0 ratings0% found this document useful (0 votes) 43 views20 pagesModule1-DSP Algorithms and Architecture
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content,
claim it here.
Available Formats
Download as PDF or read online on Scribd
Bock Diaupran ef DSP _Systert t
lpnak ee ae
Drab JA ing | a din,
Glenak | Sah Abed Nar Retensinachio
ae an Provesso’ Teen
Digital Digital.
el DSP procenovs
end
we
The block diagrans of Oe? con ith
ehuee ry Analog front ond ¢ analy ;
ot. front end onal concisk a Anti abic ing
Aller, SIH 4 ALD toner lor . lp Ds?
=the olp of fnalog Rant Ond i giver
ProceHoy « :
= ae o is eit end cathe Up tolloued
m4 pare oe se fae Limit ef
= anna i rang
ae an 4 requenty NS
€ prevent
> the Sample
She ine , 1 an
a et tonverl ig fey to O80. gy
Analog equivalert oy ‘lp ot MO or are
= Atle procening OY Sali vONG by
bo they equ vi . ss Soup
converter is also ony on 4
Gon bibbey aoe \
= Reconghwe
WqCOVA VORA AA PRATER SRW EET
Tssue to be concidered in Rani
mae a Dee ae ae si
Oe Copal
The arithmetic operation to bbe, performed ¢ pe
Preciclon required ane decided by Je application,
{Semple vale >
The vate ab whith inpul Ceumplley are received ¢
Protecced vari with Je application < Avs rake.
along sity xo algorithm complexity determing
whe a partiedoy D6 % subkrble tor &
Given application.
‘Sy Speed |
Thc depend on the technolugy. Te meek ggecitied
Shroudn pul sequavement with a giver Semple role,
tb oath be potible. Yo operate Se OP a
purticuly «tloch vole. TA speed not achievable
"N given techpoleyy 1 ofottey rechnology ov Ohex
op on rut be plored .
uy Datu xepre entabion i> :
the format dearest ot pot Be a
vepruentation depend on anifonelic precition,
Cleanse wet me
Maroy Featunn ot Liogtumecble. Digital Ggnal
=
Processors
} roulfigly - accumual Wravduoant ‘
Sth nok fequendy uted operahion Ds
SThe DSP hay Hlw Py jatpleN. | AN) kun ,
With adequate numbev 4 bik to hold Se
Sum ot produch + enplicit mulfiphy - acculak,
inghuction.
eeeee
dh Hoxvord ardvfechure -
a There are hoo My Spaces
Ve oraam Ml NY dake mY:
Y Progra y \ Dou bi i ieaciat
= Procector core tonnents bo wy
bush res two Simulbaneoud atch dy me
Portibte
> TRy corangement docblles Ja procesiur
St Zev overhead Lopptog ‘ic
= Zero overhead Looping eA That be procesior GN
erecuke, Soop, coihrouk lontumin cydag reth
Value of he boop lounter, perform & condifio
branch fo be oop ¢ decrement Ane Joop tounter .
7 DSP procetioys include spectalized hardwore fur
Zero ovevhead Jeoping .
< Spedlned odd, Y=,
> DSP pocescors. atden Sunport spedalized addreni
wnodes tha are ve yy Common SAG
procer(ing operation f aleprithrry -
Ex cnodulo Carrudan oddcening, lik reverted
addrening y +
A Digital Signal ~ Processing Sytem Ae
= eee OD
poss JAoicrg|_SBip | Sse SUR)”
Vigited Dig
Figy A digital signal processing ‘syghem
or a aigitel precesto
4 hrurebre
wunals before,
4 raly BW.
DSO syrhern wlel o computer
a ciqnalll.
Certo sl aaThey can he procet(ed with a tommster
7 The contialiating Hiller crcureg Shak he Stgnal
Yo be Sampled” doeinot contain any Frequency
qr tran half of Sampling Frequency.
= +15
7 Hmalog be digiky — Converfey Convery fe omalog
Stgnal f> digtkeut.
~The diailal Signale are given ko DCP
© Digikd “to Pmalog converter lonverly digital
Signal to analog .
The reondncehion filler remove Nigh frequency
nove due fp he skevwcase up ot Ae
DID Comventer .
(onfnuowd Fine
Signal
npled
mecca
Carmplled dakar
4 oO a : s\qnal
|
aa fized [digits
es
(oat pasa a) Ogle! fo
ee ae ale converts
. OW signal,
One yrs 0.2 1 O18 mlOgyy PSEA MOUS oS,
fay “Typical cignalt 1 SP scheme ‘
2 An analog aga Wa tonfinuow Fine , tonfinuouy |
awn Mbud2 ChanaBaixo
= the Samp proce qenerdbes a Sampled St
The Sarmpled value Wy held b ct hold e
= The. S\qnal ak Alp twnveley Elp os Called
Sampled data * signal.
7 The procecied digital signal , ot dbbeined fom
me DEP WS he Inpub Yo he DUP tomworfey,
Fhe anatoy ofp of a dIR tonvertey hal shovecatel
amplitude due fo dhe stonvedion prove vied
\ Such a -device
the procen - tonvectng an anatog signal ra digit
Signal involves Senping de sign 1 voldiog \b tr
Conversion € tonverfing th to Corret ponding digilel voll
~> the Sarnpling Frequency muh be Nigh Enough Soa,
a to avoid abicti
(>> Bltoging = Th occurs due to sampling rake being
boo ow wks re1pect bv Nyquidk rake.
2 This @ phenomenon due fo which ahgh Preq.
signal when campled Utlrg a dou sampdbing
vale become, a ow Signal dat ray
interfere with 4h Stgnal ot interes.
TH Sn skote frab Se followt
ea ‘
Peis Ve oz. 2 frax.
r oH
k > Sears phing freq, Fs i
Ts Sarp hi olerval : |
poe ighalh treg, corer ct omaleg signal. Ista,frowe = Loko
=> Obmax = DOKHD
ato satithy te’ requirement , ay antialiating tiller iy
ted,
oss Alor Feller Lindi he Prequenuy content of te
ty to ia die Sampling theorem
= ‘the laduol sampling ® be Now
3 p TEQUeN UL e rey
ou an fee nall ety " avoid Hq
ee for be iniglementation of Ae antialiotrg
‘1
» 7
Discrete Time Sequences. =
wae |reculk al Sampling an anatog signal 16
Sequence, Wepredenting She signal Samples.
xeyy= A me owee-
T=¥ Sampling interval:
xeat) =A toe DEAT where 92 0,1, 2---ehe.
For crmplicily renin ae
yuny= A ae ‘J
Sime sompling Pa ane
Substituting Ops ey
ony = cos am tnt
yen) = A cos 2 oft
Bina fh tS ON.
; | bunik- rations).
b> & coed ei au the aiglal
equedio
Yo analogyr, . SIS,
= O properby campled signal dhe dighhaul i 4
Me muih be foun Umar 4
6 muh be o be Ww. ‘
XN) Sequence called She Sinusoidal Sequence.
The complex exponential Sequence *IS given by
P(n) = es2TON , Ne Benn Oy LD eke TI
where AL =} \oteger. ;
A Sequence tho repeodty w alld a periodic Sequence,
Period Sequences YUult from sampling periodic Signa
and Sec hy the Pollouiag rdakion
UM) = LAN) , Ne --- —b6, e-+
N= Sequente period.
TL easy to Show drat See Se aetee et coal
Xin) above hay o period tote a ee |
Sequence pln) hod ‘a period €Q
Wer i fime. do
ie regpome —H04 with a
ae ee run) can be determined ;
rc e)®) = ts munya.
Oo
neestate Fouriey | Tranche (OE T) ¢ i
Gat Pouviey Vranshorm EET a
= DFT 6 wed to trunchorm a Hime domaie, Xb)
Sequene to a Freq uenuy domain LK) Sequence.
7 To tranthorm —frequeney domain YR) Sequence fo
ime domain Xin) Seyuend invese DET CryeTy
Widedt asl | &
> Algorithes Poy Fuk computation ot DET 4 rneT
axe Known od” FET algorithm «
the DET Pais > ? Reedomcs
A ge elade She time domain
ai ie ou mie ae MLW) Sequent4l
Ore called the DET pear.
pete XCk)= BE geny SHAT Keo, 1,2, — Car)
Azo
DPFT eqn. "uin) = ‘ = xtK) eatin kl ) Ne0,) 2) > --CH+!)
N => dlemanly number + cements (0 Kin) oF XU) gequend
The Qelationskip behveen OFT _¢ Frequency Retpone
ative acted ONG sac aay O)4,2)-77 (NA).
j Crom bs eqn.
Behe demon or ytk OA oblained eee
Be spared of a Beil frog, ot oN
> The ablous us to we DFT eee
Yonk oo Freq seHNe Oe a
= the corresponding analng | freq. e nq —
beeen elenens” 0) sete” |g
ine Se S08 oodb = bm = Unt = VT A
Tr Signal cttord Lenglh. oa ey
Larger be Signal reovd engin, de rally, 8 |
de Frequency Spactag. \ 7
7 deme he freq Spat )N can be inmaied a
dy Simply Oppending eG he ten) Sequence. \
Vetore, computiog Kk) ‘
The Tat Fourier Tanefam j= CFF T)
fs det compulation o DRT requ
ge Darbey comglex wulkiplies
The alpritrs ob DETELDET se power # 2 pins BAAN
4 erpboit dx penodic nahure ab he compller
exponential @STOKIN. ottaurieg {9 per LOFT
equation»
No Direc DET EET ued oer mulkelia| oo
tomputation ——fempfatton FFT clea L
ie uy Ne U0
q \G Vee @7 LO
\o 256 ee £0
by 4b.” qe 23
256 S536 Aor cog
sve Be to ose esi hoy Se
he No Jog N+ NP
Nori aN
ee Flos
algoribn.
FFT abit dbaled be
DET veo N™ ‘tomplel anublipleyRady 2. —
Fe
TV dgorithen requs
Wequirey MN Jogi
Anplicah
Policakin ob Fer ny
‘ tan be Sy we pig ee
ov signe
Signal
Mat power Spec derihy (PS0),
Spednun.
on
The FEY stlt XCK) tan be wed to compute de
Hee OA.
SC) = (Ya) 2009] = ON) KUO 0K) ¢ te, ie.
hy omociaded voir vost
yae, De signal
SOK y=). rel power der
menies. 4 wed to chonacten\
ae Frequency domed
Lineay Tine = Envartant Sykes (>
|
LU)
touch be superposition
Linea Sygtom i> A cyshem
2 € Reel een CON be applied * \y Koown 4 & Tinea éy dem.
1p sythem Shak deteribed
von ab oll Jima
© Vine. tnvextant
= by de Sane inp
Z i called me invamiant
2) ‘ cys tem >
Linear Same \ovortort Es Se nal : hl
ONE
er itorat a OF.
© Mineo deone | inva Y
tn pe Hime doral\
LIL sys GO he reprucnted
| Maly NNineaw cantar oetticient difference SDS7A unit Sample Cor ‘impule) reiporve iS uted to
thavacterize an LTE Sythem
7 Time domain tonvebtion can be cred fo deterniel
he veporye chan LYE sytem.
Convolution +
=e
7TH Is an operation hh velakel the TP folp ton
LYE sucker 5 th unit Sample Tepore.
8
Yo) 2S hen) ttn-m)= 2 Len) hin-in) = hin) € In)
wen mio
Xoo) =y Tip
SO)=r lp ‘
hun) ey Unit Cample sepore-ch be Sutter.
% => Convolution operation vepreertetion,
the bime
The ahove eqn & wed bo cormpute,
TeLpone a a Syttem Yo oo abi tromy tnpuk Seq
2 ~Tranctoren > z cate
oy Xe . :
We have uD ee gind *
© ty
xe) = Z
Tye subehrution:
58
e
SS wind @
- aay
Xo) = Be
Using
Zs
eatea he of 10)y . rs
e Ss di 2 Sanetorm ot Ytn) eet
wont &) = NW)[ xy :
ed MS Syitem funition or She Tone fe
EVE Sutkem
7 Syskers fy dy Characterize, be systers ene.
domain. ‘
Faces sdhiracterize an LIT sythem 16 epedty 15
i FEROS
=z Paley exe Fre Yook of the, denominator A tre tranttey
Poles
= Bers ame be “oohs A ch ~ numerator -
Digital Filkert \-
ace hie) Thab operatel on an
pulpub
> A filter 16 @ Seque
inpuk Sequence tn) to genete 0 Hl
Sequence wbn).
7 The genevel dittoence equadion dor an wi NID order
Auer WW given On, f :
Un) = Soyer) + e Ag U0) -
: a
Ke\
= The begs Wada Sega, content fecdbath.
= The torrent oulpub depend on pug Golpal
tn oddliion Fo tecrent pl inpuh,\hw prtiedt
ert t
XhaL4 ds bilo
el
Taam
igh Bods Sugar
ceprentation Fadel filler
o (FEM Filler \—
Exmite. Sopube Perporse SO
A simpler Yesion ck bu general Cle difference
equakon
Youn) = E_ by tly-k).
‘The aloovt 0. neh ~ qi
= The unit el sretgonge oF She FIR "ler Jets
box a fethe dine dependent ‘on Se wy
Bley coeticrents ented by BK
The Frequenry TONE cba FIR, Piller oP be
Utes) oe Re
Kz0
a 7 — tranbonn.
=)8.
Z* whe BE &
hw = al
amSince! empuulbeen ALN RW
r i aioe sat aes Shritdure
7 SYMNMNE Rear ;
prae or aati wa Ue OLE, Dinan
smoky por i Hale ain Pret
tarot tolevcde, hoe fete oppdicahory thot
Wadyantage! shor Hon.
1 Obkai, ang detived Frequency reipoye, He rumbey
coefficien i apnerally Jowye- :
ee to runbey A coctfitiends ntquive Jonge tompulhen Fi
me
SY Larger computation Kmne Linils Sampling vile. a
Ly Limtkakion at arsphing role tively eine
Sig
Example i FER blber.
Yus) = 0.5 Ltn) 10'S LUn-)
The elp of abovy eq 15 Og oF uurrert HIP en) +
‘xp Lon-)
ea comple” SUPINE ahh Glley is dbbatned
oy subghitubing , SO) te 0)
Win) = 0-6 SCn) + O'S &ln-t)
= [0S 051 ob a Sequentes
The Fequenny “ORNS
Le) = & bné
K=O : a
= ho Ee abe
Ais vos es
= OM Oe
i Hiz\= 0-5 40.52)The riagnitude Yesponte iSgiven Of
\He56)| = '06) = Col O/2_
The phae vepore given G4
24(08) = ppgyz ~ Ola* Le Oly,
Grou delay
Fee lhe. deloy, to ocala eevee
TEs obleined by oHerenbiahing ¢ negatiog he
phase respore function. k
Group ddey th
Tofinite Lempulie Mesponge (2LR) Fillers}
> Valike FER fillers, EER Ales awe ‘obioike
Surber st -Vorgulse pone Samples.
They one veuunive Piller 04 he olp depends
not only ony be pak 4 pruent ELp bul aio on
The pak olf's . bis tei
H2) = gl Pailin ak Yaa vt MU
\eagt- Gaile ia, Fay ah:
= The stability depends on Ae amber & value
cwoefhidenk.
o Tr generat Tre filler har - non Lineay prove
yUpone < doesnok provide tonttank group delay:
Ark Piller 1) unsuitable for appiicalion, ak
Canna} tolerate prove, dstorhon .
Rdsantones ey schfctenky Yo veabtze o-deiined
=Simalfer number
rupone rdative fo an FIR
Prequieriy po tee :
7 Shorter Corputestion
a tarue bordutdhh,i
Trg FiLbey
ie Aifteronce. RG. ucckion,
Ym) = Og Wore) 4 OVX)
7404 4 paskotp
= Oh
: Pruent Dip, :
The, Sroneboy by of Bay Filey abkeined 4
Wid} = 4 eg ONE
Geoqa) — (@-0-4),
HS HEA) = Otero
(e368 -o.q)
EER filler Dein
detign Procedure A an FIR filler involv be
dekermitakon of he Filler weltiverk bx.
b> 2 (THe) Ede. Wr oto
aT yt
TTA thhey Deaiga’r
Too approathy are used to dutgn LIER Alters,
YBosed on Analog Ser design technique,
21 Bored on Diver Design. i
rR fle design toved om analig Mlfer_ deaten Setriquel
= Digit TIO Fillers ont duigned eyed on epg
aller deign metrods Such 04 ee lle
design iif Chebytheu 1 {ii Ohebyhoo 2 YEN
Bikey bo salisty diller
Be Designing, co ‘
. ication he cornering \k bo fe
digital Peer wala cur app priate BODive TTR Pbbey '
Thue, method, are beyed on Seck square
HM HL bo a died Frequency wAporne-
> Thae method allow oe
Prequenuy TUPONe Specifications.
Thue axe the DS? operakiony Shab Gire wed
to change the, Sampling _yate od a Sequence.
2 Dedmation Swed fo decreaje tre Sampling
meeiniopolahon ee vires ie
> Decimation Invelues dropping cepa * volthouk
Violating Sampling theorem.
> Tnterpolation \nvolua, pote ri
Opprmpriate tondderation ko Samp,
anxound the point of \nsertion.
The Decimation Process |~ s
ye her ae Se Pot 0
Cima te 4 4 Renton \S Lo), y
Btley wed in the deci 5 re
Goltduae einciae SAS ay
= ‘ 21.62 4) 37) bebe
Ze Given LON) » ais A
; Oye Uke aoe
Detimation Factor > >:
, tesa oo) +htn).Ss 8 4
a ie
& 2 |og. ae : :
nin) Sali ie S42 60 3.9 Ry 3:6) eh
my Oyo? Bee 35h 2 4 166
Ba 20 210 919 1a oa
ln) = 102,19 0, 2.42, Sen pa an:
5
4D hg
Sly j~0 Be, 2.08] TAG
z
Yood= Wms) = Lor, 3-62, BAG, AU],
Decimation Fuchs The factoy by whith tre a
iS detimated 6 called a the ecimation fac
The 3 coukpul vdahon , fos decimation vt the
bt, ou \antegey Oe ee ™., iS gwen ay
eB
Yom) = WLm')= = XLmM-k).
coher w YY
Loy)
Figi- Dedmation proce Ling
Facky \-
sii2 To prevent Samplin theorems wolation , ‘.
the Signal is Pt bandhienited ing fA.
filler with bn evefficients .
7 Only Shore Samples Shab needs fo be hep
every oe Carpile} anc faker.
see a akin Pr Px Watexpolatia
33
yum) = 2 be wlrn-k)
K=-e
Glare © rot), = UUM, Me Oe eee
20 boy oft values ub on,
XL Ve wm} piqite
ny ana Be Yun) '
5 tales pA eicon 4
Lec
eS Figts Totexpolation proces slog &n \nter polation fucky |
> Fisk generate de Sequence wtm) by Anverking
(L-1) Zems
7A Bier wilh bk toetticienly
ee ot the Aller equals he ee
original signa sornpley Ne To cornpurting
snterpolaked signal ample -
> The inkrpolating tiller ha fowpacs tiller
ued by Rly tH Wage frequencl4 gener
by mnoreafing Pe Sample rate.