon.

    MATLAB codeclear all; close all;

    precision =0.0001

    w = [-pi:precision:pi];

    ideaL = zeros(1,length(w));

    wc = pi/2;

    orta = round(length(w)/2);

    ideaL((orta-round(wc/precision)):(orta+round(wc/precision)))=1;

    M = 10000;

    H = 0;

    forn = -M:-1

    H = H+(sin(wc*n)/(pi*n))*exp(-i*w*n);

    end

    forn = 1:M

    H = H+(sin(wc*n)/(pi*n))*exp(-i*w*n);

    end

    H = H+(wc/pi);

    plot(w,H); hold on;

    plot(w,idea,'r')

    grid

    detail

    -4 -3 -2 -1 0 1 2 3 4-0.2

    0

    0.2

    0.4

    0.6

    0.8

    1

    1.2

    1M 2

    c

    -4 -3 -2 -1 0 1 2 3 4-0.2

    0

    0.2

    0.4

    0.6

    0.8

    1

    1.2

    2c

    4M

    -4 -3 -2 -1 0 1 2 3 4-0.2

    0

    0.2

    0.4

    0.6

    0.8

    1

    1.2

    10M 2

    c

    -4 -3 -2 -1 0 1 2 3 4-0.2

    0

    0.2

    0.4

    0.6

    0.8

    1

    1.2

    100M 2

    c

    -4 -3 -2 -1 0 1 2 3 4-0.2

    0

    0.2

    0.4

    0.6

    0.8

    1

    1.2

    1000M 2

    c

    -1.571 -1.5708-1.5706-1.5704 -1.5702 -1.57 -1.5698-1.5696 -1.5694

    0.5

    0.6

    0.7

    0.8

    0.9

    1

    1.1

    10000M

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    10/33

    ilolu 10

    FOURIER TRANSFORM OF A CONSTANT SEQUENCE

    1x n 2 2jr

    X e r

    not absolutely

    summable

    or we can write as

    202 jeX

    keeping in mind that FT is periodic with 2.

    -4 -2 2 4

    (2)

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    11/33

    ilolu 11

    FOURIER TRANSFORM OF A COMPLEX EXPONENTIAL SEQUENCE

    0j nx n e 02 2jr

    X e r

    not absolutely

    summable

    or we can write as

    202 0

    j

    eX

    keeping in mind that FT is periodic with 2

    0+4 0 -2 0 0+2 0+4

    (2)

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    12/33

    ilolu 12

    FOURIER TRANSFORM OF A SINUSOIDAL SEQUENCE

    0 001

    cos( )2

    j n j nx n n e e

    r

    jrreX

    22

    00

    not absolutely summable

    or 2000

    jeX since FT is periodic with 2

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    13/33

    ilolu 13

    FOURIER TRANSFORM OF UNIT STEP SEQUENCE

    x n u n 1

    21

    j

    jr

    X e re

    not absolutely summable

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    14/33

    ilolu 14

    SYMMETRY PROPERTIES OF FOURIER TRANSFORM

    Definitions:

    Conjugate symmetric (CS) sequence. x n x n

    Conjugate antisymmetric (CaS) sequence. x n x n

    Using the above definitions, any sequence can be written as

    e o

    x n x n x n

    where

    1

    2e

    x n x n x n

    is the CS part

    and

    12

    ox n x n x n

    is the CaS part.

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    15/33

    ilolu 15

    SYMMETRY PROPERTIES

    Fundamental relations

    Let j

    x n X e

    be a FT pair. Then, the following hold:

    jx n X e since j j n

    n

    X e x n e

    [] (

    ) since

    (

    ) []

    =

    Above yields

    jx n X e

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    16/33

    ilolu 16

    The two relations above also yield:

    1)

    Re2 2

    j j

    j

    e

    X e X ex n x nx n X e

    ( CS part of jX e )

    2)

    22Im

    **

    jjj

    o

    eXeXeX

    nxnxnxj

    ( CaS part of jX e )

    3)

    Re2 2

    j j

    j

    e

    X e X ex n x nx n X e

    ( real part of jX e )

    Therefore FT of an even seq. is real!

    4) Im2 2

    j jj

    o

    X e X ex n x nx n j X e

    ( imag. part of jX e )

    Therefore FT of an odd seq. is purely imaginary!

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    17/33

    ilolu 17

    Ex:Let []and []be two real sequences with their DTFTs()and(), respectively.Let

    [] [] []Then,

    () ( )()Note that

    ()is NOT the real part of

    ().

    However,

    () () ()2 Since

    () ()() ()() Similarly,

    (

    ) () ()

    2

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    18/33

    ilolu 18

    Ex: (contd)

    []: [1 1]

    []: [1 1]

    []: [1 1 ]

    () 1

    (

    ) 1

    () 1 () 1

    () ()2 1 () ()2

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    19/33

    ilolu 19

    REAL SEQUENCES

    Based on the above relations, for real sequences ( x n x n ):

    jj

    eXeX

    , conjugate symmetrywhich implies

    Magnitude is even.... jj eXeX

    Phase is odd... jj eXeX Real part is even.. jRjR eXeX Imaginary part is odd......................... jIjI eXeX

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    20/33

    ilolu 20

    Verification by an example

    Ex: nx n a u n 1

    11

    j

    jX e a

    ae

    a)FT is conjugate symmetric:

    1

    11

    j j

    jX e X e a

    ae

    b)Real part of FT is an even function:

    21 cos

    Re1 2 cos

    j j j

    R R

    aX e X e X e

    a a

    c) jR eX is the FT of nxe :

    2

    1 cos

    1 2 cos

    j

    R

    aX e

    a a

    1

    2

    n n

    ex n a u n a u n

    d)Imaginary part of Ft is an odd function.

    2sin

    Im1 2 cos

    j j j

    I I

    aX e X e X e

    a a

    n

    e

    x n

    1 2

    1

    n

    x n

    1 2

    01

    10 a

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    21/33

    ilolu 21

    e) jI eX is the FT of nxo :

    2

    sin

    1 2 cos

    j

    I

    aX e

    a a

    1

    2

    n n

    ox n a u n a u n

    f) Magnitude of FT is an even function:

    2

    1

    1 2 cos

    j jX e X e

    a a

    g)Phase of FT is an odd function

    jjeX

    a

    aeX

    cos1

    sintan

    1

    n

    o

    x n

    1 2

    1

    4

    -4 -3 -2 -1 0 1 2 3 40.5

    1

    1.5

    2

    2.5

    3

    3.5

    4

    4.5

    5

    jX e

    8.0a

    -4 -3 -2 -1 0 1 2 3 4-1

    -0.8

    -0.6

    -0.4

    -0.2

    0

    0.2

    0.4

    0.6

    0.8

    1

    4

    4

    jeX

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    22/33

    ilolu 22

    FOURIER TRANSFORM THEOREMS

    1,F

    j j jx n X e x n F X e X e F x n

    1) F

    j jax n by n aX e bY e Linearity

    2) 00F

    j n jx n n e X e

    Time-shift

    3)

    00

    Fjj n

    e x n X e

    Freq. shift

    4) F

    jx n X e

    Time reversal

    5) jF dX e

    n x n jd

    Differentiation in frequency

    domain

    6) F

    j jx n y n X e Y e

    Convolution

    7) 1

    2

    Fj j jy n x n w n Y e X e W e d

    Modulation, windowing

    Parsevalstheorem (prove as an exercise)

    8) 22 1

    2

    j

    n

    x n X e d

    energy of the signal

    2

    jX e

    is called the energy density spectrum.

    9) 1

    2

    j j

    n

    x n y n X e Y e d

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    23/33

    ilolu 23

    Proof of (6):

    Let

    [] [][ ]= [] []

    () ?() [][ ]=

    =

    [] [ ]

    =

    =

    [] []=

    =

    ()()

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    24/33

    ilolu 24

    Proof of (8) using (6):

    Let

    [] []

    [ ]

    = []

    []

    () ( )() ( )

    [0] []

    []

    = |[]|

    =

    [0] 12 ()

    12 ()

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    25/33

    ilolu 25

    Proof of (9):

    []

    []

    =

    12 ()

    []=

    1

    2 (

    )

    []

    = ()

    12 ()()

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    26/33

    ilolu 26

    FOURIER TRANSFORM PAIRS

    n 1

    0n n 0j ne

    1 n 2 2jr

    X e r

    nx n a u n 1a 1

    1 jae

    x n u n 1

    21

    jk

    ke

    nn a u n 1a

    2

    1

    j

    j

    ae

    ae

    1 nn a u n 1a

    n n

    a u n a u n

    2

    1

    1 jae

    2 12

    n

    n n a u n

    3

    1

    1 jae

    1 !1

    1 ! !

    nn k

    a u nk n

    11 2 1

    1 !

    nn k n k n a u n

    k

    1

    1k

    jae

    1

    sin 1sin

    n

    p

    p

    r n u n

    1r

    2 2

    1

    1 2 cos j j

    pr e r e

    show using

    nx n a u n

    sincn

    n

    1,

    0,

    cj

    c

    X e

    1, 0

    0, otherwise

    n Mx n

    / 2

    sin 1 / 2

    sin / 2

    j MM

    e

    0j n

    e 02 2

    r

    r

    0cos n

    r

    j

    r

    jrere 22

    00

    n0

    sin

    r

    j

    r

    jrejrej 22

    00

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    27/33

    ilolu 27

    Ex: IDEAL LOWPASS FILTER (IMPULSE RESPONSE)

    Plots of sin sequence

    They are infinitely long sequences in

    < <

    Plots are arbitrarily in 2 0 < < 2 0

    -20 -15 -10 -5 0 5 10 15 20-0.05

    0

    0.05

    0.1

    0.15

    0.2

    -20 -15 -10 -5 0 5 10 15 20-0.1

    -0.05

    0

    0.05

    0.1

    0.15

    0.2

    0.25

    -20 -15 -10 -5 0 5 10 15 20-0.2

    -0.1

    0

    0.1

    0.2

    0.3

    0.4

    0.5

    sin cn

    n

    4c

    c

    2c

    n

    n

    n

    6c

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    28/33

    ilolu 28

    Ex: MOVING AVERAGE FILTER

    Magnitude

    Phase

    DTFT functions are plotted in or in 20

    -4 -3 -2 -1 0 1 2 3 40

    1

    2

    3

    4

    5

    6

    -4 -3 -2 -1 0 1 2 3 40

    2

    4

    6

    8

    10

    12

    -4 -3 -2 -1 0 1 2 3 4-3

    -2

    -1

    0

    1

    2

    3

    5M

    -4 -3 -2 -1 0 1 2 3 4-3

    -2

    -1

    0

    1

    2

    3

    10M

    / 2sin 1 / 2

    sin / 2

    j MM

    e

    5M

    1M

    2

    1M

    10M

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    29/33

    ilolu 29

    M = 1

    1

    12

    h n n n 21

    1 cos2 2

    jj j

    H e e e

    1

    12

    y n x n x n

    M = 4

    1

    1 2 3 45

    y n x n x n x n x n x n

    1

    1 2 3 45

    h n n n n n n

    2 3 4

    2

    11

    5

    11 2 cos 2 cos 2

    5

    j j j j j

    j

    H e e e e e

    e

    or

    2 3 41

    15

    j j j j jH e e e e e

    5 5 5

    2 2 2

    5

    2 2 2

    1 1 1

    5 1 5

    j j j

    j

    jj j j

    e e ee

    ee e e

    2

    5sin

    1 2

    5sin

    2

    je

    -4 -3 -2 -1 0 1 2 3 40

    0.1

    0.2

    0.3

    0.4

    0.5

    0.6

    0.7

    0.8

    0.9

    1

    frequency

    |H|

    -4 -3 -2 -1 0 1 2 3 40

    0.1

    0.2

    0.3

    0.4

    0.5

    0.6

    0.7

    0.8

    0.9

    1

    frequency

    |H|

    2

    5

    4

    5

    2

    5

    4

    5

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    30/33

    ilolu 30

    Ex: Express 1j

    X e , 2 jX e , in terms of jX e , the DTFT of x n .

    1 7 7x n x n x n x n x n 7

    1

    j j j jX e X e e X e

    2 4x n x n x n 4

    2

    j j j j

    X e X e e X e

    -2 0 2 4 6 8 100

    1

    2

    3

    4

    1x n

    -2 0 2 4 6 8 100

    1

    2

    3

    4

    2x n

    -2 0 2 4 60

    0.5

    1

    1.5

    2

    2.5

    3

    3.5

    4

    x n

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    31/33

    ilolu 31

    One can also write as

    () ( ) 7() () 7() since []is real()() 7()

    ()7 7() 7()

    2 Re {()+7}()7() ( ) ( 1 4)

    ()( ) ()2cos2 2 c o s2 ()()

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    32/33

    ilolu 32

    Ex: What is the inverse DTFT, [], of () 38?From the table 1 nn a u n

    2

    1

    1 jae

    for 1a

    2 1

    18

    n

    n u n 22

    11

    8

    je

    2

    3

    12 3

    8

    n

    n u n

    3

    21

    8

    2

    1

    j

    j

    e

    e

    Why does the high frequency gain of MA filter decreaseas M increases?

    Comment on the above illustrations.

    -5 0 5 10 15 20-1

    -0.8

    -0.6

    -0.4

    -0.2

    0

    0.2

    0.4

    0.6

    0.8

    1

    -5 0 5 10 15 20-1

    -0.8

    -0.6

    -0.4

    -0.2

    0

    0.2

    0.4

    0.6

    0.8

    1

  • 7/25/2019 LN5 Frequency Domain Representation Part2 FourierTransform v2

    33/33

    LCCDEs AND FREQUENCY RESPONSE

    [] [] [] () ( )() () ()

    [ ]= [ ]

    = ()=

    ()

    =

    () = ()

    =

    () () = =

    ()


Recommended