文稿案例教案230b 1 lecture_第1頁
文稿案例教案230b 1 lecture_第2頁
文稿案例教案230b 1 lecture_第3頁
文稿案例教案230b 1 lecture_第4頁
文稿案例教案230b 1 lecture_第5頁
已閱讀5頁,還剩3頁未讀 繼續(xù)免費(fèi)閱讀

下載本文檔

版權(quán)說明:本文檔由用戶提供并上傳,收益歸屬內(nèi)容提供方,若內(nèi)容存在侵權(quán),請(qǐng)進(jìn)行舉報(bào)或認(rèn)領(lǐng)

文檔簡(jiǎn)介

PAGE

2

un

z-1

AdaptiveLinearPrediction

un-1

z-1

z-1

un-m

w1*

X

w2*

X

wm

*

X

u?(n|un1,un2,...,unm)

Theideaistousealinearfiltertopredictthecurrentinput,withthedifferencebetweentheestimateandtheactualvaluehavingthenewinformation.Theestimateischosenasminimizingthevarianceofthisdifferencevariable.Foralongenoughfilter,thisresultsinthedifferencevariablebeinganuncorrelatedsequence.

EE230BProfGregPottie 3

LinearPrediction

Definethedatavectoras

u(un1,...,unm)

Wethensolvetheleastsquaresproblem

wr,where

E[uuH]

r(0)

r*(1)

r(1)

r(0)

...

r(m2)

r(m1)

r*(m1)

r*(m2) ...

r(0)

r E[uu];noteuisascalar.

r*(1)

*

*

n n

r*(m)

EE230BProfGregPottie

4

LinearPredictionII

Wecaninpracticeusethe

algorithm,usingthedifference

variableastheerrorterm.Thisisknownastheforwardpredictionerror,givenby

m

f(n)u wu

*

m

n

knk

k1

Thereissimilarlyastructureknownasabackwardpredictor,whichusesreceivedvaluestopredict(estimate)anearlierreceivedinput.

EE230BProfGregPottie

5

PredictiveDFE

Theideahereistousealinearequalizer(whichproducescorrelatednoise)andthenusealinearpredictortoremovethecorrelationinthedistortion.Sincetheforwardfiltercanbeadaptedindependently,convergenceisfaster,butusuallyattheexpenseofmoreadaptivecoefficientsintotalthantheconventionalDFE.

LEQ

y+

-

-

+

e2

e1

-

Noisepredictionfilter

+

EE230BProfGregPottie

6

PredictiveDFEII

Inthelimitofinfinitefilterlength,thepredictiveDFEhasthesameperformanceastheconventionalDFE

BoththeconventionalandpredictiveDFE’sleadtorelatedMLSEandprecodingstructures.

EE230BProfGregPottie

7

Precoding

TheDFEhastwopotentialdrawbackscomparedtotheLEQ:

AtlowerSNR,itcanbesubjecttoerrorpropagation;onewrongdecisiontriggersalongstringoferrors

Channelcodingrequiresdecisiondelay;thisisfataltoaDFE

Precodingavoidsbothofthispitfallsbymovingthefeedbacksectiontothetransmitter.Inthissectionwewillcover:

theD-transform,convenientnotationfordescribinghowthedeviceswork

modulo-reductionandwhyitisnecessary

costsandbenefitsofprecoding(nothingcomesfor )

EE230BProfGregPottie

8

D-Transform

Thisisreallyjustthez-transform,whereD=one-symboldelay=z-1.Itisconvenientsincesequencescanbeexpressedaspolynomials.

Thereceivedsequenceis

r(D)x(D)h(D)n(D)

wherex(D)xxDxD2...

0 1

2

n(D)w(D),q(D)1qD...,

q(D)

1

theMMSEpredictionerrorfilterforn(D)

h(D)h(D1)Kp(D)p(D1)

p(D)1pDpD2...,

1

2

causal,minphase,spectrallyequivalenttoh(D)

EE230BProfGregPottie

9

Infini

engthConventionalDFE

Usingthisnotation,theforwardfilterisdefinedby

c(D)q(D)[p(D)/h(D)]noisepredictor+allpasstokillprecursors

b(D)q(D)p(D);theFBFisthenb(D)1

ThefullpicturefromchannelinputtoDFEoutputisthenasbelow.

x(D)

+

n(D)

r(D)

p(D)/h(D)

decision

-

1/q(D)

c(D)

b(D)-1

w(D)

EE230BProfGregPottie

10

h(D)

q(D)

Precoder

ConsideranL-pointPAMsignalsetwhereL2,levels1,3,...

i(D)inputdatasequence(Llevels)

x(D)precoderoutput;precoderhasresponseb(D)1

Thenxkikxkjbj2Lzk,k0,

j1

z=integerchosentominimizethevalueofx2.

k

k

Withthisprocedure,xkmustliewithin[-L,L).E.g.,forL=4

2Lz(D)

+

i(D)

-

-

+

b(D)-1

x(D)

EE230BProfGregPottie

11

Precoder

ternativepointofviewistofirstformthesignal

fkikxkjbj,k0,

j1

Thenreduceto[-L,L)usingamodulo2Loperation

i(D)

+

x(D)

-

b(D)-1

EE230BProfGregPottie

12

mod

Receiver

Inthereceiver,aftertheforwardfilter(thesameasfortheDFE),get

v(D)v0v1(D),...

x(D)h(D)c(D)n(D)c(D)

y(D)w(D)

Recallc(D)q(D)[p(D)/h(D)]b(D)/h(D)

Thisiswhythenoiseiswhitefollowingtheforwardfilter

Also,givenx(D)[i(D)2Lz(D)]/b(D)

y(D)[i(D)2Lz(D)]h(D)b(D)

b(D)

i(D)2Lz(D)

h(D)

EE230BProfGregPottie

13

ReceiverII

Theelementsofy(D)lieonananexpandedintegergrid,witharangethatisthesameasaconventionalDFE.Tothisisaddedthenoise.

Tomakeadecision,firstreducemod2Lto[-L,L)toobtainthefoldedsamples

v'(D)i(D)w(D)

ThisisthesamesequenceaswouldbeseenbythedecisiondeviceoftheDFE(assumingnoerrors),becausei(D)livesonlyon(-L,L)and|zk|=0orisgreaterthanorequalto1.Theoverallsystemis

i(D)

x(D)

r(D)

v(D)

mod h(D) c(D) mod decision

-

b(D)-1

n(D)

EE230BProfGregPottie

14

BenefitsofModuloDevices

Withoutnoiseormodulodeviceswecouldhaveobtained

x(D)i(D)/b(D)

v(D)i(D)h(D)b(D)i(D)

b(D) h(D)

However,x(D)haslargepeakvaluesandthesequenceiscorrelated.

Themodulooperationinthetransmit

mountstosubtractionby

therandomsequence2Lz(D).Itreducesthepeakpoweranddecorrelatesthesequence(reducingaveragepoweralso).Withmoduloreduction,x(D)isagainapproxima ywhitewiththesamplesbeingroughlyuniformover(-L,L)(perfectlyasLgoestoinfinity,anexcellentapproximationalreadyforL=4).

EE230BProfGregPottie 15

CostsandBenefitsofPrecoding

Foruniformlydistributedxk,L=4,theaveragepoweris5.33,vs.5for4-PAM.ThissmallpowerpenaltyvanishesasLgrowslarge.

Noiseisunaffectedbythemodulooperations--sothisistheonlyperformancepenaltyforastaticchannel.Inexchangeweavoiderrorpropagationandcaneasilyusechannelcodes,whilegettingtheperformanceofaDFE

Adaptationismoredifficult:mustfirstlearnthefeedbackfilterbyadaptingaconventionalDFE,andthentransmitFBFcoefficientstothetransmitter.Thiscanbedoneonlyperiodically,ands

溫馨提示

  • 1. 本站所有資源如無特殊說明,都需要本地電腦安裝OFFICE2007和PDF閱讀器。圖紙軟件為CAD,CAXA,PROE,UG,SolidWorks等.壓縮文件請(qǐng)下載最新的WinRAR軟件解壓。
  • 2. 本站的文檔不包含任何第三方提供的附件圖紙等,如果需要附件,請(qǐng)聯(lián)系上傳者。文件的所有權(quán)益歸上傳用戶所有。
  • 3. 本站RAR壓縮包中若帶圖紙,網(wǎng)頁內(nèi)容里面會(huì)有圖紙預(yù)覽,若沒有圖紙預(yù)覽就沒有圖紙。
  • 4. 未經(jīng)權(quán)益所有人同意不得將文件中的內(nèi)容挪作商業(yè)或盈利用途。
  • 5. 人人文庫網(wǎng)僅提供信息存儲(chǔ)空間,僅對(duì)用戶上傳內(nèi)容的表現(xiàn)方式做保護(hù)處理,對(duì)用戶上傳分享的文檔內(nèi)容本身不做任何修改或編輯,并不能對(duì)任何下載內(nèi)容負(fù)責(zé)。
  • 6. 下載文件中如有侵權(quán)或不適當(dāng)內(nèi)容,請(qǐng)與我們聯(lián)系,我們立即糾正。
  • 7. 本站不保證下載資源的準(zhǔn)確性、安全性和完整性, 同時(shí)也不承擔(dān)用戶因使用這些下載資源對(duì)自己和他人造成任何形式的傷害或損失。

最新文檔

評(píng)論

0/150

提交評(píng)論