1

Digital Signal Processing

(DSP)
DSP concerns with representing the signal by sequences of finite precision
numbers (symbols) and process them. The purpose of doing this is:
 Transform a signal into a form which is more desirable (e.g., easy
to be calculated by computer.

 Signal interpretation. In such a problem , the objective of

processing is to obtain the characteristic parameters of the
signals.

2
 Applications

1. Speech processing (speech and speaker recognition, scrambling, speech

coding, etc.)
2. Signal Detection.
3. Signal compression.
4. Spectrum Estimation.
5. Radar Detection.
6. Seismology.
7. Image Processing and Biomedical Imaging.

3
Examples
Waveform of "Assalaamu Alaikum"
1

0.5
speech signal

-0.5

-1
0 0.5 1 1.5 2 2.5 3 3.5 4
4
x 10
1500
Autocorrelation

1000

500

-500
-4 -3 -2 -1 0 1 2 3 4
4
x 10

4
original image

Noisy Image

5
Filtered Image using median filter

Filtered Image using wiener filter

6
 Original image

% Original image.
I= imread('peppers.png');
figure, imshow(I)
 Distorted image

h = fspecial('motion',50,45);
filteredRGB = imfilter(I,h);
8
 y(n)=2*cos(2*pi*60*t)+cos(2*pi*200*t)+random signal
5

y(n)

-5
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Time in seconds
|Y( )|
1.5

0.5

0
0 50 100 150 200 250 300 350 400 450 500
Frequency in Hz

9
Time-domain representation for DT signals

1 for n  1,3
A. 
x ( n)   4 for n  2
0
 elsewhere

B. x(n)  ....., 0,1, 4,1, 0,....0

Or y (n)  2,  1,1, 5, 4, 0, 3

10
Some elementary DT signals

1 for n  0
1. Unit sample sequence  ( n)  
0 for n  0

2. Unit step signal

1 for n  0
u ( n)  
0 for n  0
n for n  0
3. Unit ramp signal u r ( n)  
0 for n  0
4. Exponential signal x(n)  a n for all n

Where a could be real or complex. For convergence |a|<1.

Another way of representing the following sequence:

y (n)  2,  1,1, 5, 0, 0, 3

11
y(n)  2 (n  2)   (n  1)   (n)  5 (n  1)  3 (n  4)
In general, 
y ( n)   y (k ) (n  k )
k  
Convolution
Therefore, the unit step can be represented by:

u ( n)    (n  k )
k 0

Also  (n)  u (n)  u (n  1)

12
Classification of DT signals
1. Causal x(n)=0 for n<0,
Example
x(n)  a u (n), a  1
n

2. Sinusoidal sequence
Example x(n)= A cos(w n + theta)

3. Periodic signal x(n) = x(n + N)

Example
x(n)  A cos  
n
10
Is a periodic with a
period N=20.

4. Shifting
The sequence {y(n)} is said to be a delayed version of
{x(n)} if

y (n)  x(n  k ) where k is an integer

13
5. Even or odd (symmetric or antisymmetric )
Even when x(n) = x(-n)
Odd when x(n) = -x(-n).
For any signal x(n), xe(n) =( x(n) + x(-n))/2
And xo(n) =( x(n) - x(-n))/2

6. Random and deterministic signals.

e.g., speech, noise, and exponential for deterministic.

7. Power and energy signals.
E   | x ( k ) | 2

k  

N
P limN  2 N11  | x(k ) | 2

k  N

If P is finite, the signal is power signal. Now, over a finite interval, define
14
 N
EN   | x ( n ) |2 . Then E  limN  E N
n N

1
and P  limN  EN
2N 1

Notice, if E is finite, P=0; But if E is infinite, then the average power

may either finite or infinite.

Discrete Time System (DTS)

DTS
Excitation h(n) y (n)
x (n)

Definition: DTS is a device or algorithm that transforms the input or

excitation x(n) to an output response, y(n). Mathematically, it is defined as
A unique transformation or operator that maps an input sequence x(n) into
an output sequence, y(n). 15
 
y (n)  [ x(n)] or x(n) 
 y (n)
This means y(n) is the response of the system T to the excitation x(n).

Example : Determine the response of the following systems to the input

signal x(n) characterized by:

| n | , for  3  n  3 For the following

x ( n)   cases:
 0, otherwise
a- y ( n)  x ( n  2)
b- y ( n)  max{ x ( n  1), x ( n), x ( n  1)}
n
c- y ( n)   x(k )
k  

16
The input signal is x(n)  ..,0, 3, 2,1, 0,1, 2, 3, 0, ..

a. The output is y (n)  ..,0, 3, 2,1, 0,1, 2, 3, 0, ..

b. The output is y (n)  ..,0, 3, 3,3,2,1, 2, 3, 3, 3, 0, ..

c. The output is y (n)  ..,0, 3,5,6, 6, 7, 9,12,12,12,..

Notice in part c, the system is an accumulator that computes the

running sum of all the past input values up to present time

17
 Block Diagram Representation of DTS

1. Adder

x1 (n)
+ y(n)  x1 (n)  x2 (n)
x2 (n)
a
2. Constant multiplier x(n) y ( n)  a x ( n)
3. Signal multiplier

x1 (n) X y(n)  x1 (n)  x2 (n)

x2 (n) 18
 4. Unit delay

x(n) 1 y (n)  x(n  1)

z
Example: Sketch the block diagram of the DT system
described by the following input-output relation:

y(n)  .75 y(n  1)  .125 y(n  2)  x(n)  2 x(n  1)

1 2
z
1
x(n) + + y (n)

 .75
1
z
 .125 1
z 19
 Classification of DTS

1. Static versus Dynamic

Static (memoryless ): If the output at any instant n depends at most on the
input sample at the same time but not on the past or the future
Examples:
y ( n)  a x ( n)
y ( n)  n x ( n)
y ( n)  b x 2 ( n), etc.

Examples of dynamic systems:

y ( n)  a x ( n)  b x ( n  1)
n
y ( n)   x(n  k ), etc.
k 0

20
2. Causal versus Noncausal
The DTS is causal if its output at any time n depends only on the present and
past inputs, i.e., x(n), x(n-1), …… . Otherwise the system is noncausal.
Examples:
y ( n)  a x ( n)  b x ( n  1)
n
y ( n)   x(n  k ), etc.
k 0

3. Time-Invariant versus Time-Variant

The DTS is time-invariant if a time shift in the input results in an equivalent shift
in the output.
Examples: y ( n)  [ x( n)]
y ( n  k )  [ x(n  k )]

Example of time-variant system: y ( n)  x( n) cos (n)

21
4. Linear versus Nonlinear
The DTS is linear iff
[ a1 x1 ( n)  a2 x2 ( n)]  a1 [ x1 ( n)]  a2 [ x2 ( n)]
where a1 and a2 are constants.

Otherwise, it is nonlinear. Example: y ( n)  x 2 ( n)

For the time-invariant system, the convolution is defined as:


x ( n) * h( n)   x ( k ) h( n  k )
k  

Example: Let x(n) and h(n) be defined as:

x(n)  3 (n  1)  2 (n  2)  2 (n  3)  3 (n  4)
h(n)  4 (n)  3 (n  1)  2 (n  2)   (n  3)
22
 4

2
x(n)
0

-2
0 0.5 1 1.5 2 2.5 3 3.5 4
4
h(n)

0
0 0.5 1 1.5 2 2.5 3
20
y(n)=x(n)*h(n)

10

0
0 1 2 3 4 5 6 7
n

23
5. Stable versus Unstable
Theorem. The DTS is said to be BIBO stable if and only if every bounded
input produces a bounded output.
The input x(n) is bounded if | x(n) | Bx  

The stability requires that for every bounded input there

exists a fixed positive finite value such that
| y (n) | B y   for all n

 
| y (n) ||  h( k ) x ( n  k ) |   | h( k ) x ( n  k ) |
k   k  

 Bx  | h( k ) |
k  

Now, if the system is stable ,then
 | h(k ) || Bh  
k  
24
Therefore, | y(n) | Bx Bh  

Example: Consider the nonlinear system described by the input-output equation

y (n)  y 2 (n  1)  x(n). Let x(n)  C (n) & y (1)  0

Thus, y (0)  C , y (1)  C 2 , ..., y (n)  C 2 n

It is clear that the output is unbounded when C > 1.

Therefore, the system is BIBO unstable, since bounded input does not
give bounded output.
Interconnection of Discrete-Time Systems (DTS)

x(n) 1 2 y (n)

Cascaded Connection
25
Interconnection of Discrete-Time Systems (DTS) (cont.)

1

x(n) + y (n)

2
Parallel Connection p  1  2

Notice for cascaded Connection, c  12

Notice again, if the system is linear and time-invariant, then

21  12 26
 In this course, we will consider the linear time-invariant system. For
this system, the response is


y ( n)   x ( k ) h( n  k )
k  

Example: Determine the output y(n) of a relaxed LTI system with impulse
response

h(n)  a u (n), | a | 1 when the input x(n)  u (n)

n

From the above equation (convolution),

y (0)  1, y (1)  1  a, ....

n 1
1 a
y (n)  1  a  a  ....  a 
2 n
1 a
27
Notice that y(n) = 0 for n < 0 and

1
y ()  limn y ( n) 
1 a
Later, we will see how to use the z transform to obtain a closed form
solution to y(n) provided that we know x(n) and h(n).

So, using the convolution, we can find the response to any input
signal x(n). For the previous case let a be 0.9. The response is
shown below:

28
 unit step
2

x(n) 1

0
2 4 6 8 10 12 14 16 18 20
1
h(n)

0.5

0
0 2 4 6 8 10 12 14 16 18 20
10
y(n)

0
0 5 10 15 20 25 30
n

29
Finite Impulse Response (FIR) and Infinite Impulse Response (IIR)
The previous example of h(n) represents IIR response while, the FIR is
represented by

h( n)  0 for n  0 and n  M

The convolution for such a system is

M
y ( n)   h( k ) x ( n  k )
k 0

Notice, the output y(n) at time n is the weighted linear combination of the
present x(n) and its previous values up to x(n-M).
Notice again, that the FIR has a finite memory of length M+1, while IIR
has an infinite impulse response.

30
 Examples for the convergence of Infinite Impulse Response (IIR):

 
1
1.  h( n)   a  , if | a | 1
n

n 0 n 0 1 a

a
2.  na  (1  a) 2 ,
n
if | a | 1
n 0

The convergence is obtained from the geometric series.

Examples :

1 n 1
1.  ( )   2,
n 0 2 1  .5
 
1 n 1 n 1
2.  ( 2)   ( 2 )  (1  2 ) 0.5
n2 n 0

1 2  1 n 1 2  1 n4
3. ( )  ( )  ( )  ( )  0.5
2 n4 2 2 n4 2 31
In general, the DTS is described by the difference equation:

M N
y ( n)   b( k ) x ( n  k )   a ( k ) y ( n  k )
k 0 k 1

The current output y(n) is equal to the sum of the past outputs (i.e., from
y(n-1), .. To y(n-N)) scaled by the delay dependent feedback coefficients
a_k plus the sum of the present and past coefficients scaled by the
feedforward coefficients b_k.

Implementation of DTS

The above equation can be implemented by following block diagram:

32
 bM
1
z
b2

1
z b1

1
z
x(n) + + y (n)
b0
1
 a1 z
1
 a2 z
 aN
1
z 33