Discrete-Time Fourier Transform - Lecture Notes | EECS 451, Study notes of Electrical and Electronics Engineering

Material Type: Notes; Professor: Yagle; Class: Dig Sig Proc&Analys; Subject: Electrical Engineering And Computer Science; University: University of Michigan - Ann Arbor; Term: Unknown 1989;

Typology: Study notes

Pre 2010

Uploaded on 09/02/2009

koofers-user-al2
koofers-user-al2 🇺🇸

9 documents

1 / 2

Toggle sidebar

This page cannot be seen from the preview

Don't miss anything!

bg1
EECS 451 DISCRETE-TIME FOURIER TRANSFORM
DTFT: X(e ) = P
n=−∞ x(n)ejωn =X(z)|z=ej ω (z-xform on unit circle).
Inverse: x(n) = 1
2πRπ
πX(e )ejωn . (π, π)(pπ, p +π) for any p.
Period: X(e ) is periodic with period 2π. Highest frequency: ω=π.
Dual: Fourier series: Expand X(ej ω) as a Fourier series with period 2π:
x(n)=Fourier coefficients; computed using DT F T 1formula above.
Uniform P|x(n)|<(absolutely summable) unif orm convergence:
converge lim
N→∞ max |XN(e )X(e )|= 0 where XN(ej ω) = PN
n=Nx(n)ejωn
Mean- P|x(n)|2<(finite energy)mean square convergence:
square lim
N→∞ Rπ
π|XN(e )X(e )|2 = 0. Weaker than uniform:
Sinc: x(n) = sin(Bn)
πn X(e ) = 1 if 0 |ω|< B
0 if B < |ω| π
Finite x(n) = {...0,0,3,1,4,2,5,0,0...}(x(0) = 4; finite length = 5)
length X(e ) = 3ej2ω+ 1e + 4 + 2ej ω + 5ej2ω=X(z)|z=e
signal X(e ) = [4 + 3 cos(ω) + 8 cos(2ω)] j[sin(ω) + 2 sin(2ω)]
Expo- x(n) = anu(n)X(ejω) = P
n=0 anejωn = 1/(1 ae).
nential x(n) = anu(n) + bnu(n1) X(eω) = ba
a+be abe
provided: |a|<1<|b|(stable x(n)ROC must include the unit circle |z|= 1).
DISCRETE-TIME FOURIER SERIES (DTFS)
DTFS: Xk=1
NPN1
n=0 x(n)ej2πnk/N ;x(n) = PN1
k=0 Xkej2πnk/N
Discrete+periodic in time domainDiscrete+periodic in frequency.
Basis: PN1
n=0 ej2πnk/N =nNif N divides k
0 otherwise . Orthogonal function.
Periodic: x(n), Xk, ej2πnk/N are all periodic in nand kwith periods N.
Parseval: 1
NPN1
n=0 |x(n)|2=PN1
k=0 |Xk|2=power in the periodic x(n).
Square: x(n) = 1 if 0 nL1
0 if LnN1Xk=(L
Nif N divides k; else
1
N
sin(πkL/N)
sin(πk/N)ejπk(L1)/N
Continuous L F
z=esms= m F {Px(n)δ(tn)}
Discrete Z DT F T
T ime z =e
pf2

Partial preview of the text

Download Discrete-Time Fourier Transform - Lecture Notes | EECS 451 and more Study notes Electrical and Electronics Engineering in PDF only on Docsity!

EECS 451 DISCRETE-TIME FOURIER TRANSFORM

DTFT: X(ejω^ ) =

n=−∞ x(n)e

−jωn (^) = X(z)|z=ejω (z-xform on unit circle). Inverse: x(n) = (^21) π

∫ (^) π −π X(e

jω (^) )ejωndω. (−π, π) → (p − π, p + π) for any p.

Period: X(ejω^ ) is periodic with period 2π. Highest frequency: ω = π. Dual: Fourier series: Expand X(ejω^ ) as a Fourier series with period 2π: x(n)=Fourier coefficients; computed using DT F T −^1 formula above. Uniform

|x(n)| < ∞ (absolutely summable) → unif orm convergence:

converge (^) Nlim →∞ max |XN (ejω^ )−X(ejω^ )| = 0 where XN (ejω^ ) =

∑N

n=−N x(n)e

jωn

Mean-

|x(n)|^2 < ∞ (finite energy)→ mean − square convergence: square (^) Nlim →∞

∫ (^) π −π |XN^ (e

jω (^) ) − X(ejω (^) )| (^2) dω = 0. Weaker than uniform:

Sinc: x(n) = sin( πnBn ) → X(ejω^ ) =

1 if 0 ≤ |ω| < B 0 if B < |ω| ≤ π Finite x(n) = {... 0 , 0 , 3 , 1 , 4 , 2 , 5 , 0 , 0.. .}(x(0) = 4; finite length = 5) → length X(ejω^ ) = 3ej^2 ω^ + 1ejω^ + 4 + 2e−jω^ + 5e−j^2 ω^ = X(z)|z=ejω signal X(ejω^ ) = [4 + 3 cos(ω) + 8 cos(2ω)] − j[sin(ω) + 2 sin(2ω)] Expo- x(n) = anu(n) → X(ejω^ ) =

n=0 a

ne−jωn (^) = 1/(1 − ae−jω). nential x(n) = anu(n) + bnu(−n − 1) → X(eω^ ) = (^) a+b−ejωb− (^) −aabe−jω provided: |a| < 1 < |b| (stable x(n) ⇔ROC must include the unit circle |z| = 1).

DISCRETE-TIME FOURIER SERIES (DTFS) DTFS: Xk = (^) N^1

∑N − 1

n=0 x(n)e

−j 2 πnk/N (^) ; x(n) = ∑N^ −^1 k=0 Xke

j 2 πnk/N

Discrete+periodic in time domain⇔Discrete+periodic in frequency. Basis:

∑N − 1

n=0 e

j 2 πnk/N (^) =

N if N divides k 0 otherwise

. Orthogonal function.

Periodic: x(n), Xk , ej^2 πnk/N^ are all periodic in n and k with periods N. Parseval: (^) N^1

∑N − 1

n=0 |x(n)|

2 = ∑N^ −^1

k=0 |Xk|

(^2) =power in the periodic x(n).

Square: x(n) =

1 if 0 ≤ n ≤ L − 1 0 if L ≤ n ≤ N − 1

→ Xk =

{ L

N if N divides k; else 1 N

sin(πkL/N ) sin(πk/N ) e

−jπk(L−1)/N

Continuous L ⇔ F z = es^ m s = jω m F{

x(n)δ(t − n)} Discrete Z ⇔ DT F T T ime z = ejω