P a g e |1 / 8

EE430 Digital Signal Processing

- Signal processing is concerned with the representation, transformation and manipulation of
signals as well as the information they contain.


Applications: Radar, sonar, acoustics, robotics, seismology, electronics etc.
Reference for DSP Fields: IEEE Transactions on Signal Processing
Goal of the Course: Learn about the fundamentals of DSP, to be able to use different tools of DSP in
different applications.












Signal Processing
Theory
Speech
Processing
Image
Processing
Array
Processing
Communications Adaptive
Processing

A/D
PC
DSP
FPGA
ASIC
D/A
Antenna
Microphone
C
Optical sensor etc.
C
Antenna
Generic
transducer
Loudspeaker
A generic structure for the implementation of a
digital signal processing system


P a g e |2 / 8

Advantages of digital systems:
a) Better accuracy
b) Identical blocks, circuits
c) Easy storage without loss (except A/D converters)
d) Low cost?
Main disadvantage is the speed of A/D converters and DSP units (BW < 100 MHz)
History of DSP:
During 1960s, computers proved to be useful in signal processing. FFT is rediscovered. (Gauss laid out
the fundamentals of FFT algorithm in 1805. Cooley-Tukey rediscovered it) Digital computers efficiently
implemented the signal processing algorithms using FFT.
The rapid development of integrated circuit technology lead to the development of powerful, smaller,
faster and cheaper computers and special-purpose digital hardware (FPGA). These systems are capable of
performing complex digital signal processing tasks which are hard or difficult for analog systems.
Limitations of Analog Signal Processing:
a) Accuracy : Not accurate enough, low precision
b) Cannot build all identical circuits due to drift and aging
c) Diagnostics is comparably harder
Digital Signal Processing Layers:
















Signal Processing Theory
High Level Implementation
(ex: MATLAB, MATHEMATICA, C, Hypersignal)
Low Level Implementation
[on DSP board(C, Assembler Codes, Object), FPGA, VLSI etc.]
P a g e |3 / 8

Signals, Systems and Signal Processing

A signal is a physical quantity that varies with time, space etc. A signal is a function of one or more
variables.






Further Classification of Signals:

Multichannel Signals:
1
2
3
( )
( ) ( )
( )
s t
s t s t
s t
| |
|
=
|
|
\ .
signals coming from an array of receivers

Multidimensional Signals:
1 1 2
( ) ( , ) s t s t t = image signal

Deterministic Signals:
Any signal that can be uniquely described by an explicit mathematical expression, a table or a rule is
called deterministic.
Random Signals:
Signals that evolve in time according to a statistical law or in an unpredictable manner.


Signal Types
Analog Signals
(continuous-time signals)
CT
(cont. amp cont. time)
Discrete-time signals
DT
(cont. amp. discrete time)
Digital signals
(discrete amp. discrete time)
P a g e |4 / 8

Discrete-time Signal (Sequence)








x[n] is a sequence of numbers which constitute discrete-time signal at different times. n is the address
where the value x[n] is stored in the computer and it should be integer. Noninteger indexed values are not
defined.
- Discrete-time signals can be real or complex.
- Discrete-time signal processing covers digital signal processing. In digital signal processing all
the operations are done digitally and the results are digital numbers.
- For a digital signal, the numbers are represented by only finite number of bits. Therefore there are
finite number of permissible values for x[n].

- For discrete-time signals, x[n] can have continuous amplitude values in discrete time.
Ex: CCD charged-coupled devices.

Certain DT Sequences

Unit-sample sequence:

0, 0
[ ]
1, 0
n
n
n
o
=
=
`
=
)


P a g e |5 / 8


Unit sample sequence is important since any discrete-time sequence can be represented by unit-sample
sequence. It does not suffer from the mathematical complications of the continuous-time impulse or
(Dirac delta function)
[ ] [ ] [ ]
k
x n x k n k o

=
=

Unit-step sequence:
1, 0
[ ]
0, 0
n
u n
n
>
=
`
<
)

0
[ ] [ ] [ ]
[ ] [ ] [ 1]
n
k k
u n k n k
n u n u n
o o
o

= =
= =
=



Exponential Sequences:
[ ]
n
x n Ao =
- If A and are real, then x[n] is real.
- If ||<1, then
0 1 [ ] decreases in time
1 0 [ ] decreases in time with alternating sign
x n
x n
o
o
< <
`
< <
)

- If A and are complex,







0 0
[ ] | || | [cos( ) sin( )]
Euler's relation: cos sin
n
j
x n A w n j w n
e j
u
o | |
u u
= + + +
= +



Frequency phase


0
0
( )
| | , | |
[ ] | || |
jw j
j w n n
e A A e
x n A e
|
|
o o
o
+
= =
=
P a g e |6 / 8

- ||=1 x[n] is a complex exponential sequence
Complex exponentials (or sinusoids)
0
[ ] (real A)
jw n
x n Ae =
- Periodicity of complex exponentials in frequency
0 0 0 0
( 2 ) 2
1
2
[ ]
cos(2 ) sin(2 ) 1
j w n jw n jw n jw n
j n
x n Ae Ae e Ae
e n j n
t t
t
t t
+
= = =
= + =


- Periodicity in time.







- There are only N distinct periodic complex exponentials which are periodic by N. They are called
as harmonically related complex exponentials.

th
2
k harmonic
[ ] , 0,1, 2,..., 1
j kn
N
k
s n e k N
t
= =

Fundamental period: N
Fundamental frequency:
0
1
f
N
=
Let x[n] be a periodic sequence with period N,
(2 1)
2
3
2
1
j n
j
j
e
e j
e j
t
t
t
+
=
=
=
1
0 0 0 0
( )
[ ] [ ],
jw n jw n N jw n jw N
x n x n N n
Ae Ae Ae e
+
= +
= =
0
0
0
rational
1
2
2
requirement for periodicity in time
jw N
e
w N k
N
w k
t
t
=
=
=
P a g e |7 / 8

2
1
0
[ ] Fourier series representation
N
j kn
N
k
k
x n c e
t

=
=

Examples:
0
0
0
2
[ ] sin(5 ), 5
5
2 2 2
is rational
5 5
w n
x n n kn n
N
w
w
t
t t
t
t t
t
= =
=
= =


2
[ ] sin(2 ), not rational, N=?
2
x n n
t
t = =

Even & Odd (Conjugate Symmetric, Conjugate Anti-symmetric) Sequences

Even(Conjugate Symmetric) Sequence:
*
[ ] [ ] x n x n =
Odd Sequence:
*
* *
*
[ ] [ ], [ ] ( )
[ ] [ ] [ }, ( ) ( ) ( )
1 1
[ ] [ ] [ ] ( ) [ ] [ ]
2 2
1
[ ] [ ] [ ] (
2
DTFT jw
jw jw jw
e o R I
DTFT jw jw jw
e R
DTFT
o I
x n x n x n X e
x n x n x n X e X e jX e
x n x n x n X e X e X e
x n x n x n jX
=
= + = +
( ( = + = +

( =

*
1
) [ ] [ ]
2
jw jw jw
e X e X e ( =


Energy and Power Signals:

Energy of a signal,
2
| [ ] | (real or complex signals)
n
E x n

=
=


2 2
5 should be integer so
5
5,10,15...
fundamental period: N=2
k N k N
N
k
= =
=
P a g e |8 / 8

If E is finite, then x[n] is an energy signal.

Average power,
2
1
lim | [ ] |
2 1
N
N
n N
P x n
N

=
=
+

Define
2
| [ ] |
1
lim
2 1
N
N
n N
N
N
E x n
P E
N
=

=
=
+

- If E is finite, P = 0
- If E is infinite, P may be finite or infinite.
- If P is finite, x[n] is a power signal.
Transformation of the independent variable (time):
Time-shift:
positive time delay
[ ] [ ]
negative time advance
k
x n x n k
k

=
`

)

1
[ ] [ 1] unit delay x n z x n

Time reversal:
{ [ ]} [ ] time reversal
{ [ ]} [ ] k>0 time delay
{ { [ ]}} { [ ]} [ ]
{ { [ ]}} { [ ]} [ ]
k
k k
k
FD x n x n
TD x n x n k
TD FD x n TD x n x n k
FD TD x n FD x n k x n k
=
=
= = +
= =


Not commutative