communication systems lecture 3class.icc.skku.ac.kr/~dikim/teaching/3032/notes/eee3032p... · 2020....

21
1 Communication Systems Lecture 3 Dong In Kim School of Info/Comm Engineering Sungkyunkwan University

Upload: others

Post on 29-Jan-2021

2 views

Category:

Documents


0 download

TRANSCRIPT

  • 1

    Communication SystemsLecture 3

    Dong In KimSchool of Info/Comm Engineering

    Sungkyunkwan University

  • 2

    OutlineEnergy spectral density and autocorrelation

    For energy signalsPower spectral density and autocorrelation

    For power signalsLinear systems

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

  • 3

    Energy Spectral Density and Autocorrelation

    Recall that for energy signals2 2 21( ) ( ) ( )

    2E x d X d X f dfτ τ ω ω

    π

    ∞ ∞ ∞

    −∞ −∞ −∞= = =∫ ∫ ∫

    energy spectral density (Joules/Hz)2( ) ( )G f X f

    What is the time-domain counterpart for G(f)?[ ] [ ]1 1 * 1 1 *( ) ( ) ( ) ( ) ( )G f X f X f X f X f− − − − = = ∗ F F F F

    * *( ) ( )X f x τ↔ − ( )Note: ( ) ( )X f x τ− ↔ −

    2( ) ( ) j fX f x e dπ ττ τ∞

    −∞= ∫

    * * 2 * 2 '( ) ( ) ( ') 'j f j fX f x e d x e dπ τ π ττ τ τ τ∞ ∞

    −∞ −∞= = −∫ ∫

    Change of variable 'τ τ= −From

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

  • 4

    [ ]1 *( ) ( ) ( )G f x xτ τ− = ∗ −F

    To find , define *( ) ( )x xτ τ∗ − *( ) ( ),y xτ τ= −*

    * *

    ( ) ( ) ( ) ( ) ( ) ( )

    ( ) ( ) ( ) ( )

    x x x y x t y t dt

    x t x t dt x t x t dt

    τ τ τ τ τ

    τ τ

    −∞

    ∞ ∞

    −∞ −∞

    ∗ − = ∗ = −

    = − = +

    ∫ ∫

    Energy Spectral Density and Autocorrelation

    Time-domain autocorrelation function:* * *( ) ( ) ( ) ( ) ( ) ( ) ( )x x x t x t dt x t x t dtφ τ τ τ τ τ

    ∞ ∞

    −∞ −∞∗ − = − = +∫ ∫

    2( ) ( ) ( )G f X f φ τ= ↔ Fourier transform pair

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

  • 5

    Energy Spectral Density and Autocorrelation

    If x(t) is real:

    Autocorrelation measures the similarity between the signal and a delayed version of the signal.

    * * *( ) ( ) ( ) ( ) ( ) ( ) ( )x x x t x t dt x t x t dtφ τ τ τ τ τ∞ ∞

    −∞ −∞∗ − = − = +∫ ∫

    ( ) ( ) ( ) ( ) ( ) ( ) ( )x x x t x t dt x t x t dtφ τ τ τ τ τ∞ ∞

    −∞ −∞∗ − = − = +∫ ∫

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

  • 6

    Energy Spectral Density and Autocorrelation

    Get energy E from

    2( ) ( ) j fG f e dfπ τφ τ∞

    −∞= ∫

    2( ) ( ) ( )G f X f φ τ= ↔Relationship between G(f) and ( )φ τ

    ( )φ τ

    2 2( ) ( ) ( )E x d X f df G f dfτ τ∞ ∞ ∞

    −∞ −∞ −∞= = =∫ ∫ ∫

    Parseval’s theorem:

    (0) ( )G f df Eφ∞

    −∞= =∫0τ = ⇒

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

  • 7

    Example – a square pulse

    ( ) ?φ τ =

    a

    Ax(t)

    2A a

    τ

    t

    τ

    Ax(t-τ)

    tE=?

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

  • 8

    Example – a gated sinusoid

    Each period we get a peak (but smaller)

    ( ) ?φ τ =x(t)

    τ

    t

    x(t-τ)

    τt

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

  • 9

    Example

    ~ 0 if and are completely unrelatedas in “noise”

    What value of τ will give the biggest ?: every point lines up, all products positive

    ( )x t

    *( ) ( ) ( )x t x t dtφ τ τ∞

    −∞= −∫

    ( )x t ( )x t τ−

    ( )φ τ0τ =

    t

    ( )x t τ−

    τ

    t

    τ

    ( )φ τ

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

  • 10

    ExampleFind the autocorrelation, energy spectral density and energy when ( ) ( )tx t e u tα−=

    1( )Xj

    ωα ω

    =+

    22 2

    1 1 1( ) ( )G Xj j

    ω ωα ω α ω α ω

    = = ⋅ =+ − +

    [ ]1( ) ( )x G fφ τ −=F from tables1

    2e α τ

    α−=

    ?E =

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

  • 11

    OutlineEnergy spectral density and autocorrelation

    For energy signalsPower spectral density and autocorrelation

    For power signalsLinear systems

  • 12

    Power Spectral Density and autocorrelation

    Time-average autocorrelation function of power signals:

    * *

    * *

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

    1 ( ) ( ) lim ( ) ( )2

    T

    T T

    T

    T T

    R x t x t x t x t dtT

    x t x t x t x t dtT

    τ τ τ

    τ τ

    →∞ −

    →∞ −

    − = −

    = + = +

    Periodic signals: 0

    *

    0

    1( ) ( ) ( )T

    R x t x t dtT

    τ τ+∫In Chap 5, we will define the statistical (ensemble) average autocorrelation function for random signals, and the notation for statistical average: ( )x t

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

  • 13

    *1( ) lim ( ) ( )2

    T

    T TR x t x t dt

    Tτ τ

    →∞ −+∫

    *1Note: (0) lim ( ) ( )2

    T

    T TR x t x t dt P

    T→∞ −= =∫

    Power Spectral Density and autocorrelation

    Power spectral density (of power signals):

    [ ]

    [ ]

    2

    -1 2

    ( ) ( ) ( )

    ( ) ( ) ( )

    j f

    j f

    S f R R e d

    R S f S f e df

    π τ

    π τ

    τ τ τ

    τ

    ∞−

    −∞

    −∞

    =

    = =

    F

    F

    ( )0 ( )P R S f df∞

    −∞= = ∫ : so S(f) is the power spectral density.

    (compare with energy signals)

    DONG IN KIM

    DONG IN KIM

  • 14

    Properties of autocorrelation:21 . R (0 ) ( ) : P o w er.x t=

    R (0) R ( ) fo r a ll .τ τ≥

    Power Spectral Density and autocorrelation

    * *1 1( ) lim ( ) ( ) lim ( ) ( )2 2

    T T

    T TT TR x t x t dt x t x t dt

    T Tτ τ τ

    →∞ →∞− −− = +∫ ∫

    *2 . R ( ) ( )Rτ τ− =

    If x (t) rea l, R ( ) ( ) ( ) even .R Rτ τ τ− = ⇒23. lim ( ) ( ) if x(t) does not have periodic component.R x t

    ττ

    →∞=

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

  • 15

    OutlineEnergy spectral density and autocorrelation

    For energy signalsPower spectral density and autocorrelation

    For power signalsLinear systems

  • 16

    Linear Systems

    Linear systems:

    x(t) y(t)H

    [ ][ ] [ ]

    1 1 2 2

    1 1 2 2

    ( ) ( ) ( )

    ( ) ( )

    y t a x t a x t

    a x t a x t

    = +

    = +

    H

    H H

    Time invariant systems:[ ] [ ]0 0( ) ( ) ( ) ( )y t x t y t t x t t= ⇒ − = −H H

    Linear time invariant (LTI) systems:Both linear and time invariant

    DONG IN KIM

    DONG IN KIM

  • 17

    Linear SystemsImpulse response of an LTI system

    [ ]( ) ( )h t tδ= H

    ( ) ( ) ( )x t x t dλ δ λ λ∞

    −∞= −∫

    Any input can be written in terms of ( )tδ

    In freq domain: ( ) ( ) ( )Y f H f X f=

    x(t) y(t)h(t)

    Send x(t) to an LTI system h(t):[ ]( ) ( ) ( ) ( ) ( ) ( )

    ( ) ( )

    y t x t x t d x h t d

    x t h d

    λ δ λ λ λ λ λ

    λ λ λ

    ∞ ∞

    −∞ −∞

    −∞

    = = − = −

    = −

    ∫ ∫

    H H

    Only need to know h(t) to get y(t)

    DONG IN KIM

  • 18

    Input-Output Relationship for Spectral Density

    *( ) ( ) ( )yG f Y f Y f=* *( ) ( ) ( ) ( )X f H f X f H f=

    2 2 2( ) ( ) ( ) ( )xH f X f H f G f= =

    x(t) y(t)h(t)

    Energy signals:

    Output spectral density is a scaled version of input spectral density.

    2( ) ( ) ( )y xG f H f G f=

    The same is true for power signals, proved in Chap 5.2( ) ( ) ( )y xS f H f S f=

    ( ) ( ) ( )Y f H f X f=

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

  • 19

    Input-Output Relationship for Autocorrelation Function

    [ ]

    2

    21 1

    21 1

    ( ) ( ) ( )

    ( ) ( ) ( ) ( )

    ( ) ( )

    ( ) ( )

    y x

    y y x

    x

    h x

    G f H f G f

    R G f H f G f

    H f G f

    R R

    τ

    τ τ

    − −

    − −

    =

    =ℑ =ℑ =ℑ ⊗ℑ

    = ⊗

    x(t) y(t)h(t)

    Energy signals:

    Convolution property still holds for the autocorrelation function.The same is true for power signals.

    ( ) ( ) ( )y h xR R Rτ τ τ= ⊗

    ( ) ( ) ( )y t h t x t= ⊗

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

  • 20

    Example: Autocorrelation Function

    0

    0

    00 0

    0

    0

    /2* *

    /20

    2 2/2*

    /20

    22

    1( ) ( ) ( ) ( ) ( )

    1

    mn

    T

    x T

    m n mj j tT T Tn mT n m

    njT

    nn

    R x t x t x t x t dtT

    x x e e dtT

    x e

    π τ π

    δ

    π τ

    τ τ τ−

    −∞ ∞

    − =−∞ =−∞

    =−∞

    − = −

    =

    =

    ∑ ∑∫

    Time-average autocorrelation function of a periodic signal:

    2 2

    0

    ( ) , x n x nn n

    nS f x f P xT

    δ∞ ∞

    =−∞ =−∞

    = − = ∑ ∑

    DONG IN KIM

    DONG IN KIM

    DONG IN KIM

  • 21

    Example: Power Spectral Density

    Power spectral density of a periodic signal:

    22

    0

    22

    0 0

    ( ) ( )y nn

    nn

    nS f H f x fT

    n nx H fT T

    δ

    δ

    =−∞

    =−∞

    = −

    = −

    22

    0y n

    n

    nP x HT

    =−∞

    = ∑

    DONG IN KIM

    DONG IN KIM