A

SUMMER TRAINING REPORT

ON
Digital Signal Processing
Submitted as a partial fulllment of Bachelor of Technology
Session-2010-2011
Submitted To:- Submitted By:-
Mr. Ritesh Saraswat Vivek Chouhan
H.O.D 08EMBEC058
ECE Department Final Year ECE
Under the Guidance of Seminar Incharge
Mrs. Abha Singh
M. L. V. Textile and Engineering College.
(An Autonomous Engineering College of Govt. Of Rajasthan)
Bhilwara(Rajasthan).
Preface
Summer training is an important part of the engineering curriculum. The B.Tech.
course summer training helps a student in getting acquainted with the manner
in which his knowledge is being practically used outside his institute and this is
normally dierent from what he has learnt from books. Hence, when the student
switches from the process of learning to that of implementing his knowledge, he
nds an abrupt change.
The need of training arises for doing things yosurself, understanding its way.
Practical exposure for doing things makes a person conversant to the technicalities
involved in any job. To overcome the problem of prot and growth through the
soundest utilization of human capacities eective recruitment and selection process
in rst step. If it is not done well no amount of training, supervision or incentive
make for it. In view of such benets, imparting of vocational training has been made
as an integral part of any academic structure.
Acknowledgment
Before I get into thick of the things I would like to add a few heartfelt for the people
who were part of my training in numerous ways, people who gave unending support
right from the stage the training was conceived.
In the beginning of the report I want to give my thanks to Mr.Ritesh Saraswat
(Head of Department- Electronics Communication ) for all his encouragement and
appreciations that I have received from him.
I am thankful to all the faculty members because in the supervision, suggestions and
guidance of them I gain all the knowledge experience. It gives me immense pleasure
to acknowledge my humble, sincere gratitude to Training incharge.
I am indebted to all my elders and friends for inspiring me to have my training with
immense dedication.
Vivek Chouhan
Contents
1 Institution prole 3
1.1 Location . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 MATLAB & DSP implementation techniques 4
2.1 MATLAB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.1.1 Strengths of MATLAB . . . . . . . . . . . . . . . . . . . . . . 4
2.1.2 Weakness of MATLAB . . . . . . . . . . . . . . . . . . . . . . 4
2.1.3 MATLAB Interface . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.4 Useful Commands . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1.5 M Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.6 Functions of M-Files . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.7 Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.8 Subplots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.9 Three dimensional Plots . . . . . . . . . . . . . . . . . . . . . 9
2.1.10 Surface Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2 Theory of Digital Signal Processing . . . . . . . . . . . . . . . . . . . 11
2.3 Advantage of Digital Signal Processing over Analog Signal Processing 12
2.4 DSP Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.4.1 Analog to Digital conversion technique . . . . . . . . . . . . . 13
2.4.2 Discrete Fourier Transform . . . . . . . . . . . . . . . . . . . . 16
2.4.3 Discrete Time Fourier Transform . . . . . . . . . . . . . . . . 16
2.4.4 Fourier Transform . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.4.5 Continuous Time Fourier Transform . . . . . . . . . . . . . . . 17
2.4.6 Inverse Transforms . . . . . . . . . . . . . . . . . . . . . . . . 18
2.4.7 The Heisenberg Uncertainty Principle . . . . . . . . . . . . . . 18
2.4.8 Linear time-invariant systems (LTI) . . . . . . . . . . . . . . . 19
2.5 Filter Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.5.1 IIR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1
2.5.2 FIR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.5.3 Design of Linear Phase FIR lter based on Fourier Series
method: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.5.4 Window based Linear Phase FIR lter design . . . . . . . . . 22
2.5.5 Other Windows . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.6 DSP System Properties . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.6.1 Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.6.2 Causality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.6.3 Time-invariance . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.6.4 Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.6.5 Linearity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.7 Applications of DSP . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.7.1 Biomedical . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.7.2 Speech Applications . . . . . . . . . . . . . . . . . . . . . . . 27
2.7.3 Communication . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.7.4 Radar and Sonar . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.7.5 Image Processing . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.7.6 Multimedia . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.8 Limitations of DSP . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.8.1 Aliasing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.8.2 Limitations of DSP Frequency Resolution . . . . . . . . . . . 29
3 Conclusion 30
4 References 31
4.1 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
List of Figures
2.1 Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 Subplots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 Three dimensional plots . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.4 Surface Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.5 Digital Signal Processing . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.6 Analog to digital converter . . . . . . . . . . . . . . . . . . . . . . . . 14
2.7 Types of Sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.8 Waveform of ECG . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3
Chapter 1
Institution prole
1.1 Location
ASP Training Institute,
Nagori Garden,
Bhilwara
1.2 Field
I had completed my summer trainig in Digital Signal Processing and its applications.
1
Chapter 2
MATLAB & DSP implementation
techniques
2.1 MATLAB
MATLAB is an interactive software tool that is easy to use and yet extremely
powerful for solving science and engineering problems.
MATLAB = MATrix LABoratory.
2.1.1 Strengths of MATLAB
Relatively easy to learn. Easy to interpret and x errors.
MATLAB can be used as a calculator or a programming language.
graphics to analyze and visualize your data.
Many built-in functions and toolboxes (mathematical, statistical and engineer-
ing) that are accurate and fast.
Interfaces to external languages, such as C, C++.
2.1.2 Weakness of MATLAB
MATLAB is not a general purpose programming language.
2
More expensive than conventional Fortran or C compiler.
MATLAB is an interpreted language (making it mostly slower than a compiled
language such as C, C++)?.
MATLAB is designed for scientic computation and not suitable for other things
(such as parsing text)
2.1.3 MATLAB Interface
Command Window- You can type any command at >> prompt.
Command history- Gives a list of recent commands.
Current directory- Allows user to access les in the directory.
Work space- Allows user to access all information about variable that user dene.
2.1.4 Useful Commands
>> help abs
ABS(X) is the absolute value of the elements of
X. When X is complex, ABS(X) is the complex
modulus (magnitude) of the entries of X.
See also sign, angle, unwrap.
This is useful when you know the exact MAT-
LAB command. It will help you with the syntax etc
>> clear
This command removes all variables from the current
workspace
>> exit
This command lets you exit from MATLAB
>> a = 2; b = 3;
>> who
Your variables are:
a b
Displays all the variables you have introduced in the
workspace.
whos gives more detailed information on the
variables.
>> save My a b
Saves variables a and b to le My.mat
>> load My
Loads the variables from the le My.mat
INPUT COMMENT OUTPUT
a=2 Assigns value to a variable a=2
b=3 You can suppress screen output by b=3
d=a+b 5 Assign result of a+b to d
a+b by Defalt assignment is made to ans ans=5
1234/5678 Divide number 0.2173
2
5
Power of a number 32
pi Defalt display for 3.1416
ceil(5.4) Round towards + inity 6
log10 Lograthim to base e 2.3026
2.1.5 M Files
1. Interactive programming is useful only in the initial stages.
2. M-les are text les that contain a sequence of MATLAB commands to achieve
the required goals.
3. You can enter the commands into a le and save it. Then just type in the
le name inside the MATLAB command window. All the commands will be
executed and the results will be available in the MATLAB workspace.
2.1.6 Functions of M-Files
1. A function le is also an M-le where all variables in the function le are local.
2. MATLAB functions can return vectors, matrices.
2.1.7 Plots
MATLAB is very good at visualizing mathematical functions. Use plot command
for the basic plot.
>> x=-8:0.1:8;
>> y=sin(x);
>> z = cos(x);
>> plot(x,y,+-,x,z,o-)
>> legend(sin(X),cos(X)
Figure 2.1: Plots
2.1.8 Subplots
subplot partitions a graphics window to show many plots.
>> x=-8:0.1:8;
>> y=sin(x);
>> z = cos(x);
>> subplot
>> plot(x,y)
>> ylabel(sin(X))
>> subplot
>> plot(x,z,r)
>> xlabel(X)
>> ylabel(cos(X))
Figure 2.2: Subplots
2.1.9 Three dimensional Plots
plot3 is a three dimensional analogue of plot:
>> t = -4*pi:pi/16:4*pi;
>> x = sin(t);
>> y = cos(t);
>> z = t;
>> plot3(x,y,z)
Figure 2.3: Three dimensional plots
2.1.10 Surface Plots
It plots the surface.
>> surf(x,y,z)
>> xlabel(X)
>> ylabel(Y)
>> zlabel(Z=(X2)+(Y2))
Figure 2.4: Surface Plots
2.2 Theory of Digital Signal Processing
Digital: operating by the use of discrete signals to represent data in the form of
numbers.
Signal: a parameter (electrical quantity or eect) that can be varied in such a way
as to convey information.
Processing: a series operations performed according to programmed instructions.
Changing or analyzing information which is measured as discrete sequences of num-
bers.
Figure 2.5: Digital Signal Processing
2.3 Advantage of Digital Signal Processing over Analog Signal Process-
ing
Digital system can be simply reprogrammed for other applications / ported to dier-
ent hardware / duplicated. (Reconguring analog system means hardware redesign,
testing, verication.)
DSP provides better control of accuracy requirements. (Analog system depends on
strict components tolerance, response may drift with temperature.)
Digital signals can be easily stored without deterioration. (Analog signals are not
easily transportable and often cant be processed o-line.)
More sophisticated signal processing algorithms can be implemented. (Dicult to
perform precise mathematical operations in analog form.)
2.4 DSP Techniques
Analog signal must be converted into Digital form (Discrete) before DSP techniques
can be applied. The analog signal is basically denoted as x[t] or xa[t] because it
varied by time. The analog signal comes in form of sinusoid (sine or cosine wave).
The Analog signal is digitized by using Integrated Electronic Circuit device called
an Analog-to-Digital Converter (ADC). The output of ADC will be in the form of
binary number that represents the analog signal such as electrical voltage. The ana-
log signals are always come with noise. Thus the noise ltering is needed before the
signal goes to ADC. The ltering can be done by using DSP techniques. The special
purpose microprocessors are designed to carry out application of DSP. It is named
as Digital Signal Processors (DSPs) and used in real time application. Quantization
is the conversion of discrete-time continuous-valued to discrete-time discrete- valued
(digital) signal. The dierence of this is called Quantization Error. The coder in
ADC will convert the output of the Quantizer to b-bit binary sequence that can be
read by DSPs (Digital Signal Processors). The DAC, will perform a reverse opera-
tion of ADC in order to generate back analog signal.
2.4.1 Analog to Digital conversion technique
Analog to digital conversion technique consists of three steps to digitize an analog
signal:
1. Sampling
2. Quantization
3. Binary encoding
Before we sample, we have to lter the signal to limit the maximum frequency of the
signal as it aects the sampling rate. Filtering should ensure that we do not distort
the signal, i.e. remove high frequency components that aect the signal shape.
Figure 2.6: Analog to digital converter
Sampling
Analog signal is sampled every TS secs. Ts is referred to as the sampling interval.
fs = 1/Ts is called the sampling rate or sampling frequency. There are 3 sampling
methods:
1. Ideal - an impulse at each sampling instant.
2. Natural - a pulse of short width with varying amplitude.
3. Flattop - sample and hold, like natural but with single amplitude value
The process is referred to as pulse amplitude modulation PAM and the outcome is
a signal with analog (non integer) values.
Figure 2.7: Types of Sampling
Quantization
Sampling results in a series of pulses of varying amplitude values ranging between
two limits: a min and a max.The amplitude values are innite between the two lim-
its. We need to map the innite amplitude values onto a nite set of known values.
This is achieved by dividing the distance between min and max into L zones, each
of height
= (max - min)/L
The midpoint of each zone is assigned a value from 0 to L-1 (resulting in L values).
Each sample falling in a zone is then approximated to the value of the midpoint.
Encoding
Each zone is then assigned a binary code. The number of bits required to encode the
zones, or the number of bits per sample as it is commonly referred to, is obtained
as follows:
nb = log2L (2.1)
Given our example, nb = 3 The 8 zone (or level) codes are therefore: 000, 001, 010,
011, 100, 101, 110, and 111.
2.4.2 Discrete Fourier Transform
Like Fourier Series-There is another, only slightly dierent, way to write a discrete-
time periodic signal as a sum of complex exponentials of frequency 0k.WecallitthediscreteFouriertransform(DFT), butitisverysimilartothediscreteFourierseries(DFS).
X

k
=
p1

n=0
x(n)e
ik
0
n
(2.2)
x(n) =
1
p
p1

k=0
X

k
e
ik
0
n
(2.3)
We simply move the 1/p term outside the sum. The terms of the DFT are thus
similar to the DFS-
Xk = p Xk They also give an idea of the relative scaling of the frequency com-
ponents.
2.4.3 Discrete Time Fourier Transform
Note that for a signal with period p, when determining Xk we just need to sum
over 1 period and start the sum at any point
X() =

n=
x(n)e
in
(2.4)
Though the DFT is not dened for aperiodic signals, we have a name for the DFT
equation extended to an innite sum-
X() =

n=
x(n)e
in
(2.5)
2.4.4 Fourier Transform
The Fourier series represents a periodic signal in terms of frequency components:
x(t) =

k=
X
k
e
ik
0
t
(2.6)
x(n) =
p1

k=0
X
k
e
ik
0
n
(2.7)
We get the Fourier series coecients as follows:
X
k
=
1
p
_
p
0
x(t)e
ik
0
t
dt (2.8)
X
k
=
1
p
p1

n=0
x(n)e
ik
0
n
(2.9)
The complex exponential Fourier coecients are a sequence of complex numbers
representing the frequency component 0k.
2.4.5 Continuous Time Fourier Transform
We can extend the formula for continuous-time Fourier series coecients for a peri-
odic signal,
X
k
=
1
p
_
p
0
x(t)e
ik
0
t
dt (2.10)
to aperiodic signals as well. The continuous-time Fourier series is not dened for
aperiodic signals, but we call the formula
X() =
_

x(t)e
it
dt (2.11)
2.4.6 Inverse Transforms
If we have the full sequence of Fourier coecients for a periodic signal, we can re-
construct it by multiplying the complex sinusoids of frequency 0k by the weights
Xk and summing:
x(t) =

k=
X
k
e
ik
0
t
(2.12)
x(n) =
p1

k=0
X
k
e
ik
0
n
(2.13)
We can perform a similar reconstruction for aperiodic signals-
x(t) =
1
2
_

X()e
it
d (2.14)
x(n) =
1
2
_

X()e
in
d (2.15)
These are called the inverse transforms.
2.4.7 The Heisenberg Uncertainty Principle
The delta functions are localized in time; they are nonzero at just one point and
zero everywhere else.But the frequency spread of the delta functions is not local-
ized. We showed that X() is always 1; it never dies out.
For sinusoids, the opposite is true. They never die out in time, but the frequency
spread is just one point. The pulse function was somewhat localized in time, and
somewhat localized in frequency (the sinc function dies out asymptotically).This
is the Heisenberg uncertainty principle: the product of the time spread and fre-
quency spread of a function can never be less than a dened minimum nonzero
value.
We know we can represent functions in terms of frequency components (sinusoids).
These basis functions are nonzero at single points in the frequency domain, but
never die out in the time domain.We can also represent functions in the time do-
main. Using the sifting property, we can represent any function in terms of deltas.
For example, imagine every discrete time function as a train of appropriately scaled
Kronecker deltas. These basis functions are single points in time, never dying out in
frequency.We can also represent functions in terms of other basis functions, some-
what localized in time and frequency, like the pulse and sinc.
These functions are referred to as wavelets, and they form time-frequency rep-
resentation of a signal.
2.4.8 Linear time-invariant systems (LTI)
Let hk[n] be the response to d[n-k] (an impulse at n = k)
y[n] = T {x[n]}
= T
_

k=
x[k] [n k]
_
(2.16)
y[n] =

x[k] T {[n k]}

(2.17)
2.5 Filter Design
Two important classes of lters based on impulse response.
1. Finite Impulse Response (FIR)
2. Innite Impulse Response (IIR)
Expressing lter functions.System function representation;
H(z) =

M
k=0
b
k
z
k
1 +

N
k=1
a
k
z
k
(2.18)
Dierence Equation representation;
N

k=0
a
k
y(n k) =
M

k=0
b
k
x(n k) (2.19)
Each of this form allows various methods of implementation.The eq can be viewed
as a computational procedure (an algorithm) for determining the output sequence
y(n) of the system from the input sequence x(n).
Dierent realizations possible with dierent arrangements of eq.
Filter Design Issues:
1. Realizable
2. Stable
3. Sharp Cuto Characteristics
4. Minimum order
5. Generalized procedure
6. Linear phase characteristics
7. Issues considered for lter implementation:
8. Simple design
9. Structuredness modularity
10. Generalization of design any lter type.
11. Cost of implementation.
12. Software/hardware realization.
2.5.1 IIR
Out put is a function of past o/p, present and past i/ps, Recursive nature, Poles
and zeros, Sharp cuto chas. achievable with min order.
Dicult to have linear phase chas over full range of freq.
2.5.2 FIR
Advantages:-
1. Inherently Stable
2. Linear phase characteristics possible
3. Simple implementation both recursive and nonrecursive structures possible
4. Free of limit cycle oscillations when implemented on a nite-word length digital
system
Disadvantages:-
1. Sharp cuto at the cost of higher order
2. Higher order leading to more delay, more memory and higher cost of implemen-
tation.
Methods of designing FIR lters:-
1. Fourier series based method
2. Window based method
3. Frequency sampling method.
2.5.3 Design of Linear Phase FIR lter based on Fourier Series method:
Motivation: Since the desired freq response Hd(ej) is a periodic function in with
period 2, it can be expressed as fourier series expansion.
H
d
(e
j
) =

n=
h
d
(n)e
jn
(2.20)
where h
d
(n) are fourier series coefficients
h
d
(n) =
1
2
_

H
d
(e
j
)e
jn
d
This expansion results in impulse response coecients which are innite in dura-
tion and non causal. It can be made nite duration by truncating the innite length.
The linear phase can be obtained by introducing symmetric property in the lter
impulse response, i.e., h(n) = h(-n)
It can be made causal by introducing sucient delay (depends on lter length).
Stepwise procedure: From the desired freq response using inverse FT relation ob-
tain hd(n) Changes the innite length of the impulse response to nite length with
( assuming M odd).
h(n) = h
d
(n) for (M 1)/2 n (M 1)/2 = 0 otherwise (2.21)
Introduce h(n) = h(-n) for linear phase characteristics.
Write the expression for H(z); this is non-causal realization.
To obtain causal realization H(z) = z -(M-1)/2 H(z)
H
d
(e
j
) = 1 for

2
||

2
= 0 for

2
||
(2.22)
2.5.4 Window based Linear Phase FIR lter design
The arbitrary truncation of impulse response obtained through inverse Fourier rela-
tion can lead to distortions in the nal frequency response. The arbitrary truncation
is equivalent to multiplying innite length function with nite length rectangular
window, i.e.
h(n) = hd(n)w(n) (2.23)
where w(n) = 1 for n = (M-1)/2
H(e
j
) = Hd(e
j
) W(e
j
) (2.24)
where W(e j ) is the FT of window function w(n).
The FT of w(n) is given by
W(e
j
) =
sin(M/2)
sin(/2)
(2.25)
Suppose the lter to be designed is Low pass lter then the convolution of ideal
lter freq response and window function freq response results in distortion in the
resultant lter freq response. The ideal sharp cuto chars are lost and presence of
ringing eect is seen at the band edges which is referred to Gibbs Phenomenon.
This is due to main lobe width and side lobes of the window function freq response.
The main lobe width introduces transition band and side lobes results in rippling
characters in pass band and stop band.Smaller the main lobe width smaller will be
the transition band.
The ripples will be of low amplitude if the peak of the rst side lobe is far be-
low the main lobe peak.How to reduce the distortions?
Increase length of the window as M increases the main lob width becomes nar-
rower, hence the transition band width is decreased With increase in length the side
lobe width is decreased but height of each side lobe increases in such a manner that
the area under each sidelobe remains invariant to changes in M. Thus ripples and
ringing eect in pass-band and stop-band are not changed.
Choose windows which tapers o slowly rather than ending abruptly. Slow tapering
reduces ringing and ripples but generally increases transition width since main lobe
width of these kind of windows are larger.
Ideal window characteristics- Window having very small main lobe width with most
of the energy contained with it (i.e.,ideal window freq response must be impulsive).
1. Window design is a mathematical problem
2. More complex the window lesser are the distortions
3. Windows better than rectangular window are, Hamming, Hanning, Blackman,
Bartlett, Traingular,Kaiser.
4. Rectangular window
2.5.5 Other Windows
Henning windows
w
han
(n) = 0.5(1 cos
2n
M 1
) for 0 n M 1 (2.26)
Hamming windows
w
ham
(n) = 0.54 0.46 cos
2n
M 1
for 0 n M 1 (2.27)
Blackman windows
w
blk
(n) = 0.42 0.5 cos
2n
M 1
+ 0.08 cos
4n
M 1
for 0 n M 1 (2.28)
Bartlett (Triangular) windows
w
bart
(n) = 1
2|n
M1
2
|
M 1
for 0 n M 1 (2.29)
Kaiser windows
w
k
(n) =
I
0
_

_
_
M1
2
_
2

_
n
M1
2
_
2
_
I
0
_

_
M1
2
_ for 0 n M 1 (2.30)
2.6 DSP System Properties
Memory
Causality
Stability
Time invariance
Linearity
2.6.1 Memory
A system is memoryless if y[n] = f ( x[n] )i.e. it sees only present values.
A system has memory if y [n] depends on previous values.
it can also depend on present and future values.
2.6.2 Causality
A system is causal if the output y[n] depends only on present and/or past values.
On-line systems are causal by denition.
2.6.3 Time-invariance
A system is time invariant if a shift in the input causes a corresponding shift of
the output.
For all n0 x1 [n] = x [n-n0] gives y1[n] = y [n-n0.
2.6.4 Stability
A system is stable if every bounded input sequence produces a bounded output
i.e. it never diverges.
2.6.5 Linearity
linear systems obey the principle of superposition.
Additive property.
T{x1[n] + x2[n]} = T{x1[n]} + T{x2[n]} = y1[n] + y2[n] (2.31)
Scaling property
T{ax1[n]} = aT{x1[n]} = ay[n] (2.32)
Altogether:
T{ax1[n] + bx2[n]} = aT{x1[n]} + bT{x2[n]} (2.33)
More generally, If x[n] = k ak xk[n] then y[n] = k ak yk[n] where yk[n] is
the system response to the input xk[n]
A sequence can be represented as a linear combination of delayed impulses:
x[n] =

k=
x[k] [n k] (2.34)
2.7 Applications of DSP
It has so many applications in each and every eld these days like,Biomedical.
2.7.1 Biomedical
Analysis of biomedical signals, diagnosis, patient monitoring, preventive health
care, articial organs.
Electrocardiogram (ECG) signal provides doctor with information about the
condition of the patients heart.
Figure 2.8: Waveform of ECG
Electroencephalogram (EEG) signal provides Information about the activity
of the brain.
2.7.2 Speech Applications
Examples
Noise reduction Reducing background noise in the sequence produced by a
sensing device (microphone).
Speech recognition Dierentiating between various speech sounds.
Synthesis of articial speech Text to speech systems for blind.
Speech Recognization
Applications
Command and control of appliances and equipment.
Telephone assistance systems.
Data entry.
Speech controlled toys.
Speech and voice recognition security systems.
To control and command an appliance (computer, VCR, TV security system,
etc.) by speaking to it, will make it easier, while increasing the eciency and
eectiveness of working with that device.
At its most basic level speech recognition allows the user to perform parallel
tasks, (i.e. hands and eyes are busy elsewhere) while continuing to work with
the computer or appliance.
2.7.3 Communication
Examples
Telephony transmission of information in digital form via telephone lines,
modem technology, mobile phones.
Encoding and decoding of the information sent over a physical channel (to
optimize transmission or to detect or correct errors in transmission).
2.7.4 Radar and Sonar
Examples
Target detection position and velocity estimation.
Tracking.
2.7.5 Image Processing
Examples
Content based image retrieval browsing, searching and retrieving images
from database.
Compression -reducing the redundancy in the image data to optimize trans-
mission / storage.
Image enhancement.
2.7.6 Multimedia
Generation storage and transmission of sound, still images, motion pictures.
Examples.
Digital TV.
Video conferencing.
2.8 Limitations of DSP
2.8.1 Aliasing
Most signals are analogin nature, and have to be sampledloss of informationwe
only take samples of the signals at intervals and dont know what happens in be-
tween aliasingcannot distinguish between higher and lower frequenciesSampling
theorem: to avoid aliasing,sampling rate must be at least twice the maximum
frequency component (bandwidth) of the signals.
Sampling theorem says there is enough information to reconstruct the signal,
which does not meansampled signal looks likeoriginal one limitations of DSP -
AntialiasFiltercorrect reconstruction is not just connecting samples with straight
linesneeds antialiaslter (to lter out all high frequency components before sam-
pling) and the same for reconstruction it does remove information though.
2.8.2 Limitations of DSP Frequency Resolution
Most signals are analogin nature, and have to be sampledloss of informationwe
only take samples for a limited period of timelimited frequency resolutiondoes
not pick up relativelyslow changes.
Chapter 3
Conclusion
The main achievements of the training in Digital Signal Processing are that i got
familiar with the latest technologies and principles of Digital Signal Processing
Technology.
The main achievement could be said to get knowledge about recent technologies
of DSP and completed the syllabus of DSP.
28
Chapter 4
References
4.1 Bibliography
1. Digital Signal Processing
Simon Haykins
Sanjay Sharma, Analog Digital communication
Aian V. Oppenheim Ronald W. Schafer
2. MATLAB
MATLAB 6.5 manual
29