1Wri

teanoteonLinearAdapt
iveEqual
i
zerbas
edonMMSE(
Mini
mum
Meansquar
eerr
or)techni
que.
Li
nearAdapt i
veEqualizerbasedonMMSE( Mini
mum MeanSquar eError
)techni
queisasi
gnal
process
ingalgor
ithm usedtomiti
gatetheeff
ectsofchanneldi
stor
tionandnoisein
communi cat
ionsyst
ems .Thepurposeofanequali
zeristorest
orethedist
ortedsi
gnalt
oit
s
ori
ginalf
orm bycompens ati
ngforthechanneli
mpairments.

TheMMSEt echniqueisawi delyusedapproachinadapt

iveequali
zati
onbecauseitaimsto
minimizethemeans quareerrorbetweentherecei
vedsi
gnalandthetransmit
tedsignal
.By
minimizi
ngtheer ror
,theequal i
zercanest
imateandcorr
ectforthechanneldi
stor
ti
on,
res
ulti
nginimpr oveds i
gnalqualit
y.

Her
e'sabr
iefex
planat
ionofhowt
heLi
nearAdapt
iveEqual
i
zerbas
edonMMSEt
echni
que
wor
ks :

1.I
nit
ial
izat
ion:Theequal
i
zerisi
nit
ial
i
zedwit
hini
ti
altapwei
ght
sorcoef
fi
ci
ent
s.Thes
e
coeff
ici
entsareadjus
teddur
ingtheadapt
ati
onprocesst
omini
mizet
hemeansquareer
ror
.

2.Tr
aini
ng:Atr
aini
ngsequenceorpi
lotsymbol
saret
rans
mitt
edthr
oughthechannel
,whi
ch
areknownatboththetr
ansmitt
erandrecei
ver
.Ther
ecei
vedsymbol
sareusedtoest
imat
e
thechannel
res
ponse.

3.ChannelEst
imati
on:Usingtheknowntraini
ngs
ymbolsandt
her
ecei
vedsymbol
s,t
he
equali
zerest
imatesthechannelr
espons
e. Thi
ses
timat
ioni
sus
edt
oadjustt
hetapweight
sof
theequali
zer
.

4.Equal
izati
on:Theequali
zerusest
heestimatedchannelres
ponsetocompensat
ef ort
he
channeldis
tort
ionint
hereceivedsi
gnal
.Itappli
esali
nearcombinati
onofther
eceived
symbolswiththetapweightstogener
ateanes ti
mateofthetrans
mitteds
ymbols
.

5.Err
orCalcul
ati
on:Thediffer
encebetweent
heest
imatedt
rans
mi t
teds
ymbolsandthe
recei
vedsymbolsiscalcul
atedtodet
erminet
heer
ror.
Theerr
orisusedt
oupdatethetap
weightsoftheequali
zer.

6.TapWei
ghtUpdat e:Thet
apwei ghtsareupdat
edus i
nganadaptiveal
gori
thm,suchasthe
l
eastmeansquares(LMS)orrecurs
iveleas
tsquar
es(RLS)algor
it
hm.Thegoalist
omi ni
mize
themeansquar
eer r
orbetweent heesti
matedandreceivedsymbol
s.

7.I
terati
on:St
eps3t o6ar
erepeat
edi
ter
ati
vel
yunt
ilt
hemeans
quar
eer
rori
smi
nimi
zedora
convergencecri
ter
ioni
smet.
2Wr
it
eanot
eonZer
ofor
cingequal
i
zer
.

I
nt elecommuni cat ion, I
nt er symbol i
nt erference( I
SI)isaf orm ofdi storti
on
ofas ignal inwhi chones ymbol interfer eswi t
hs ubs equents ymbol s.The
pres enceofI SI int hes ystem i ntroduceser rorsint hedeci siondevi ceat
ther ecei verout put .Ther ef ore, i
nt hedes i
gnoft het ransmi tti
ngand
receivingf i
lt
er s, theobj ect ivei stomi ni mizet heef fectsofI SI,andt hereby
delivert hedi gi tal dat at oi tsdes t
inationwi ththes mal l
ester r
orrat e
pos sible.Inthi spaper ,
wedes ignazer of orcingequal i
zer ,whi chisat ype
oflinearequal izerus edt omi t
igatet heef fectsofI SI (
Inters ymbol
I
nt erference) . Thezer of or cingequal izerdoest hisbyf indi ngt heinver se
oft hei mpul ser es pons eoft hechannel .TheI SIcanbes tudi edus i
ngeye
diagr am, whi chover laysmanys ampl esofas i
gnal andcangi vea
graphi cal repr es ent at i
onoft hes ignal char acteri
s t
ics.Int hispapert he
resultshavebeens hownwi thpr opereyedi agrams .I
nt hispaperweal so
i
nves ti
gat et hebi ter rorr at e, whichi sus edt oes timat etheper f
ormance
char act eri
sticsoft hezer of or ci
ngequal izer.
3Expl
aint
heCDMAs
yst
em bas
edonI
S95f
orwar
dli
nk

 CELPcodergenerat
esavar
iabl
edat
arat
eof9600,
4800,
2400,
and1
200hi
ts/
s,(
bas
ed
speechact
ivi
ty)i
nthef
rameof20ms.
 Dat
aisencodedbyar
ateof1
/2,
cons
trai
ntl
engt
hK-
0convol
uti
onal
encoder
.
 Forlowerspeechactivi
ty4800,2400,
or1200bit
s/st
heout
putsymbol
sfr
om t
he
convoluti
onalencoderarerepeat
edei
thertwi
ce,f
ourti
mes
,orei
ghtti
messoast
o
maintainacons t
antbitr
ateof9600bits/
s
 Blockint
erl
eaveri
susedtoovercometheef
fect
sofbur
ster
ror
sthatmayoccurdur
ing
thetransmi
ssi
onthroughthechannel
.
 Scr
ambl
eri
sus
edf
orDat
aEnc
rypt
iont
omakecal
lmor
esecur
e.
 Scr
ambl
erwi
l
lrandomi
zesdat
aandpr
event
sthet
rans
iti
onofal
ongs
eri
esof1
'sor'
s
 Bl
ocki
nter
leaverdat
arat
eof192kbpsarescrambledbymult
ipl
i
cat
ionwithl
ungcode
wi
thchi
prateof12288Mchips/
sandisdecimatedbyfact
orof64t
o1 9.
2kehi
ps/
s.
 Thel
ongcodei
sus
edt
oident
if
yacal
lofaMSont
hef
orwar
dandr
ever
sel
i
nks
 Hadmar
dorWal
shcodes
equenceofl
engt
h64i
sas
signedt
oeachchannel
 64or t
hogonalsequencesareass
ignedtoeachBS(64Chamels)
,Onechanneli
sused
totrustpi
lots
ignal
,whichi
susedmeas ur
ethechannel
charact
eri
sti
cs(
incl
udessi
gnal
str
engt handthecarr
ierphaseof
fert
)
 Anotherchannel
isus
edt
opr
ovi
det
imes
ynchr
oni
zat
ion,
unechannel
forpagi
ng
act
ivi
ty,
 Remai
ning61channel
sforus
er
 Eachus
erdat
aismulti
pli
edbyHadmardsequence,t
heresul
ti
ngserpeneri
sspreadby
twoPNser
pensone.
in-phas
eandotheri
nquadraturephaseofl
engthN=2^1 5
 Di
ff
erentBSar
eident
if
iedbydi
ff
erentof
fset
soft
hes
ePNs
equences
.
4Expl
aint
heCDMAs
yst
em bas
edonI
S95r
ever
sel
i
nk

 Ar
ateof1
/3K-
9convol
uti
onal
codei
sus
edi
nther
ever
sel
i
nk

.
Thecodedbi
trat
eis28.
8kbps

 Foreach20msframet
he576encodedbi
tsar
ebl
ocki
nter
leaved
andpassedt
othemodul
ator

 Thedat
aismodulatedusi
nganM-64or
thogonal
signal
sof
Hadi
mardsequenceseachofl
engt
h64

 6bitbl
ockofdat
aismappesi
nto64Hadamar
dseq=>(
64/
6)
*28.
8=307.
8

 64-ar
yor
thogonal
modul
ati
onus
ingt
hes
ameWal
shf
unct
ioni
nthe
for
wardl
ink

 Cont
rar
ytot
hef
orwar
dli
nk,
usedf
oror
thogonal
dat
amodul
ati
on

 OneWal
shf
unct
ioni
str
ans
mit
tedf
ors
ixcodedbi
ts

 Modul
ateds
ymbol
rat
e:28.
8kbps
*64chi
ps/
6codedbi
ts=307.
2kcps

 Thes
ignal
iss
preadbyl
ongPNcodewi
thar
ateof1
.2288Mchi
ps/
s.