NCIT College

Digital Signal Analysis and Processing (3-1-2)

Evaluation Practical Total

Theory
Sessiona 30 20

Final
Total 20 100
Course Objectives:
1. To provide knowledge on digital signal procesi techniques.
2. To design and implement IR and FIR digital filter.

Course Contents:

1. Signals, Systems and Signal Processing (5 hrs)

1.1Basic elements of Digital Signal Processing
1.2 Energy signal, Powersignal
1.2 Need of Digital Sigpal Processing over Analog Signal Processing
14 Sampling continuous signals, and spectral properties of sampled signals.

2. Discrete-time Signals and System (9 hrs)

2.1Elementary discrete-time signals
2.2 Discrete time Fourier series and properties
2.3 Discretetime Fourier transform and properties
2.4 Discrete time system properties
2.5 Properties of Linear Time-Invariant systems (LTI)
2.6 LTI convolution sunm characterized by constant coefficient difference equations
2.7 Stability of LTI systems, Implementation of LTI system.
2.8 Frequency Response of LTI systems

3. Review of 2-Transfornm (4 hrs)

3.1Definition z-transfom
3.2 Convergence of Z-transforn, Region of convergence
3.3 Properties of Z-Transforn (linearity, time shift, multiplication by exponential sequence,
differentiation, time reversal, convolution, multiplication)
3.4 Inverse z-transform-by long division, by partial fraction expansion

4. Discrete Fourier Transform (8 hrs)

4.1Definition and application
4.2 Frequency response of
LTI system
4.3 Forward and Reverse transform
4.4Properties of the Discrete Fourier Transform: linearity and Symmetry, time shift,
Irequency shift, duality, convolution, multiplication, conjugation & conjugate symmetry
4.5 Basic concept of Fast Fourier Transform
(FFT) algorithim
 (6 hrs)
5. Discrete Filter Struceture,
5.1 FIR filter overview
sampling, lattice
.2 Structures for FIR Filter (direct, cascade, frequency
5.3 1IR Filter overview
lattice ladder)
, cascade, lattice,
4 Structure for lHR filter (direct form 1&effects on location of poles, and zero
.Quantization of filter coefficients and
(7 hrs)
6. FIR Filter Design
6.1 Gibbs phenomena in FIR filter design
hanning window, hamning
.2 Filter Design by Window method (rectangular window,
window)
6.3 Filter design by Kaiser window
6.4 Filter design by frequency sampling
6.5 Filter design using the Remez exchange algorithm
7. IIR Flter Design (6hrs)
7.1 Filter design using low pass approximations Butterworth filter.
7.2 Filter design using impulse invariance method
7.3 Filter design using bilinear transformation
7.2 Properties of Chebyshev & Elliptic filters
7.3High pass, Band pass and Notch filters.

Laboratory
1. Overview of DSP 1ools.
2. Scaling, dynamic range and noise behavior of a recursive digital filter
3. Response of a non-recursive digital filter, Implementation in Inpulse Invariant and Bilinear
Transformation.
4. Bandpass filters implemented using cascade second order systems
5. Design of FIR filter
6. Design of IIR filter

Reference Books:
A.V. Oppenheim, Disecrete-Time Signal Processing, Prentice Hall, 1990.
1. J.G. Proakis and D.G. Manolakis, Digital Signal Processing,.Prentice Hall of India.
3. S.K. Mitra, Digital Signal Processing, A Computer-based Approach, McGraw Hill.
 chapter1
Signals, Systems and Sgnal processing

LA Sianalid dekined a 2ny phssical: quarhity that Varies

widh ime Spateorany otherindeperdet variablekt.
specchSignal, imageSignal, e sinut.
A Systemiadelined 2A a phi Gicaldeuice ha pextosro
oheraion an a Siqnal. G
a H r used o xeduce dhe_noiseand inderferencei
The SysBem ucharadteizd by he type aperahonthat it
LeLtom% an thesonal
thr £ame 1£ Hhe opezation io lineät the Sygtem called
incai IE the ope yation 0on-ineaz the Systemin Saidto
be noninear and So oxthi Such operaBionsare uAually .

efexrtd to a SLanalL_pzaceAing
Siana analysis descrihea dhedeldof cudy whnsegpal
tColle.cdinderctand anddeduceinfornatinnaa
intellience tom VariouASignals
Digital Signa prncesing:
DiitalSiana prncPingis the prore 0f£ analyzin9 and
Lmndifying 2 Sionado 0pimize o impnve ikeffiienty
pernmanie I pimarily Aed detect erars, and
o e r and Cornpre/D_analb
.
signa in daanAit
Hinynluelves 2pplying vadaus mathernathicaand Compltahona
alqnrithms toAnalo4 nd diqita SinalA Aa pancuice -
Sianalshai c f hiabher quality than the 0iginal Signal
TtacilitaBe) he charin9 Of a Snglepmteoramang
n 0fsiqnals ky hme- shain4

NCIT College
 cmenis04 Diaidal Signa_pz0cOn
native
a 0 aBezn.
g A o p i a v i d e s

DiaidalStnoa220ce Siqnal
adluAtraBed

Method for oceinq he 2analog

in fi beloo.
D/a Analng
Analoq Alp
Diaiia
ACanvetter 0utpu
Convezter Stanal
Siana LP3ocO>07
Dialtal
Digital .J
Output SÍ gna
input sI gnaJ
Digita
of basiC 2lemets
fBlock dia9ramn
SyStemn
Signa) p3D(emin
n_nature
Of ihe_Signals generaBed are a0alog
SoundViden, ekTi Tf Such
temppreMureSeismicSiona

sianals are zntemed by Dsp System,-hen dhe Signalbs

munt ke digiti2ed. Henre input in qiven dhmughAlD
overier and OuipukJA 0tained thanuah DlA Conveter
Ihe baic eleroents ofDsp Spstem are
i)A/D_lhnverttr
il Digidal Stgna pa0tenor
i D/a_Converrr
iA/p lonverter
It Convefts analog input to diqiial input, uhith va
Lappingriade ab 2ninput o dhe diqital _procepos. The-
lAID Conveter_delemine Samplinq ale and quantirob D
eIDIin_digitizinq oPeratisn
 iiDigital Siqna) pXaCemo
The diaital Signa poceoa ay be a large pabg1ammabl
digita Computer0 aS.mal1 mic0prvLedor panararomed to
perform the deaired operahon on the nput sipnal.It
Lpey fox msaoplilicahon atteruakon, fiHeingSpectoal analysis
feature exaachon etc operaion on digidaldata.

ii)DLA ConvertIL
Same af dhe pmmoMed Signals ae aequired hacr in dhei
analngAorm -for example, SOund.imaqe, video aze dequired in
analnghm. Hence autput o£DiqitalSiona DCOMOr
iven_tnDIA COnverter 2hich_Converts dialtal_Ottput d
DSP iHs analog equivalent

dvantaaesADiitalaveyAnalog SiananeMing
Digital pznazammahle Sustermallnws lezibilityin 1econtigumtg
the digiial signal ncexiln9 operatiom Simply changing
he xoqram
2 Botter_Contol of_acuracs in _digida) SystemS Cornpared
o a02lo yshems.
2 Piqldal siqna arp easily Stoaed on maqneticmedia (tape
disk taithoutdeteninzaton or ioss af Signal As a
fonsequ ene he stanal becorne transfartable and an be
Pr0 Ceed Df-line.
h e digiial Siqna pocexsng methad alpws for the
implemerhadn_of_more sophisdi(aded Ginal piceMing
2laaithms.
12igida implemeniatonhf dhe Stqnad po.ceing Syslern
cheaper man 3nølpg.
 and
of demp: 9.ng
indepcndent
s u s i e m s a r e

gilai
Oiher exianalpa1ameic5
to
dolerances
Comot
Seosihve

Cicuits lem
iqa
Value. L 2 s C a d e d _ u i d h o u t
any_hadi
ding-
&DiglBzlSysiem Canbe

DblemS

Disadvaniages of DSP uith selatively_

dtoSigna
DSpHechniques_are
limide
bandoidth_
banduidtha it qequizes higher
10 uide bancduidths gequires
equires
extxemely
Sanas having
Canverttrs andfact digital signal
high speed Aln achieve in 15P._
EOCOMDIS, which are dith cultto
applicahong._
expenaive fr Sall
2The DSp ystems 2re
Amicahions 0f DSP

Speech pancesing
Speech 3 signal. najjal Ceing of Speech
1-9imension2a
is applied to uoide 3ange Of speechD0hleroc Such 2A
Cadirg-
Speechspeclau a0alysiS.Channe Vocnders.,speech
el: psp 2pplied tan speech enhantement,Speech analii-
nd Synthe1 Speech TOCogniHHon.
Imageproceingi-
Any 2-dirnensiona an im2ge, Agidal pncele-
aiemn_Called
of inaqe geauire -DimensionAl Dsp -Bools Such 2 DEL-
FFT ala0ridha and z-tansfoms.ProceNing of eledica
Sga exdnacded Hzoro image b ysigidal dechni que).
 ncld.e
eimajer2ai1on-andreCaxding, image Compaemiln
imaDe econstnucdion_imageeSHoxation andimage
enhanementi
aRadar Sianal pDLeing:
Ladar slandAfiYKadio_deiecion and Langing
mpnuemen in signal pmceing possIhle by_dígiialL
dechrolngy DevelopmenofDSP ban Ikd. to qzeader Sophic
itahonof 1ada ackinalgarithms
ADigital ammunicabion
Appicahan fpsp_in digital Communilationspecially
dete lommuicaion ompiCes Of digltal transmision
win PCM, digital Sunfkhing AinG DM. echo CnoBM
and digdal tape- xetordersi Dcpin Jele Communication_
SStems are dound. do be CoSt e-flective due to mal
3vailalilily of medium2nd laqe caled'gita ICG.
bspecdzalAnalysis
ireaueny domain analysiL iseasily and eflcively _possible
indigiBall Sqnal pmteing ung EET alqoidhd. The^e
algorithm 12dure Com2tational omplexityand alsa reduca
Coroput2tioa dime
b Sonar Sianal mleMing:
Sanaz_Standa forSound NavigaHon and
Kangna.Sonar
w_Wedto deleimnehe arge velocity 2nd direction_
Of aets tha are secoie_Top Hhe
0bseryor. DS P-
Can b ed toprmcoM Sanar
Slgnals or the puipoSe d
navigaHon_and_tanging-
 Consumera2pl caiOna_

Condznl appialinnA Tnduxtaia appli.ation

EndcstaimnentAppli.aipr
eit medital engineeing-
Miliiary lnefense_applications_ Telemmmunicatlon

Audo.maHve
AD and Da anvesinn Such_ahSpeecb
Mos Signals ofpracical inierest
binlagical signa ard
SeiL S i g n a l s .T a d a sSignab Yideo
CommunicationSignals
Such2A audin and
AD
1gnas are analag digital:meaa,ifn_Aid
ID_pro.COM analnq Signal by d i g i t a l o m i ë to
neceMany t Contthem intn
havingAnie
LonuertAhem Ho a Sequence of number
Analog to diqstalA/h
Tecision. This p0ce duresis_caled
onverSbn_and the Corzexpondinq
deicediare caled
AlDoneders. (AN
AD_onversion is 2 three Step ot.eM.This pDce
in Tatrd_in_ o. elota
A nvertev
Kt
Sampler &ua i zer Cod er

Analogi Diqital
DISCYee me uatized Signa
Sigha SLqna Stgna

9 CasiC_paris Aly Converc}e r

 Sampling
This iS the ConverSI on df a _ortinudw- Hime
Stonal int
a_discrcte- 1me Signalohiatned by takinq "sarmpln"
o the (oniLUDW) -Hme_Slanal at discretc time indan.
Ihud.if c)iS dhe inpLt do 1he Sampler the outut is
n1) E Xn) Whene Ti alled he Samplinq inierval

uanhizaion
hhis in the onvexsion0f a discrete- Hme ondinunu
yalued S1gna) into a discrete Hme discrete=valued ldigitd2
Stana The valuR q_each signal Samle isepresented
by 2 yalue Selectrd drom 2 finiieSet poSSihle
valea The didference bettyeen unauani2ed Sarnple an)
and he quanhzed atput 2 tn Called dthe quan-hzation
erni-
3.Codin9
1n he Codinq an (eM, e2ch disrete value K¡cn2
geNreAerted hy a k bit binary Sequence.
Dla lanyerSino
In_ many Cabes0f pMaciCal interetlég:Speech pmceNhna
indesirahle ta onvert the pmteMed Signah into
analbgform. The 70 CeM of ConyerHnga digii Sional
into an analo9 Sianal is Knoton digital to analog-
A) Conyersion
Al Dl onverts cannech dots'in a digital_srgna
bminq ome kind of in-texpolatiom, úhaSe accuaaty
Oependb on 1he Quallty he Dla onveSinn_DIvteA
 PPDNmalfon
GtaircaDe

Ongina

opnda
Signa

6T

Dla ConVelter-
hold
9 : ZeD - otder
Samplei
gelonAtuded hom he
Iheanaln sianal Can be nOI any dictorion
Lihoutany InsS of information high
aaASu£ioientiy
ided dhatAhe Sampling
Called Ali3 ing
2Ynid the_2oblem Commonly

her? the high fieguenty

Al2sino s 2 henomenon
toith fach
nmDD ek 0 he Sampled ignal inter fale
odher becaune f in2dequate Samplin
Herterene
highfreg. lompoDENES

Alasnq leadh do chistorcion in ietnVered_Sianal. This

the eAnon ohy Sampingre4 Should be at leaat
hoieCe he bendwidtb of the Signal
 SamplingOf rinuOUASigna
Samfli9 he pLLM 04iee Can verSion Ofthe Conjinuau
time Signal 7,d IniD_discrete-time Signal n) ie
-c0<n <o

heze
X() dhe discreie -dime Stqnal abdained by
ta Xing
Saople" at the_2nah Signa eyery Iseonda
his pmce dureS illuatrated in_ diq belota

t)(T)

Sapler
12 3 GS 63 9 9

fr penDdic Sarnpnq 0 an ahalp9 Signal

The ime interva T between Sutceive sample is called

he Sanplin g period or sampingindervaand ik noCLpmca
alled_Calle.d Lamplin4 ate,Samplea per Selond) or
Samplinq

lénodic aophinq establishes _a elatonship betoeen dhe

ie
YaNALLe t and n_ 0f Coodinun w -ime 2nd discaet e-+me
San al 2
t e nT :D

A2TLAulLthene exists2 Yelahonshi p betueen 1he tieq

Vatiabl E r hr aalng Siqnab and rea Vonable

11
 Cor t) _screte lime siona he
SinuSeidai siynal
analo9
ei u enides20

for
t ) = A Cos (2xEt + 01
Sampes_LSec
Ahen5armpled peindtcally3
E

EnT t09)
n
NO) A_Cos_(2x
nT)

Xn =ACos (2xn F 0)
Fs

Campann e anoith_ to) A _ o s ( 2 x t n , t 9 ) oe 9

Also, oe Eno, that maximum Tange discrej time

4requencíe iS

y Substingthe_value 4f:f and w e

2T 2 27

F s As
 eobseve dthat he undamental di£lente
ICotinuov time 2nd discacle lime Siqn between
ange ah in their
Dlue Ofthefreq. Vadables Fand
A and
Thus, eli0dicsampling
o a
imlieda mappinq 0f dhe Corhnubu- time Stqna
h e Naiable infiniledrq, ange
E(or n) indo affniie
Yangeforhe Variahle f(or a)
Since the highe Kreq. in_a
_dscrede :Hme 4iana is
o 0rf: iiHAblloosdhat, wiHh a
22Be Hhe S2mphng
ae Lamepondinqhigheat_Yaluen E and L
ahd-

Samplin henrero
AaBed hat a CortinuoA timeSian2
can he
ompletely exeentcd in Samplea 20d e CoxeredL
bacie f the-
Samplingdi quenty 21damay-
Nyquist ateE2fmax
Nyqws+ nterval 2x

Kesalution E ma *min
L 1
Anin -Amox
Leno, levels.

13
 V
V
V
 RAdiqiial Communica.H0 lhox caznie) bnary coded unass
neprelentngSample 12n Inpub Sina)
t) 3os 600 r t t 2 cos 00 Ath e link
10,000 bits /Sec andeach ioput Sasnple oeraBed
intn 102 differcnt_ValBaje lexels is quanh2es

ihad he sâmplingre and hldin

what i the Nqulst Tate for eq.2
the ignal ,tt2-
ii) aha ar21the frequentienin
esulhng diCcrete Hma_
Signal tn)2
hat he e coutian A 2
50 Here, giveA

dzts 3:CoS 6no at_+2 (DS 200 t

bit TaBe =
10,000 bits/sec
no. 9uanized leve (U402
e b e the_nn.
we En0t0_
hik/Sarmple
b=109,L) lo9,(1024) 10 bis /Sample.
Fn0 , it
ale E Camples /Sec_ bits /sarople
) Saoplesec i gae 10,000 1000 Hz
bits/sample
o ldn dreq 000 K0D Hz
2

15
 and
samplina icqG) I000 Hz
Saldinqdseq = 500 Hz

hmm e" 2
O
I,C 0S 2x (300)t +2 Cos 2x (400)t

200 H2 &A3
yqust. iNeTva)
&Aa2

00
Nyqust dale in given by
NyQuiSE 12Be E 2 imo
2 x 00Mz:
200 H2.

sii
eplhcingt by ie t : n in egl1? we ct

2t0S 2x l200)n 2CoS 2 (400)

3 oS 2a (300) n 2 (os
2A (u0o) D
000 00D

3 (oS 23 0.2) n t 2 CoS 2a o.9)n

' 0.3 & . 4 Hz

i eoiuion,_A Xmax i n AtAa --A +Aa))- 5--5)

L-1 024 -1 02-)

'A= 0.00 945

o19fa
b8Lansidehe the anan Sgnal.1,4)2_cos 10.0.
Dnelerpine thtmin Sm)in 9 ale requited to
avoid aliain
iiSuppa.se that he 5ignal is Sample d at he ate
what o he diSCrelcime Siqnal F200 Hz,
0ltained aflcr 5amping?
iismse that dhe 5ig Sampled_at the_ gate F 6 Ha
Lhal i the disereie
inwhato dhe reg
AmeSignalOitalnedafler 5amplin
EKKaf a Sinusoid that yields
Samplea idendical do thoseahtalnedin _part (i)._

S 2Here Xti) 3 osl00xt_

' F 5o Mz.
The fre4 of the analb q Signal F:50 Hz
Accordin g to nyquist rate
ienyquist T2te1EV2 2ma 2 x 50 Hz. 100Mz

Hence, the min. :Sampling Tate Tequired ta aois aiain9

s 00 Hz.

52pled a , f 200 Hz, the drscrele- hme

Signalin (ie, tu t in eg )
Cn) 3 CoS 2 50 200

.'. den) 3 CoS ()n

ii)T S1gn Sample d 3t fs 5 Hz, then

tn Cos 2 50 3 Co S n
5
3 Cos(2 2 )n 3 Co S n

17
 nihe Sampina. 2ale s a 5 Hz hve

F fF 5 f
he req. heSinu S0Jd.in pat ) iS

Hence F 5x a 5, N2
cleany, he Sinusold SiqnaU,
2C) 3 CoS 2r Ft
Cos 5nAt

amped at 3 5 samplen /secields idenh Pa Samgle)

Hence F:50kz
2n alia o fE 25 hz for dhe
Samplin.q Tatef A5
1 3 - >

Ator 2 digital System: With a max audio inpu equy

42) o 6kHz determinee_the min. Sample 2athe 2nd he alia
Arequenty pmduced i£ a 2 K Hz_audio Sgnal uere allpwed
Ao ener A0alng tb digitaConverter:
So Her, g* n Let sabe there4
3udio:.
mar Hz
Sample aate 2 2 frae max 2* 6 k Hz 1 2 k k

Samplina freq. fs) =

+KH2
 (onaldertheSianalIzH)= o Cos 2r ino D)t +5 Cos 2%(5000)1_
iatn be S2mpled
neermne he nyqui Mte tor thir Stanal_
dhe siqma sampledat: M2 ,*hi he ignal_
he elaened om Sanple 2-
So ere, Gt)=1d Cos 2x(Lo D0 ) t 5 (os
25500 0 E
Clearly the signa g4) Conaim fwo CoSi ne woave)

An5
ine, th high et fre. in. 12tt) fmo h : Boo0 Hz
According donyqult (iteia
Sorn 2 tmay 2me 2*5000 1000DE
iSine, dhe Sqnal JA Sanled 21 f 41<Mz =yo00
hen YzLA Saropleo, u»e_ahtain xc by t7 }

13
 CID00) D t5 Cos 2x500D) 000

=o Cos2141n 5cos 2A 10-

Io o 2 )n+ 5 os 27)n_

.n) 1S Cos 2 (t)D

tohich sa pled Siqna,

when thS Siqna reconstruced, dhen

Dt hooot
1(0S 23(1000)

Here, ue ohServethat the retanstu ded Siana Gnia

niy ane freq lpoo HzAnd amplitde S S.
1hu the effe ct fa= 5obn Hz Compledely
lost2nd g.malidu.de E j incTe 20ed. This Shoos
h a t ioitn dneSamling- 2ie 4000 H2,the S19nal

eCovered nm ts S2sple This n becawe

nô
n e amplin heoTtmin np Saisied
The ny qli rate 2 Calculaied: in' i s o kHz
minSasplinq re4 Shoull he Ib XHz Ao avoid
2li2Sing

dislrete dime Sna Z0)=66 Cos (o XD

The
Qualtized Laiih
t s 2re yequi red_in he A/D_lonyerte 1f de
 max he thal Can beeanstaucted ramabove
Sigha nSoHz,dedemine the ehnituded
Lontinunw ime Signal
So ere, qven, cn) 6b Cos Co.1 an)
-65 (os 27 ( )n
20

fre E ) : Hz & A 6.5

Te50ltio A 0.0. max. freq. ( mox- 500 H

'. Sampli n req. (4s)= 2

fmax 2x 500 Hz. 1000 Hz

we en 0
A max min A,- (-A)
L-
L-I|

0T 0-0l 6.5+65
L 1301
L-1

now n0. bit kample logL) = 109,(1301) = 10.34S

Let TeCon7UCHed Con-t}nuo ime SIgra

So
2Ct) 6.5 Cos 27(i000XD.05) t
-

,H) E 65 los 27 (50)t_

A diqital_ DmmunICati o n link binary-Codèd 07es

Teproentinq Sacmple Of_an input Siqal
At) 2 Cos 5000 t +2 51n 6000 &_

21
 The link i olerated at18,000 LH/sec3nd each inputs
guanii2ed into12 dlfeni votage_Jevoh.
Lahat Jhe _discrete -di me Siqna obBained. fH ev sa mprg2
i l whad_i eSaludion, A2
SDI Le Knouw,
b L =
l09 lo9, ( 512) =
S bis/sarpe
we kmoo,

F (sampleslsec) bid,rale 18000 - 2000 52nle

itssample 9

Samping txeq LEYE 2A00 Sagle) [sec

Her
2 (+)= 3 Cos poon+ + 29) n 6000 t
21 =3 oS 2x(2Soo) t t 2sin 27 (3000)
Lohich han. A£3
Fa2000 Hz L A2za
AAfr Sampling i e t oY

ZD) E 3 tnS 27_2c00nt2 s)n a 2x(2000)n

2000
200
3 Cos 27(1.2s)n+2 Sin 27 (15)7

Iphiths the equired discrele -tine ignalDhtzined

aHer2oping

Tecoluion (A) = "xpin XrayAytA25

Kain Knoy
5-(5)
512-)
.D 0.0195