DIGITAL SIGNAL PROCESSING 701Digital...COURSE NAME DIGITAL SIGNAL PROCESSING COURSE CODE: EC 701 Dr....
Transcript of DIGITAL SIGNAL PROCESSING 701Digital...COURSE NAME DIGITAL SIGNAL PROCESSING COURSE CODE: EC 701 Dr....
COURSE NAME
DIGITAL SIGNAL PROCESSING
COURSE CODE: EC 701
Dr. Mrutyunjay Rout
Dept. of Electronics and communication Engineering
NIT Jamshedpur
1NIT Jamshedpur
NIT Jamshedpur
2
Course DescriptionUNIT-I: DSP Preliminaries, Sampling, DT signals, sampling theorem in time domain, sampling of analog
signals, recovery of analog signals, and analytical treatment with examples, mapping between analog
frequencies to digital frequency, representation of signals as vectors, concept of Basis function and
orthogonality. Basic elements of DSP and its requirements, advantages of Digital over Analog signal
processing.
UNIT-II: Discrete Fourier Transform, DTFT, Definition, Frequency domain sampling , DFT, Properties of
DFT, circular convolution, linear convolution, Computation of linear convolution using circular
convolution, FFT, decimation in time and decimation in frequency using Radix-2 FFT algorithm, Linear
filtering using overlap add and overlap save method, Introduction to Discrete Cosine Transform
UNIT-III: Z transform, Need for transform, relation between Laplace transform and Z transform, between
Fourier transform and Z transform, Properties of ROC and properties of Z transform, Relation between
pole locations and time domain behaviour, causality and stability considerations for LTI systems, Inverse
Z transform, Power series method, partial fraction expansion method, Solution of difference equations.
UNIT-IV: IIR Filter Design, Concept of analog filter design (required for digital filter design), Design of
IIR filters from analog filters, IIR filter design by approximation of derivatives filter design by impulse
invariance method, Bilinear transformation method, warping effect. Characteristics of Butterworth filters,
Chebyshev filters and elliptic filters, Butterworth filter design, IIR filter realization using direct form,
cascade form and parallel form, Finite word length effect in IIR filter design.
UNIT-V: FIR Filter Design, Ideal filter requirements, Gibbs phenomenon, windowing techniques,
characteristics and comparison of different window functions, Design of linear phase FIR filter using
windows and frequency sampling method. FIR filters realization using direct form, cascade form and
lattice form, Finite word length effect in FIR filter design, Multirate DSP, Introduction to DSP Processor
Concept of Multirate DSP, Sampling rate conversion by a non-integer factor, Design of two stage sampling
rate converter, General Architecture of DSP, Introduction to Code composer studio, Application of DSP to
Voice Processing, Music Processing, Image processing and Radar processing
NIT Jamshedpur 3
Books
Text Books:
1. John G Proakis and Manolakis, βDigital Signal Processing Principles,
Algorithms andApplicationsβ, Pearson, Fourth Edition, 2007.
2. S.Salivahanan, A. Vallavaraj, and C. Gnanapriya, βDigital Signal
Processingβ, TMH/McGraw Hill International, 2007.
Reference Books:
1. S.K. Mitra, βDigital Signal Processing, A Computer-Based Approachβ,
Tata Mc Graw Hill, 1998.
2. Ifaeachor E.C, Jervis B. W., βDigital Signal processing: Practical
approachβ, Pearson publication, Second edition, 2002.
3. Johny R. Johnson, Introduction to Digital Signal Processing, PHI, 2006.
Lecture: 1-8
Introduction to Digital Signal
Processing
4NIT Jamshedpur
Lecture: 1-8β’ Signal processing emerged soon after World War I in the
form of electrical filtering.
β’ With the invention of the digital computer and the rapidprogress in VLSI technology during the 1960s, a new wayof processing signals the signal processing is term asdigital signal processing.
β’ Digital signal processors take the real world signals likeaudio, video, speech etc., that have been sampled andquantized and then mathematically manipulate them.
β’ Signals need to be processed so that the information thatthey contain can be displayed, analyzed, or converted toanother type of signal that may be of use.
5NIT Jamshedpur
Lecture: 1-8
6NIT Jamshedpur
What is Signal?
β’ Anything that carries information and represents
as a function of independent variables such as
time, space, temperature, pressure, etc.
β’ Any physical quantity that can be varied in such a
way as to convey information.
β’ A signal is any quantity that depends on one or
more independent variables.
Lecture: 1-8
7NIT Jamshedpur
Example of Signalβ’ A radio signal represents the strength of an electromagnetic wave
that depends on one independent variable, namely time is a 1-Dsignal.
β’ Image is a 2-D signal.
β’ A video signal is a 3-D signal.
β’ Natural signals:β Signals produced by the brain and heart
β Signals originating in galaxies, astronomical images etc.
β Speech signals, sounds made by dolphins
β Signals produced by lightning, the atmospheric pressure etc.
β’ Man-made signals:β Signals originating from satellites, radio, telephone, TV
β Signals due to ECG, EEG etc.
β signals generate from musical instruments
Lecture: 1-8
8NIT Jamshedpur
β’ Signal Processing: Process of operation in which the characteristics of asignal such as amplitude, shape, phase, frequency, etc. undergoes a change.
OR
Signal processing is the analysis, interpretation and manipulation of anysignals like sound, images etc.
β’ Types of signal processing:
βAnalog Signal Processing
β Digital Signal Processing
Analog Signal
Processing
Analog SignalX(t)
Analog Output Signal
y(t)
Sample and
Hold
AnalogInput Signal
X(t)
Analog Output Signal
y(t)
A/D
Converter
Digital Signal
Processor
D/A
Converter
β’ Digital Signal processors (DSP) take real-world signals like audio, video,
pressure, temperature etc. that have been digitized and then mathematically
manipulate them
Lecture: 1-8
9NIT Jamshedpur
Components of a DSP System
Lecture: 1-8
10NIT Jamshedpur
β’ Advantages of Digital Signal Processing:βGreater Accuracy
βCheaper
βEase of Data storage
βEasy Operation
βFlexibility
βMultiplexing
β’ Limitations of Digital Signal Processing:βAntialiasing Filter
βBandwidth limited by Sampling Rate
βQuantization Error
Lecture: 1-8
11NIT Jamshedpur
β’ Applications of Digital SignalProcessing:βIn Communication
βConsumer Application (e.g., TV, FMradio etc.)
βImage processing
βIn Biomedical
βIn Radar and Sonar
βIn Speech and Music
Lecture: 1-8
12NIT Jamshedpur
β’ Any unwanted signal interfering with the main signal
is termed as noise. So, noise is also a signal but
unwanted.
β’ Classification of Signals:
Depending on the independent variables and the value of the
function defining the signal.
1. Continuous-Time (CT) and Discrete-Time(DT) Signals
2. Continuous-valued and Discrete-valued Signals
3. Multichannel and Multidimensional Signals
4. Deterministic and Random Signals
Lecture: 1-8
13NIT Jamshedpur
Continuous-Time (CT) and Discrete-Time (DT) Signals:
β’ Continuous-Time (CT) Signal:
β’ A CT Signal is a signal that is defined at each and every instant of time.
It can be represented as x(t), where t is the independent variable.
β’ This type of signal shows continuity both in amplitude and time. These
will have values at each instant of time. Sine and cosine functions are
the best example of Continuous time signal.
Lecture: 1-8
14NIT Jamshedpur
Continuous-Time (CT) and Discrete-Time (DT) Signals:
β’ Discrete-Time (DT) Signals:
β’ A DT signal is a signal that is defined at discrete instant of time. It can
be represented as x(nT), where n is an integer and T is the time interval
between two consecutive signal values (Sampling period).
β’ This type of signal shows continuity in amplitude but discrete in time.
β’ Relationship between time variables t and n of CT and DT signals.
Lecture: 1-8
15NIT Jamshedpur
Representation of Discrete-Time (DT) Signals:
β’ Graphical Representation
β’ Functional Representation
β’ Tabular Representation n β¦ -3 -2 -1 0 1 2 3 β¦
X[n] β¦ 0 0 0 1 1 1 1 β¦
β’ Sequence Representation . . . 0 0 0 ΰΈ1β
1 1 1 . . .
Lecture: 1-8
16NIT Jamshedpur
β’ Continuous Valued and Discrete Valued Signals:
β’ Values of CT or DT signals can be continuous or discrete.
β’ If the signal takes on all possible values on a finite or an
infinite range, it is said to be a Continuous valued signal.
β’ If the signal takes a set of discrete values, it is called Discrete
valued signal.
β’ Continuous time and continuous valued : Analog signal.
β’ Continuous time and discrete valued: Quantized signal.
β’ Discrete time and continuous valued: Sampled signal.
β’ Discrete time and discrete values: Digital signal.
Lecture: 1-8
17NIT Jamshedpur
Multichannel and Multidimensional Signals:
β’ Multichannel Signal:
β’ Signal is generated from multiple sources.
β’ For example: Electrocardiography (ECG) 3 lead and 12 lead signal.
π₯ π‘ =
π₯1(π‘)π₯2(π‘)π₯3(π‘)
β’ Multidimensional Signal:
β’ If the signal is function of one independent variable is called one
dimension signal otherwise the signal is called M-dimensional signal
β’ For example: Video signal, I(x,y,t) is a 3-Dimensional signal because I is
the function of three independent variables (x,y,t).
Lecture: 1-8
18NIT Jamshedpur
Deterministic and Random Signals:
β’ Deterministic Signal:
β’ A signal is said to be deterministic if there is no uncertainty with respect
to its value at any instant of time. Or, signals which can be defined
exactly by a mathematical formula are known as deterministic signals.
β’ This signal is predicted at any time.
β’ Random Signal:
β’ A signal is said to be Random if there is
uncertainty with respect to its value at some
instant of time
β’ Random signals cannot be described by a
mathematical equation.
β’ Random signals are modelled in probabilistic
terms.
Lecture: 1-8
19NIT Jamshedpur
Standard Discrete-Time Signals:
β’ Unit Step Sequence:
β’ Unit Sample Sequence (Impulse Sequence):
β’ The unit step sequence can be written in terms of delayed impulses as
π’ π = πΏ π + πΏ π β 1 + πΏ π β 2 +β― = Οπ=0β πΏ[π β π]
Lecture: 1-8
20NIT Jamshedpur
Standard Discrete-Time Signals:
β’ Unit Ramp Sequence:
β’ Exponential Sequence:
β’ Exponential sequence are important in representing and analyzing liner time
invariant systems.
β’ An exponential signal can either have exponentially rising or falling
amplitude depending upon its exponent value.
β’ The general form of an exponential sequence is given by π₯ π = πΌπ β π .
β’ If Ξ± is real then the sequence is real.
Lecture: 1-8
21NIT Jamshedpur
β’ Exponential Sequence:
Lecture: 1-8
22NIT Jamshedpur
β’ Exponential Sequence:
β’ When βΞ±β is complex, a more general case to consider is the
complex exponential sequence:
π₯ π = π΄πΌπ where πΌ = πΌ ππππ and A = π΄ ππΟ
π₯ π = π΄πΌπ = π΄ ππΟ |πΌ|ππππππ
= π΄ |πΌ|πππ(πππ+Ο)
= π΄ |πΌ|πcos (πππ + Ο) + π π΄ |πΌ|πsin (πππ + Ο)=Re π₯(π) + ππΌπ π₯(π)
Polar form
β’ If πΌ < 1, the real and imaginary part of the sequence
magnitude oscillate with exponentially decaying envelopes.
β’ If πΌ > 1, the real and imaginary part of the sequence
magnitude oscillate with exponentially growing envelopes.
Lecture: 1-8
23NIT Jamshedpur
β’ Exponential Sequence:
Lecture: 1-8
24NIT Jamshedpur
β’ Exponential Sequence:
β’ When πΌ =1, x(n) is referred to as the discrete-time complex
sinusoidal sequence and has the form:
π₯(π) = π΄ cos (πππ + Ο) + π π΄ sin (πππ + Ο)
β’ For complex sinusoidal sequence, the real and imaginary part of
the sequence magnitude oscillate with constant envelopes.
Lecture: 1-8
25NIT Jamshedpur
Operation on Discrete-Time Signals:
β’ Signal processing is a group of basic operations applied to an
input signal resulting in another signal as the output.
β’ The basic set of operations are:
β’ Time Shifting
β’ Time Scaling
β’ Time Reversal
β’ Signal Multiplier
β’ Signal Addition
OperationDT input
SignalX(n)
DT Output Signaly(n)
Lecture: 1-8
26NIT Jamshedpur
Operation on Discrete-Time Signals:β’ Time Shifting: The name suggests, the shifting of a signal in time.
This is done by adding or subtracting an integer quantity of the shift
to the time variable in the function.
β’ Subtracting a fixed positive quantity from the time variable will shift
the signal to the right (delay) by the subtracted quantity.
β’ Adding a fixed positive amount to the time variable will shift the
signal to the left (advance) by the added quantity.
Delay kDT input
SignalX(n)
DT Output Signal
y(n)=x(n-k)
Advance kDT input
SignalX(n)
DT Output Signal
y(n)=x(n+k)
Lecture: 1-8
27NIT Jamshedpur
Operation on Discrete-Time Signals:β’ Time Shifting (Delay): Right Shifting
Delay kDT input
SignalX(n)
DT Output Signal
y(n)=x(n-k)
π₯[n]= 0 0.25 0.75 ΰΈ1β
0.75 0.25 0 π₯[n-3]= ΰΈ0β
0.25 0.75 1 0.75 0.25 0
Lecture: 1-8
28NIT Jamshedpur
Operation on Discrete-Time Signals:β’ Time Shifting (Advance): Left Shifting
Advance kDT input
SignalX(n)
DT Output Signal
y(n)=x(n+k)
π₯[n]= ΰΈ1β
2 3 4 π₯[n+1]= 1 ΰΈ2β
3 4
Lecture: 1-8
29NIT Jamshedpur
Operation on Discrete-Time Signals:β’ Time Scaling: Time scaling compresses or dilates a signal by
multiplying the time variable by some quantity.
β’ If the quantity is greater than one, the signal becomes narrower and
the operation is called decimation.
β’ If the quantity is less than one, the signal becomes wider and the
operation is called expansion or interpolation, depending on how the
gaps between values are filled.
Time Scaling
k=2
DT input SignalX(n)
DT Output Signal
y(n)=x(2n)
Time Scaling
k=1/2
DT input SignalX(n)
DT Output Signal
y(n)=x(n/2)
Compress the signal x(n)
Expand the signal x(n)
Lecture: 1-8
30NIT Jamshedpur
Operation on Discrete-Time Signals:β’ Time Scaling (Compress): Signal becomes narrower
Time Scaling
k=2
DT input SignalX(n)
DT Output Signal
y(n)=x(2n)
π₯[n]= 0 0.25 0.75 ΰΈ1β
0.75 0.25 0 π₯[2n]= 0 0 0.25 ΰΈ1β
0.25 0 0
Lecture: 1-8
31NIT Jamshedpur
Operation on Discrete-Time Signals:β’ Time Scaling (Expand): Signal becomes narrower
π₯[n]= 0 0.25 0.75 ΰΈ1β
0.75 0.25 0 π₯[2n]= 0 0 0.25 ΰΈ1β
0.25 0 0
Time Scaling
k=1/2
DT input SignalX(n)
DT Output Signal
y(n)=x(n/2)
Lecture: 1-8
32NIT Jamshedpur
Operation on Discrete-Time Signals:β’ Time Reversal: This operation is the reversal of the time axis, or
flipping the signal over the y-axis.
β’ Folding the sequence x[n] about n=0.
β’ Mathematically, it is expressed as x[-n]
Time ReversalDT input
SignalX(n)
DT Output Signal
y(n)=x(-n)
π₯[n]= β3 β 2 β 1 ΰΈ0β
1 2 3 π₯[2n]= 3 2 1 ΰΈ0β
β 1 β 2 β 3
Lecture: 1-8
33NIT Jamshedpur
Operation on Discrete-Time Signals:
β’ Time βScaling and Time- Shifting operations are not
commutative.
β’ Time-Reversal and Time-Shifting operations are not
commutative.
β’ Time βScaling and Time-Reversal operations are commutative.
β’ All above operations are based on transformations of the
independent variable i.e., discrete time n.
Lecture: 1-8
34NIT Jamshedpur
Sequence of Operations:
Step 1: First delay or advance the signal i.e., first operation is the
Time-Shifting.
Step 2: Perform Time-Scaling and/or Time-Reversal on the shifted
signal
ππ’ππ π‘πππ: ππππ‘πβ π₯ βπ + 2 πππ π₯ βπ β 2 π€βππ π₯ π = ΰΈ2β
1 3 5 8
πβπππ‘ ππ¦ π ππππ π₯(π) = πβπππ‘ ππ¦ π π₯(βπ) = π₯(βπ + π)
πΉπππ πβπππ‘ ππ¦ π π₯(π) = πΉπππ π₯(π β π) = π₯(βπ β π)
Lecture: 1-8
35NIT Jamshedpur
Operation on Discrete-Time Signals:
β’ Scalar Multiplication (Amplitude Scaling): The signal x(n) is
multiplied by a scalar factor βaβ.
DT input SignalX(n)
DT Output Signal
y(n)=a. x(n)
a
If π₯ π = ΰΈ2β
1 3 5 8 and a = 2 then π¦ π = ΰΈ4β
2 6 10 16
β’ For Example:
Lecture: 1-8
36NIT Jamshedpur
Operation on Discrete-Time Signals:
β’ Signal Multiplier: Multiplication of two signals to form
another sequence.
DT input SignalX(n)
DT Output Signal
y(n)=a. x(n)
a
If π₯ π = ΰΈ2β
1 3 5 8 and a = 2 then π¦ π = ΰΈ4β
2 6 10 16
β’ For Example:
Lecture: 1-8
37NIT Jamshedpur
Sampling Process
β’ To be able to process a continuous valued continuous-time i.e
analog signal by a digital processor, we must first sample it to
generate a discrete-time signal the quantize it to get a quantized
discrete-time signal.
β’ A sampling system comprises three main components:
β Sampler
β Quantizer
β Encoder
Lecture: 1-8
38NIT Jamshedpur
Sampling Processβ’ Sampling is defined as, βThe process of measuring the instantaneous values
of continuous-time signal in a discrete form.β
β’ Sample is a piece of data taken from the whole data which is continuous in
the time domain.
β’ When a source generates an analog signal and if that has to be digitized,
having 1s and 0s i.e., High or Low, the signal has to be discretized in time.
This discretization of analog signal is called as Sampling.
Lecture: 1-8
39NIT Jamshedpur
Sampling Processβ’ To discretize the signals, the gap between the samples should be fixed. That gap
can be termed as a sampling period Ts.
β’ ππππππππ πΉππππ’ππππ¦ =1
ππ = ππ
β’ The sampling rate fs denotes the number of samples taken per second, or for a
finite set of values.
β’ For an analog signal to be reconstructed from the digitized signal, the sampling
rate should be highly considered. The rate of sampling should be such that the
data in the message signal should neither be lost nor it should get over-lapped.
Hence, a rate was fixed for this, called as Nyquist rate.
β’ The sampling theorem, which is also called as Nyquist theorem, delivers the
theory of sufficient sample rate in terms of bandwidth for the class of functions
that are bandlimited.
β’ ππ = 2π΅
β’ The sampling theorem states that, βa signal can be exactly reproduced if it is
sampled at the rate fs which is greater than twice the maximum frequency B.β