ELEC442: DSP

DTFS, DTFT, DFS, DFT, FFT

o Introduction
o DTFS & Properties

o FT of periodic signals

o DFT & Properties: Sampling of the DTFT

o DTFT, DTFS, DFT, DFS, FFT, ZT: numerical

o Summary
Dr. Aishy Amer
Concordia University
Electrical and Computer Engineering

Figures and examples in these course slides are taken from the following sources:
•A. Oppenheim, A.S. Willsky and S.H. Nawab, Signals and Systems, 2nd Edition, Prentice-Hall, 1997
•A.V. Oppenheim, R.W. Schafer and J.R. Buck, Discrete-Time Signal Processing
•M.J. Roberts, Signals and Systems, McGraw Hill, 2004
•J. McClellan, R. Schafer, M. Yoder, Signal Processing First, Prentice Hall, 2003
•Slides 2-22 are from http://metalab.uniten.edu.my/~zainul/images/Signals&Systems
 Periodic DT Signals

 A DT signal is periodic with period

where is a positive integer if

 The fundamental period of is the

smallest positive value of for which the
equation holds
 Example:

is periodic with fundamental period

2
 Fourier representation of
signals
 The study of signals and systems using sinusoidal
representations is termed Fourier analysis, after Joseph
Fourier (1768-1830)
 The development of Fourier analysis has a long history
involving a great many individuals and the investigation of
many different physical phenomena, such as the motion
of a vibrating string, the phenomenon of heat propagation
and diffusion
 Fourier methods have widespread application beyond
signals and systems, being used in every branch of
engineering and science
 The theory of integration, point-set topology, and
eigenfunction expansions are just a few examples of
3 topics in mathematics that have their roots in the analysis
of Fourier series and integrals
 Approximation of Signals by
Sinusoids
 A signal can be approximated by a sum of many
sinusoids at harmonic frequencies of the signal f0
with appropriate amplitude and phase
 The more harmonic components are added, the
more accurate the approximation becomes
 Instead of using sinusoidal signals, mathematically,
we can use the complex exponential functions with
both positive and negative harmonic frequencies
 A Fourier representation is unique, i.e., no two
same signals in time domain give the same function
in frequency domain
4
 Overview of Fourier Analysis
Methods
Periodic in Time Aperiodic in Time
Discrete in Frequency Continuous in Frequency
Continuous  CT Fourier Series : CT - PT  DT  CT Fourier Transform: CT  CT

in Time
 x(t )e
T  j t
1
ak   x(t )e  jk0t dt X ( j )  dt
T 0 

Aperiodic in  CT InverseFourier Series : DT  CT - PT  InverseCT Fourier Transform: CT  CT


 1
Frequency
 X ( j ) e
j t
x(t )   ak e jk0t x(t )  d
k  
2 

Discrete in  DT Fourier Series DT - PN  DT - PN  DT Fourier Transform: DT  CT  P 2

Time N 1 
X [k ]   x[n]e  j0 kn X ( e j )   x[n]e
n  
 j n

n 0

 InverseDT Fourier Series DT - PN  DT - PN  InverseDT Fourier Transform: CT  P 2  DT

N 1
Periodic in 1
Frequency
x[n] 
N
 X [k ]e 
k 0
j 0 kn
x[n] 
1
2  X (e
j
) e j n d 
2

5
 Overview of Fourier Analysis
Methods
Variable Period Continuous Discrete
Frequency Frequency

DT x[n] n N  k
k  2k / N

CT x(t) t T  k
k  2k / T
• DT-FS: Discrete in time; Periodic in time; Discrete in Frequency; Periodic in Frequency
• DT-FT: Discrete in time; Aperiodic in time; Continous in Frequency; Periodic in Frequency
• CT-FS: Continuous in time; Periodic in time; Discrete in Frequency; Aperiodic in Frequency
• CT-FT: Continuous in time; Aperiodic in time; Continous in Frequency; Aperiodic in Frequency
6
 Negative frequency?

7
 Negative Frequency?

8
 Negative Frequency?

9
 Outline

o Introduction
o DTFS & properties
o DTFT of periodic signals
o DFT: Sampling of the DTFT
o DTFT, DTFS, DFT, DFS, FFT, ZT: numerical (Matlab)
o Summary

10
 Discrete-Time Fourier Series (DTFS)

 Given a periodic sequence~

x[n]with period N so that ~
x[n]  ~
x[n  rN]

~ 1 ~
 The FS can be written as x[n]   Xk e j2  / Nkn
N k
 (Recall: the FS of continuous-time periodic signals require infinite many complex exponentials)
 Not that for DT periodic signals we have

e j2 / Nk mNn  e j2 / Nkne j2mn  e j2 / Nkn

 Due to the periodicity of the complex exponential we only need N exponentials
for DT FS N 1
~ 1 ~
x[n]   Xk e j2  / Nkn
N k 0
 The FS coefficients can be obtained via

~ N 1
~
Xk    x[n]e  j2  / N kn

n0
11
 DTFS Pair

 For convenience we sometimes use WN  e j2 / N

 Analysis equation ~
N 1
Xk    ~
x[n]W kn
N
n0

~ 1 N 1 ~
 Synthesis equation x[n]   Xk WNkn
N k 0

12
 Concept of DTFS

13
 The DTFS

 Note: we could divide x[n] or X[k] by N

14
 The DTFS

15
 The DTFS

16
 The DTFS

17
 The DTFS

18
 The DTFS

19
 Example: periodic square

20
 Example: periodic square

- We know that

21
 Example: periodic square

22
 Example: periodic square

23
 Example: periodic square

 DTFS of an periodic rectangular pulse train

 The DTFS coefficients
~ 1  e  j2  / 10 k5  j4 k / 10  sin k / 2 
4
Xk   e  j2  / 10 kn
  e
n0 1  e  j2  / 10 k sink / 10 

24
 Example: periodic impulse train
~

1 n  rN
 DFS of a periodic impulse train x[n]   n  rN 
r   0 else

 Since the period of the signal is N

~ N 1
~
N 1
Xk    x[n]e  j2  / N kn
  [n]e  j2  / N kn
 e  j2  / N k 0
1
n0 n0

 We can represent the signal with the DTFS coefficients

as 
~
N 1
1
x[n]   n  rN 
r  

N k 0
e j2  / N kn

25
 Properties of DTFS
~ ~
x1 n DFS
 X1 k 
~ ~
x n
2 DFS
 X2 k 
 Linearity ~ ~
a~
x1 n  b~
x2 n  aX1 k   bX2 k 
DFS


~ ~
x n DFS
 Xk 
 Shifting ~ ~
x n  m  e  j2 km / NXk 
DFS

~
e j2 nm / N~
x n DFS
 Xk  m
 Duality

~ ~
x n DFS
 Xk 
~
Xn DFS
 N~
x  k 

26
 Properties of DTFS

27
 Summary of Properties

28
 Symmetry Properties

29
 Periodic Convolution
 Take two periodic sequences
x1 n  X 1 k 
~ ~

DFS

x2 n  X 2 k 
~ ~

DFS

 Form the product

X 3 k   X1k X 2 k 
~ ~ ~

 The periodic sequence with given DTFS can be written as

N 1
x3 n   ~
~ x1 m~
x2 n  m
m 0

 Periodic convolution is commutative

N 1
x3 n   ~
~ x2 m~
x1 n  m
30 m 0
 Periodic Convolution

31
 Outline

o Introduction
o DTFS & properties
o DTFT of periodic signals
o DFT: Sampling of the DTFT
o DTFT, DTFS, DFT, DFS, FFT, ZT: numerical (Matlab)
o Summary

32
 The DTFT
 DT Fourier Transform: • DTFT represents a DT aperiodic signal as
a sum of infinitely many complex
DT  CT  P 2
exponentials, with the frequency varying

continuously in (-π, π)
j
X (e )   x[n]e
n  
 jn
• DTFT is periodic
 only need to determine it for
 InverseDT Fourier Transform:
CT  P 2  DT
1

j jn
x[n]  X ( e ) e d
2 2

 DTFT is continuous in frequency

33
 
The DTFT X ( e j )   x[ n
n  
]e  jn

 From the numerical computation viewpoint, the

computation of DTFT by computer has several
problems:
 The summation over n is infinite
 The independent variable  is continuous

 DTFT and z-transform are not numerically

computable transforms

34
 FS versus FT

 Aperiodic signals can be viewed as a periodic signal

with an infinite period
 FS: a representation of periodic signals as a linear
combination of complex exponentials
 The FS cannot represent an aperiodic signal for all times
 FT: apply to signals that are not periodic
 The FT can represent an aperiodic signal for all time
DTFS: DTFT :
N 1 
X [k ]   x[n]e  j0 kn
DT - PN  DF - PN
j
X (e )   x[ n
n  
]e  jn
DT  CT  P 2
n 0
N 1
1

1
 X (e
j0 kn j
x[n]  X [ k ]e DF - PN  DT - PN x[n]  )e jn d CT  P 2  DT
N k 0 2 2
35
 The FT of Periodic Signals
 Periodic sequences are not absolute or square summable: no DTFT exist
 We can represent them as sums of complex exponentials: DTFS

 We can combine DTFS and DTFT

 Periodic impulse train with values proportional to DTFS coefficients

   2N X~ k     2Nk 

~
X e j 
k    

This is periodic with 2 since DTFS is periodic

36
 The FT of Periodic Signals
 The inverse transform can be written as

2 ~ 2k  jn
 


 
1 2  ~ 1 2 

2 
0 
X e j e jn d 
2 
0 

k   N
X k  

 
N 
e d
2k
 2   2k  jn N 1
1

~
  1
 X k e
~ j n

N k  
X k 0 
 

 
N


e d 
N k 0
N

   2N X~ k     2Nk 

N 1 2k ~
X e j 
 k e
1 ~ j n

 FT Pair:
~
X [ n]  X N
k    
N k 0

 Example:

37
 Example

~
 Consider the periodic impulse train p[n]   n  rN
r  

 The DTFS was calculated previously to be

~ j
   2N    2Nk 

Pe 

k   

 Therefore the FT is
~
P k  1 for all k

38
 Finite-length x[n] & Periodic Signals

 Convolve with periodic impulse train

 
x[n]  x[n]  ~
~ p[n]  x[n]   n  rN   xn  rN
r   r  

 The FT of the periodic sequence is

~ j ~
        2N    2Nk 

X e  X e j P e j  X e j

k   
2  j N  
2 k
~ j
  2k 

Xe  
k   N
X e   
N 

  
 This implies that
~  j 2Nk 
Xk   X e   
  X e j

2 k
  N

DFS coefficients of a periodic signal = equally spaced

39 samples of the FT of one period
 Finite-length x[n] & Periodic Signals

40
 Example

 Consider
1 0  n  4
x[n]  
0 else

 The FT is
sin5 / 2
 
X e j  e  j2
sin / 2
 The DFS
coefficients
~ sink / 2
Xk   e  j4 k / 10 
41 sink / 10 
 Outline

o Introduction to frequency analysis

o DTFS & properties
o DTFT of periodic signals
o DFT: Sampling of the DTFT
o DTFT, DTFS, DFT, DFS, FFT, ZT: numerical (matlab)
o Summary

42
 Sampling the DTFT:
Sampling in frequency domain


 In the DTFT
j
X (e )   x[ n
n  
]e  jn

 The summation over n is infinite

 The independent variable  is continuous

 DTFT is not numerically computable transform

 To numerically represent the continuous frequency

DTFT, we must take samples of it  DFT
43
 Sampling the DTFT:
Review to sampling
 Sampling is converting x(t) to x[n]   
x n  xc nT    n  
 T : sampling period in second; fs = 1/T : sampling frequency in Hz
 s=2fs : Sampling frequency in radian-per-second
 In frequency domain: convolution of X(jw) with an impulse train
1 
X s  j    X c  j   k s 
T k  
Creates replica of the FT of x(t); Replica are periodic with s
If s< N sampling maybe irreversible due to aliasing of images

X c  j 
-N N

X s  j  s>2
- N
3s 2s s -N N s 2s 3s
X s  j  s<2
44 3s - N
2s s -N N s 2s 3s
 Sampling the DTFT:
Sampling in frequency domain

 Consider an aperiodic x[n] with a DTFT x[n]  X e DTFT j


 Assume a sequence is obtained by sampling the DTFT
 
X k   X e j
~
  2 / N k
 
 X e j 2 / N k ; 0  k  L  1

Since the DTFT is periodic, the resulting sequence is also

periodic


~
X k  could be the DFS of a sequence
 The corresponding sequence is
N 1
~
x [ n] 
1

~
X k e j  2 / N kn
; 0  n  N  1 and 0  k  L  1
N k 0
45
 Sampling the DTFT

 
X k   X e j
~
  2 / N k

 X e j 2 / N k 
 We can also write it in terms of the z-transform


X k   X z  z e 2 / N k  X e j 2 / N k
~


 The sampling points are shown in figure

46
 Sampling the DTFT

 The only assumption made on x[n]: its DTFT exist

    xme
 N 1
X e j  jm

X k   X e j 2 / N k
~
 ~
x [ n] 
1
N
 k e j 2 / N kn
~
X
m   k 0

 Combine the equations gives

 
N 1
 j  2 / N km  j  2 / N kn
~
x [ n] 
1
  xm e e
N k 0  m   

 1 N 1 j 2 / N k n m   
  xm  e    x m  pn  m
~
m    N k 0  m
 Term in the parenthesis [] is N 1 
pn  m    n  m  rN 
1
~
N
e
k 0
j  2 / N k  n  m 

r  
  

47
x [n]  xn   n  rN  
~
 xn  rN 
r   r  
 Sampling the DTFT

FS are samples of the FT of one period

48  FS are still samples of the FT; But, one period is no longer identical to x[n]
 Sampling the DTFT
DFS coefficients of a periodic sequence obtained
through summing periodic replicas of aperiodic original
sequence x[n]
If x[n] is of finite length & we take sufficient number of
samples of its DTFT, x[n] can be recovered by
~x n 0  n  N  1
xn  
 0 else
No need to know the DTFT at all frequencies, to recover
x[n]
 DFT: Representing a finite length sequence by
samples of DTFT
49
 Sampling in the frequency domain

~
 The relationship between x[n ] and one period of x [n] in the
under-sampled case is considered a form of time domain
aliasing
 Time domain aliasing can be avoided only if x[n] has finite
length
 just as frequency domain aliasing can be avoided only for
signals being band-limited
 If x[n] has finite length N and we take a sufficient number L
of equally spaced samples of its FT, then
 the FT is recoverable from these samples
 equivalently x[n] is recoverable from ~
x [ n]
 Sufficient number L means: L>=N
 We must have at least as many frequency samples as the
50 signal’s length
 The DFT
 Consider a finite length sequence x[n] of length N
xn  0 outside of 0  n  N  1
 For x[n] associate a periodic sequence
 The DFS coefficients of the periodic sequence are samples of the

DTFT of x[n]
x n   xn  rN 
~
r  

 Since x[n] is of length N there is no overlap between terms of x[n-rN]

and we can write the periodic sequence as
x n  xn mod N   xn N 
~
 To maintain duality between time and frequency
 We choose one period of ~ Xk as the DFT of x[n]

 X k  0  k  N  1
~
X k   X k mod N  X k N 
~
X k   
51  0 else
 The DFT
N 1
X k    ~
~
x [n]e  j 2 / N kn
n 0
N 1
 Consider the DFS pair ~
x [ n] 
1
 k e j 2 / N kn
~
X
N k 0

 The equations involve only one period so we can

write  N 1
 ~ 1
 
N 1 ~
 x [n]e
X k    n 0
 j  2 / N kn 
0  k  N  1 x[n]   N 
k 0
X k e j  2 / N kn
0  k  N 1

 0 else  0 else

N 1
1 N 1
X k    x[n]e  j  2 / N kn
x[n]   X k e j 2 / N kn
n 0 N k 0
0  k  L  1, L  N 0  k  L  1, where L  N

52
 The DFT pair X k  x[n]
DFT
 DFT: x[n] finite duration

53
 DFT: Example 1

 DFT of a rect. pulse x[n], N=5

 Consider x[n] of any length L>5
 Let L=N=5
 Calculate the DFS of the
periodic form of x[n]
~ 4
Xk    e  j2 k / 5 n

n0

1  e  j2 k

1  e  j2 k / 5 
5 k  0,5,10,...

0 else
54
 DFT: Example 1

 Let L=2N=10
 We get a different set
of DFT coefficients
 Still samples of the
DTFT but in different
places
x[n] = Inverse X[k]
depends on relation L
&N
55
 DFT: Example 1
summary

 The larger the DFT size K, the more details of the INVERSE DFT,
i.e., x[n ] can be seen
56
 DFT: example 2

0  k  L  1, where L  N

57
 DFT: example 3

0  k  L  1, where L  N
58
 DFT: example 3

59
 Properties of DFT (very similar to that of DTFS)

 Linearity x1 n 

DFT
X 1 k 
x2 n 
DFT
X 2 k 
ax1 n  bx2 n  aX 1 k   bX 2 k 
DFT

xn DFT
 Xk 
 Duality
Xn DFT
 Nx k N 
60
 Example: Duality

61
 Circular Shift property

- Consider N - length x[n] defined for 0  n  N - 1

- x[n]  0 for n  0 and n  N
- For an arbitrary m, the shifted y[n]  x[n - m],
may no longer be defined over the range 0  n  N - 1
 We cannot apply linear shift :    [n - m]  

" Circular shift": y[n] must be always be in the range 0  n  N - 1

y[n]  xn  m N  , wheren N  n modulo N

- Circular shift by m is equivalent to a left circular shift by (N - m)

- A circular shift m  N is equivalent to circular shift by m N
62
 Circular Shift property
xn 
DFT
X k 
xn  mN  0  n  N -1  X k e  j 2k / N m
DFT

63
64
 Circular Shift property

65
 Circular Convolution Property
x3 [n]  x1[n] N x2 [n]

~
x [ n]

0  n  N 1

• Linear convolution: one sequence is multiplied by a time–

reversed and linearly-shifted version of the other

•Circular convolution: the second sequence is circularly time-

reversed and circularly-shifted it is called an N-point circular
66
convolution
 Circular Convolution Property

     
 Linear convolution : y[n]  x[n]* h[n], so Y e j  X e j H e j
- Circular convolution : w[n]  x[n] N h[n] , so W(k)  X(k)H(k)

   
- DFT from Y e j is Y(k)  Y e j 2 / N k , so Y (k )  X (k ) H (k )

 If x[n] and h[n] of length N , then

w[n] of length N BUT y[n] has max. length of ( 2 N-1)
 To get DFT of y[n]: form y~
x [n] a periodic sequenceof period N, then
 
w[n]  r
 y[n  rN ] 0  n  N  1
 
 0 else

67
 Circular Convolution  Linear Convolution with time aliasing
 Circular Convolution:
example 1
 Circular convolution of two
finite length sequences
x1[n]   [n  n0 ] x1[n]   [n  n0 ]

X 1[k ]  WNkn0
X 3 [k ]  WNkn0 X 2 [k ]
N 1
x3 n   x1 mx2 n  m N 
m 0

N 1
x3 n   x2 mx1 n  m N 
m 0

68
 Example 2: L=N

 Two rect. X[n]: L=N=6

1 0  n  K  1
x1 n  x2 n  
0 else

 DFT of each sequence

N 1  j 2 kn k 0
N
X 1k   X 2 k    e N

n 0 0 else

 Multiplication of DFTs
N 2 k 0
X 3 k   X 1 k X 2 k   
0 else

 Inverse DFT
N 0  n  N 1
x3 n  
69
0 else
 Example 2: L=2N

 Augment zeros to each

sequence L=2N=12
 The DFT of each
sequence 2 Lk
j
1e N
X1 k   X2 k   2 k
j
1e N

 Multiplication of DFTs
2
 j
2 Lk

1  e N 
X3 k    2 k 
1e Nj

 
70
 x[n] = Inverse DFT X[k] is not unique; depends on L and N
 Circular convolution example

71
72
 Symmetry Property

73
 Symmetry Properties

74
 Outline

o Introduction to frequency analysis

o DTFS & properties
o DTFT of periodic signals
o DFT: sampling of the DTFT
o DTFT, DTFS, DFT, DFS, FFT, ZT: numerical (matlab)
o Summary

75
 Discrete-time signal transforms

76
 Numerical Calculation of FT

1. The original signal is digitized

2. A Fast Fourier Transform (FFT) algorithm
is applied, which yields samples of the FT
at equally spaced intervals

 For a signal that is very long, e.g., a

speech signal or a music piece,
spectrogram is used
 FT over successive overlapping short
intervals

77
 Matlab examples: DTFT
 Suppose that:

 Analytically, the DTFT is

 X(ej): continuous function of 
 X(ej): periodic with period 2
 Plot it using

78
 Matlab examples: DTFT

Signal x[n] DTFT

79
 Matlab examples: DFT
 Close form X(ej) not always easy
 To plot |X(ej)|, we sampled from 0 to 2
 In code: w and X are vectors
 Small step size 0.001 to simulate continuous frequency
 Workaround: DFT
 Uniform L-samples from DTFT from 0 to 2
 Takes discrete values and returns discrete values
 No need to find |X(ej)| analytically
 Fast implementation using the fast Fourier transform (FFT)
 Matlab: fft(x,L)
• L: number of samples to take
• More L  more resolution
• Default L is N=length(x)

80
 Matlab examples: DFT

 Calculating the DFT

 Plotting the DFT against k

81
 Matlab examples: DFT

Notes:
 Default L=32 gives bad
resolution
 information lost
 x-axis not useful

Cannot find fundamental

frequency 3/8

82
 Matlab examples: DFT

 Effect of increasing L (better resolution)

• L=64

• L-128

83
 Matlab examples: DFT
 Obtaining the frequency (x-axis)

 Spike at 3/8=1.17
 Spike at 2-3/8 = 5.11
 FFT calculates from 0 to 2
 More familiar to shift using
fftshift

84
 Matlab examples: DFT

 Spikes at 3/8 and -3/8

85
 Matlab examples: DFT

 Sometimes we want
frequency in Hz

86
 Matlab examples: DFT

 |X[k]| vs. k  |X(ej)| vs.   |X(f)| vs. f

 Discrete  Continuous  Continuous
 DFT  By interpolating DFT  f = (/ 2) fs
 fs : sampling frequency
 fft values divided by N
 Peak at 0.5 (half our
87 amplitude of 1)
 Matlab examples: DFS
 No special function
 Same as DFT

 Provided signal corresponds to 1 period

88
 Matlab examples: z-Transform

 Suppose that:

89
 Matlab examples: z-Transform

90
 Matlab examples: z-Transform

91
 Matlab examples: z-Transform

 Evaluate H2(ej) directly from z-Transform

92
 Matlab examples: z-Transform

 Finding z-Transform analytically

93
 Outline

o Introduction to frequency analysis

o DTFS & properties
o DTFT & properties
o FT of periodic signals
o DTFT, DTFS, DFT, DFS, FFT, ZT: numerical (matlab)
o FTT
o Summary

94
 FFT: Fast Fourier transform

 FFT is a direct computation of the DFT

 FFT is a set of algorithms for the efficient
and digital computation of the N-point DFT,
rather than a new transform
 Use the number of arithmetic multiplications
and additions as a measure of
computational complexity

95
 FFT
1 N 1
x[n]   Xk e j2  / Nkn
N k 0
N 1
 The DFT pair was given as Xk    x[n]e  j2  / Nkn
n0
 Baseline for computational complexity:
 Each DFT coefficient requires
• N complex multiplications
• N-1 complex additions
 All N DFT coefficients require
• N2 complex multiplications
• N(N-1) complex additions
 Complexity in terms of real operations
• 4N2 real multiplications
96 • 2N(N-1) real additions
 FFT
 Most fast methods are based on symmetry
properties
 Conjugate symmetry
e j2 / Nk Nn  e j2 / NkNe j2 / Nk n  e j2 / Nkn
 Periodicity in n and k
e j2 / Nkn  e j2 / Nk nN  e j2 / Nk Nn
 The Second Order Goertzel Filter
• Approximately N2 real multiplications and 2N2 real additions
• Do not need to evaluate all N DFT coefficients
 Decimation-In-Time FFT Algorithms
 (N/2)log2N complex multiplications and additions
97
 Symmetry and periodicity of complex
exponential
 Complex conjugate symmetry

WNk [ N  n ]  WN kn  (WNkn )*  Re{WNkn}  j Im{WNkn}

 Periodicity in n and k

WNkn  WNk ( n  N )  WN( k  N ) n

 For example
Re{x[n]}Re{WNkn}  Re{x[ N  n]}Re{WNk [ N n ]}
 (Re{x[n]}  Re{x[ N  n]}) Re{WNkn}

98 The number of multiplications is reduced by a factor of 2

 Outline

o Introduction to frequency analysis

o DTFS & properties
o DTFT & properties
o FT of periodic signals
o DTFT, DTFS, DFT, DFS, FFT, ZT: numerical (matlab)
o FTT
o Summary

99
 Overview of signal transforms
Variable Period Continuous Discrete
Frequency Frequency

DT x[n] n N
 k
k  2k / N

CT x(t) t T  k
k  2k / T
• DT-FS: Discrete in time; Periodic in time; Discrete in Frequency; Periodic in Frequency
• DT-FT: Discrete in time; Aperiodic in time; Continous in Frequency; Periodic in Frequency
• CT-FS: Continuous in time; Periodic in time; Discrete in Frequency; Aperiodic in Frequency
• CT-FT: Continuous in time; Aperiodic in time; Continous in Frequency; Aperiodic in Frequency
• DFT: Discrete in time; Aperiodic in time; Discrete in Frequency; Periodic in Frequency;
finite-duration x[n]
• DFS: Discrete in time; Periodic in time (make finite-duration x[n] periodic);
100 Discrete in Frequency; Periodic in Frequency;
 Relationships between signal transforms
Sample in time
Continuous-time Sampling period = Ts Discrete-time
analog signal analog sequence
x(t) C x [n] D

Laplace Continuous z-Transform Discrete-Time Discrete

Transform Fourier Transform X(z) Fourier Transform Fourier Transform
X(f) j C X(k)
X(s) C z  re X(
s = σ+jω C  
2nk
 x [n] z -n
N 1

- j 2ft  C
 x [n] e
-j
x(t) e dt
 x [n] e
 - jn N


st n = 
x(t) e dt 
n = - n=0 D
 -  f     z   0    2 0  k  N 1
-  s  
j
ze
s = j Sample in
2f =2f
frequency,
 =  Ts,
 = 2n/N,
scale
N = Length
101 amplitude
C Continuous-variable D Discrete-variable of sequence
by 1/Ts
 Fourier versus Cosine Transform

Recall: the cosine wave starts out 1/4th later in its

period
It has an offset
 Common to measure this offset in degree or radians
 One complete period equals 360° or 2π radian
 The cosine wave thus has an offset of 90° or π/2
 This offset is called the phase of a sinusoid
 We cannot restrict a signal x(t) to start out at zero
phase or 90° phase all the time
Must determine its frequency, amplitude, and phase to
uniquely describe it at any one time instant
 With the sine or cosine transform, we are restricted to
102
zero phase or 90° phase
 DCT: One Dimensional

1 n 1
 (2t  1) f 
Xf  Cf  xt cos
 

2 t 0
 2n 
where
 1  n = size
 , f  0
Cf   2  x = signal
1, f  0 
  X = transform coefficients

103