Advanced DSP

Lecture 1
Review
Signals & Systems
 The Delta Function
• Dirac Delta (generalized) Function

0, t  0 0, t  t0  (t )
 (t )    (t0 )  
 , t  0 , t  t0
t
For any general function:

  (t )   (t  t0 )dt  (t0 )


• Sifting Property: Any continuous function can be expressed

in terms of the delta function as:

x (t )   x( )   (t   )d

 The Delta Function
 Discrete Dirac Delta Function

0, n0 0, nk 1  [n ]

 [n ]    [n  k ]  
1, n0 1, nk

n
For any general function:
x[n ]   [n ]  x[0]   [n ]

 Sifting Property: Any sequence can be expressed in terms

of the delta function as:


x[n ]   x[k ]   [n  k ]
k  
 Complex Exponentials

• Continuous
x (t )  C  e j 0t 
f0  0 frequency
 C  cos 0t   j sin 0t  2

• Discrete
x[n ]  C  e j 0n Frequencies at
o±2n are
 C  cos 0n   j sin 0n  identical
 LTI Systems & Convolution
Satisfies linearity and time invariance

x[n] h[n] y[n]

y[n ]  x[n ]  h[n ]

 
  x[k ]  h[n  k ]   h[k ]  x[n  k ]
k   k  
• Impulse response – conv()
• Cascade system
• Convolution matrix – convmtx()
 LTI Systems
• Linear constant coefficient difference equation

N M
 d k y[n  k ]   pk x[n  k ]
k 0 k 0

Constants

• For causal systems:

N M
dk p
y[ n ]    y[n  k ]   k x[n  k ]
d
k 1 0
d
k 0 0

• Impulse response can be computed from difference

equation - impz()
 Fourier Analysis
Signals

Periodic (wo) Aperiodic

Discrete Continuous Discrete Continuous

DFS FS FT
Infinite time Finite time

DTFT DFT
 Fourier Series
A continuous time signal, periodic with T can be represented as an
infinite sum of weighted complex exponentials that are
harmonically related. This representation is called the Fourier
series representation.
  2
jk t
jk 0 t
x(t )   k
a e   ak e T
k   k  
2
1 1  jk t
 jk 0 t
ak   x(t )  e dt   x(t )  e T dt
T T
T T
1
  x(t ), e jk 0 t 
T
 Fourier Series
If the periodic signal is also discrete, then we only need finite
number of complex exponentials to represent the signal. This
representation is called Discrete Fourier Series representation.
2
jk n
x[n]   ak e jk n   ak e
0 N
k  N  k  N 
2
1 1  jk n
 jk 0 n
ak   x[n]  e
N n  N 
  x[n]  e
N n  N 
N

1
  x[n], e  jk 0 n 
N
 Fourier Transform
A continuous time, aperiodic signal can be represented as an infinite
sum of (integral) of complex exponentials that are NOT harmonically
related. This representation is called the Fourier Transform, and it is
closely related to Fourier series representation.
In fact, if the aperiodic signal is thought of as a periodic signal with an
infinite period, then the FT of this signal is the envelope of the FS
coefficients of the corresponding periodic signal. Formally,

1 j t
x(t ) 
2  X ( j  )  e d


 j t j t
X ( j )   x (t )  e d  x (t ), e 

 Discrete Time Fourier
Transform (DTFT)
A discrete time signal can be obtained from a continuous time signal
by appropriately sampling it. Therefore, DTFT of a discrete signal is
periodic replicas of the FT of the corresponding continuous time
signal. DTFT is always periodic by 2. Formally,

1 j j n
x[ n] 
2  X ( e )  e
2

X ( e j )   x[ n ] e  j n
n  
 Discrete Fourier Transform (DFT)
• The relationship between FT and FS shows itself between the DTFT and
DFT as well: DTFT is the envelope of DFT, or DFT is the sampled version
of DTFT, much like FT being the envelope of FS, and FS being sampled
version of the FT.
• DFT is used for representing finite duration discrete time sequences. The
finite duration signal is made periodic by replicating the signal, which then
samples the DTFT of the corresponding aperiodic signal. Formally,

N 1 2
1 j k
~
x [ n]  
N k 0
X [k ]  e N
~x [n], 0  n  N 1
x[n]  
2 0, otherwise
N 1 j k
~
X [k ]   x [ n]  e N
n 0
 Properties of DTFT

• Time shift
• Frequency shift
• Multiplication / convolution
• Derivative in frequency domain
• Difference in time domain
Sinc Function

sinc                     
 Weighted sincs

Sum of weighted sinc functions to create a continuous waveform

from discrete-time samples