GHVLJQ Ri Zluhohvv FRPPXQLFDWLRQ V - VWHPV LVVXHV RQ V - Qfkurql) DWLRQ/ Fkdqqho HVWLPDWLRQ DQG Pxowl0Fduulhu V - VWHPV

Ghvljq ri Zluhohvv Frppxqlfdwlrq V|vwhpv
Lvvxhv rq V|qfkurql}dwlrq/ Fkdqqho

Hvwlpdwlrq dqg Pxowl0Fduulhu V|vwhpv
Iuhguln Wxiyhvvrq
Oxqg 5333
ll
Ghsduwphqw ri Dssolhg Hohfwurqlfv

Oxqg Xqlyhuvlw|
Er{ 44;
VH0554 33 OXQG
Vzhghq
Qr1 4<
LVVQ 47350;995
LVEQ <40:;:7038<05
f Iuhguln Wxiyhvvrq/ 5333

h{fhsw zkhuh rwkhuzlvh vwdwhg
Sulqwhg lq Vzhghq e| NIV DE/ Oxqg
Dxjxvw 5333
Devwudfw
Wklv wkhvlv ghdov zlwk fhuwdlq dvshfwv lq wkh ghvljq ri zluhohvv frppxqlfdwlrqv
v|vwhpv1 Lw lv irfxvhg rq sureohpv uhodwhg wr wkh preloh ru zluhohvv fkdqqho=
v|qfkurql}dwlrq/ fkdqqho hvwlpdwlrq dqg ghvljq ri zluhohvv ruwkrjrqdo iuhtxhqf|
glylvlrq pxowlsoh{ +RIGP, v|vwhpv1 Wkhuh lv d vkruw lqwurgxfwlrq wr wkh hog
ri zluhohvv v|vwhpv dqg d ghhshu uhylhz ri shuylrxv zrun dqg wkh vwdwh ri wkh
duw lq hdfk ri wkh uhvhdufk hogv1 Wkurxjkrxw wkh wkhvlv wkh jrdo kdv ehhq
wr dqdo|}h wkh sureohpv dqdo|wlfdoo|/ ghulylqj h{suhvvlrqv iru wkh uhvxowlqj elw
huuru udwh +EHU, ru v|qfkurql}dwlrq shuirupdqfh1
Wkh zrun rq fkdqqho hvwlpdwlrq lv irfxvhg rq wkh lq xhqfh ri wkh slorw
sdwwhuq lq RIGP v|vwhpv dqg rq wkh srvvlelolw| ri frpshqvdwlqj wkh vljqdo
lq dgydqfh iru lpsdluphqwv lqwurgxfhg e| wkh fkdqqho1 Wkh EHU lv ghulyhg
dqg frpsduhg zkhq xvlqj glhuhqw slorw sdwwhuqv lq dq RIGP v|vwhp1 Iru
wkh suh0frpshqvdwhg v|vwhp wkh EHU/ pd{lpxp eorfn ohqjwk/ dqg uhvxowlqj
shdn0wr0dyhudjh srzhu udwlr duh lqyhvwljdwhg zkhq frpshqvdwlqj wkh srzhu
dwwhqxdwlrq dqg2ru wkh skdvh urwdwlrq lqwurgxfhg e| wkh fkdqqho1
Wkh zrun rq zluhohvv RIGP v|vwhp ghvljq lv gluhfwhg wrzdugv suhdpeoh0
edvhg v|qfkurql}dwlrq dqg rswlpl}dwlrq ri wkh vxe0fkdqqho edqgzlgwk1 D qhz
v|qfkurql}hu vwuxfwxuh lv lqwurgxfhg iru wlph dqg iuhtxhqf| v|qfkurql}dwlrq1 Lq
wklv v|qfkurql}hu wkh rughu ri fruuhodwlrq dqg pxowlsolfdwlrq kdv ehhq fkdqjhg
frpsduhg wr frqyhqwlrqdo v|qfkurql}huv/ zklfk uhvxowv lq vljqlfdqwo| lpsuryhg
ghwhfwlrq shuirupdqfh1 Wkh vxe0fkdqqho edqgzlgwk lv rswlpl}hg edvhg rq wkh
h{shfwhg Grssohu iuhtxhqf| dqg h{fhvv ghod| ri wkh fkdqqho1
Wkh xvh ri vxshulpsrvhg slorw vhtxhqfhv lv lqwurgxfhg iru wlph dqg iuh0
txhqf| v|qfkurql}dwlrq lq RIGP v|vwhpv/ dv zhoo dv iru fkdqqho hvwlpdwlrq lq
dw idglqj fkdqqhov1 Dv rssrvhg wr frqyhqwlrqdo slorw0edvhg v|vwhpv/ wkh slorw
vhtxhqfh lv olqhduo| dgghg wr wkh gdwd vhtxhqfh dqg wkhuhiruh wudqvplwwhg vlpxo0
wdqhrxvo| dqg lq wkh vdph edqg dv wkh gdwd vhtxhqfh1 Wkh shuirupdqfh zkhq
xvlqj vxshulpsrvhg slorw vhtxhqfhv iru v|qfkurql}dwlrq dqg iru mrlqw fkdqqho
hvwlpdwlrq dqg ghwhfwlrq lv dqdo|}hg1 Wkh vfkhph lv dovr frpsduhg wr rwkhu
zhoo nqrzq slorw0edvhg dqg glhuhqwldo ghwhfwlrq phwkrgv1
Nh| zrugv= Zluhohvv/ ruwkrjrqdo iuhtxhqf| glylvlrq pxowlsoh{/ v|qfkur0
ql}dwlrq/ fkdqqho hvwlpdwlrq/ ghwhfwlrq/ slorw sdwwhuq/ suh0frpshqvdwlrq/ suh0
frglqj/ vxe0fkdqqho edqgzlgwk/ suhdpeohv/ vxshulpsrvhg slorw vhtxhqfhv
lll
ly DEVWUDFW
Suhidfh
Zkhq L vwduwhg dv d Sk1G1 vwxghqw p| dgylvru/ Wruohly Pdvhqj/ h{sodlqhg wr

ph wkdw RIGP kdg d srwhqwldo ri ehlqj dq h!flhqw dqg dwwudfwlyh prgxodwlrq
whfkqltxh iru ixwxuh zluhohvv dqg preloh v|vwhpv1 Wkhuh zhuh/ krzhyhu/ vwloo
vrph nh| frpsrqhqwv wkdw kdg wr eh lpsuryhg dqg uhtxluhg ixuwkhu uhvhdufk1
Fkdqqho hvwlpdwlrq zdv rqh ri wkhvh duhdv/ v|qfkurql}dwlrq dqrwkhu1 H!flhqw
orz0frpsoh{ fkdqqho hvwlpdwlrq whfkqltxhv zhuh uhtxluhg wr hqdeoh frkhuhqw
ghwhfwlrq dqg kljkhu rughu prgxodwlrq vfkhphv/ zkloh urexvw v|qfkurql}dwlrq
whfkqltxhv zhuh uhtxluhg iru rshudwlrq lq kljko| glvwxuehg hqylurqphqwv1
Zh ghflghg wr vwduw zlwk fkdqqho hvwlpdwlrq dqg d vwxg| ri fkdqqho hvwlpd0
wlrq whfkqltxhv1 Lw zdv riwhq vwdwhg wkdw wkh orvv dqg ryhukhdg fdxvhg e| wkh
slorw v|perov lq frkhuhqw v|vwhpv zhuh wrr odujh1 Wkhuhiruh zh orrnhg iru/ dqg
irxqg/ d zd| wr uhgxfh wkh qxpehu ri uhtxluhg slorw v|perov1 Zkhq vwxg|lqj
wkh lq xhqfh ri wkh fkdqqho lq RIGP v|vwhpv zh uhdol}hg wkdw wkh v|perov
frxog qrw eh duelwudulo| orqj> wkh fkdqqho fkdqjhv kdyh wr eh vpdoo zlwklq hdfk
v|pero1 Rq wkh rwkhu kdqg/ wkh h{fhvv ghod| ri wkh fkdqqho fdoohg iru orqj
v|perov1 Zh wkhuhiruh vwduwhg wr lqyhvwljdwh wkh rswlpdo v|pero ohqjwk +dqg
wkhuhe| wkh rswlpdo vxe0fkdqqho edqgzlgwk, dqg irxqg dq dqdo|wlfdo h{suhv0
vlrq ri krz wr pdnh d frpsurplvh ehwzhhq wkhvh fkdqqho uhodwhg hhfwv1
Zkhq suhvhqwlqj wkh uhvxowv rq fkdqqho hvwlpdwlrq L jrw wkh txhvwlrq iurp
d jxhvw surihvvru/ Plnh Idxonqhu/ li L frxog vd| vrphwklqj derxw wkh wlph
d fkdqqho hvwlpdwh lv ydolg1 Wkh dqvzhu L jdyh zdv wkdw lw ghshqgv rq wkh
Grssohu iuhtxhqf|/ exw L zdv qrw deoh wr jlyh dq| jxuhv1 Zh glvfxvvhg wkh
txhvwlrq pruh ghhso| dqg Plnh vxjjhvwhg xvlqj dq dqdo|vlv vlplodu wr wkh fkdq0
qho hvwlpdwlrq sdshu wr ghwhuplqh wkh pd{lpdo ghod|1 L fkdqjhg wkh dqdo|vlv
dffruglqjo| dqg ghulyhg h{suhvvlrqv iru wkh pd{lpdo eorfn ohqjwkv lq d v|vwhp
zkhuh fkdqqho hvwlpdwlrq wdvnv duh shuiruphg lq wkh edvh vwdwlrq rqo|1
Rqh gd| dq DVLF ghvljqhu dw wkh ghsduwphqw/ Vkrxvkhqj Kh/ dvnhg ph
zk| SQ0vhtxhqfhv zhuh qrw xvhg iru v|qfkurql}dwlrq lq RIGP v|vwhpv1 Kh
kdg vrph sureohp jhwwlqj wkh v|qfkurql}dwlrq xqlw wr zrun surshuo| dqg zh
wkrxjkw wkdw pd|eh rqh frxog frpelqh d SQ0vhtxhqfh zlwk wkh RIGP vljqdo
lq vrph zd|1 L glvfxvvhg wkh sureohp zlwk Shwhu Krhkhu/ Plnh Idxonqhu dqg
Ryh Hgiruv dqg zh ghflghg wkdw wkh lghd zdv zruwk ixuwkhu lqyhvwljdwlrqv1 Wkh
uhvxowlqj vxshulpsrvhg slorw v|vwhp vkrzhg jrrg v|qfkurql}dwlrq shuirupdqfh
y
yl SUHIDFH
dovr iru orz VQUv> lw uhdoo| vhhphg wr eh d jrrg dowhuqdwlyh wr frqyhqwlrqdo v|q0
fkurql}dwlrq1 L vwduwhg wr sod| durxqg zlwk wkh sdudphwhuv ri wkh vxshulpsrvhg
slorw vhtxhqfhv dqg frpsduhg wkh uhvxowv wr wkdw ri frqyhqwlrqdo v|qfkurql}d0
wlrq1 E| xvlqj rqo| wkh slorw vhtxhqfh wkh ghwhfwlrq shuirupdqfh zdv pxfk
ehwwhu frpsduhg wr frqyhqwlrqdo v|qfkurql}dwlrq dqg wkhuhiruh zh dqdo|}hg
wklv suhdpeoh0edvhg v|qfkurql}dwlrq v|vwhp ixuwkhu1 Zh zhuh dovr wklqnlqj
derxw wkh srvvlelolwlhv ri xvlqj wkh vxshulpsrvhg slorw vhtxhqfhv iru fkdqqho
hvwlpdwlrq vxfk wkdw qr rwkhu slorw v|perov zhuh uhtxluhg1 Shwhu Krhkhu fdph
xs zlwk d vxlwdeoh uhfhlyhu vwuxfwxuh dqg klv vlpxodwlrqv vkrzhg yhu| surplv0
lqj uhvxowv1 Odwhu/ L pdgh dq dqdo|wlfdo dqdo|vlv dqg lqyhvwljdwhg wkh v|vwhp
ixuwkhu1 Diwhu wkdw/ Ryh dqg L zhuh glvfxvvlqj krz wr frpsduh wkh ghwhfwlrq
vfkhph wr rwkhu vfkhphv1 Zh wulhg wr xvh wkh dqdo|vlv ghulyhg iru vxshulpsrvhg
slorw vhtxhqfhv dqg lw zrunhg zhoo iru wkh prvw lpsruwdqw ghwhfwlrq phwkrgv1
Vr lq wklv zd| zh zhuh deoh wr frpsduh wkh ehkdylru iru glhuhqw Grssohu udwhv
dqg VQU ydoxhv1
Dv vhhq deryh pdq| shuvrqv kdyh ehhq lqyroyhg lq p| uhvhdufk1 Doo wkh wlph
Wruohly kdv ehhq e| p| vlgh wr khos ph/ exw prvwo| dv jxlgh iru ixwxuh uhvhdufk
duhdv1 Wkh uhvhdufk kdv ehhq irfxvhg rq fkdqqho uhodwhg sureohpv lq zluhohvv
frppxqlfdwlrq v|vwhpv/ prvwo| v|qfkurql}dwlrq dqg fkdqqho hvwlpdwlrq sure0
ohpv1 Wkh lvvxhv suhvhqwhg uh hfw wkh uhvhdufk duhdv L kdyh ehhq lqyroyhg lq
gxulqj p| |hduv dw wkh Ghsduwphqw ri Dssolhg Hohfwurqlfv dqg vkrxog wkhuhiruh
qrw eh vhhq dv d frpsohwh ryhuylhz ri wkh v|vwhp ghvljq surfhvv1
Wkh wkhvlv frqvlvwv ri dq ryhuylhz ri zluhohvv frppxqlfdwlrq v|vwhpv/ d
edfnjurxqg ri wkh uhvhdufk duhd dqg vhyhq sdshuv1 Wkhuh lv d vkruw ghvfulswlrq
ri wkh sdshuv dqg wkhlu pdlq uhvxowv diwhu wkh edfnjurxqg vhfwlrq1 Wkh sd0
shuv kdyh ehhq uhirupdwwhg wr lpsuryh uhdgdelolw|/ exw h{fhsw zkhuh h{solflwo|
pdunhg wkh frqwhqw lv wkh vdph dv ruljlqdoo| sxeolvkhg1 Wkh lqfoxghg sdshuv
duh=
D I1 Wxiyhvvrq dqg W1 Pdvhqj1 Slorw dvvlvwhg fkdqqho hvwlpdwlrq iru RIGP

lq preloh fhooxodu v|vwhpv1 Lq Surf1 LHHH Yhklfxodu Whfkqrorj| Frqihu0
hqfh/ sdjhv 496<4976/ Skrhql{/ XVD/ Pd| 4<<:1
E I1 Wxiyhvvrq dqg W1 Pdvhqj1 Rswlpl}dwlrq ri vxe0fkdqqho edqgzlgwk iru

preloh RIGP v|vwhpv1 Lq G1 Hyhulww dqg P1 Uxpvhzlf}/ hglwruv/ Pxowl0
dffhvv/ prelolw| dqg whohwud!f 0 dgydqfhv lq zluhohvv qhwzrunv/ sdjhv 436
4471 Noxzhu Dfdghplf Sxeolvkhuv/ Gruguhfkw/ Wkh Qhwkhuodqgv/ 4<<;1
F I1 Wxiyhvvrq/ P1 Idxonqhu/ dqg W1 Pdvhqj1 Suh0frpshqvdwlrq iru Ud|ohljk

idglqj fkdqqhov lq wlph glylvlrq gxsoh{ RIGP v|vwhpv1 Dffhswhg iru
sxeolfdwlrq lq Zluhohvv Shuvrqdo Frppxqlfdwlrqv/ 53331
G I1 Wxiyhvvrq dqg R1 Hgiruv1 Suhdpeoh0edvhg wlph dqg iuhtxhqf| v|q0

fkurql}dwlrq iru RIGP v|vwhpv1 Vxeplwwhg wr LHHH Mrxuqdo rq Vhohfwhg
Duhdv lq Frppxqlfdwlrqv/ Mdqxdu| 53331
yll
H I1 Wxiyhvvrq/ P1 Idxonqhu/ S1 Krhkhu/ dqg R1 Hgiruv1 RIGP wlph

dqg iuhtxhqf| v|qfkurql}dwlrq e| vsuhdg vshfwuxp slorw whfkqltxh1 Lq
Surf1 Hljkwk Frppxqlfdwlrq Wkhru| Plql0Frqihuhqfh lq frqmxqfwlrq
zlwk LHHH LFF *<</ sdjhv 44844</ Ydqfrxyhu/ Fdqdgd/ Mxqh 4<<<1
I I1 Wxiyhvvrq dqg S1 Krhkhu1 Fkdqqho hvwlpdwlrq xvlqj vxshulpsrvhg
slorw vhtxhqfhv1 Vxeplwwhg wr LHHH Wudqvdfwlrqv rq Frppxqlfdwlrqv/
Dsulo 53331
J I1 Wxiyhvvrq dqg R1 Hgiruv1 D frpsdudwlyh dqdo|vlv ehwzhhq glhuhqw
ghwhfwlrq vfkhphv lq dw Ud|ohljk0idglqj fkdqqhov1 Lq Surf1 LHHH Yhklf0
xodu Whfkqrorj| Frqihuhqfh/ sdjhv 45380453</ Wrn|r/ Mdsdq/ Pd| 53331
Sduwv ri wkh frqwhqwv lq wkh wkhvlv dqg sdshuv kdyh dovr ehhq sxeolvkhg
hovhzkhuh1 Lq sulqflsoh L kdyh ehhq lqyroyhg lq irxu glhuhqw uhvhdufk surmhfwv/
zkhuh L kdyh ehhq wkh dxwkru ru fr0dxwkru ri wkh iroorzlqj sxeolfdwlrqv=
Fkdqqho hvwlpdwlrq lq RIGP v|vwhpv ^9<` ^:9` ^::` ^;3` ^;4` ^;5`
Zluhohvv RIGP v|vwhp ghvljq ^:3` ^:<` ^;6`
Suhdpeoh0edvhg wlph dqg iuhtxhqf| v|qfkurql}dwlrq iru RIGP ^:5` ^:6`
^:7`
Fkdqqho hvwlpdwlrq dqg v|qfkurql}dwlrq e| vxshulpsrvhg slorw vhtxhqfhv
^65` ^66` ^:4` ^:8` ^:;`1
ylll SUHIDFH
Dfnqrzohgjphqwv
Gxulqj p| wlph dv d Sk1G1 vwxghqw/ L kdyh kdg wkh rssruwxqlw| wr jhw wr nqrz
d orw ri lqwhuhvwlqj dqg fohyhu shrsoh1 L zrxog olnh wr wkdqn |rx doo zkr/ lq
vrph zd| ru wkh rwkhu/ kdyh frqwulexwhg wr wklv zrun ru wr wkh juhdw vrfldo
hqylurqphqw1
Hvshfldoo|/ L zrxog olnh wr wkdqn Ryh Hgiruv iru ehlqj vxfk d jrrg iulhqg/
iru fuhdwlqj vxfk d jrrg dwprvskhuh dqg iru vkdulqj klv zlgh nqrzohgjh derxw
doprvw hyhu|wklqj1 Wkhuh lv dozd|v wlph iru txhvwlrqv dqg glvfxvvlrqv/ qr
pdwwhu li wkh| duh derxw ghwdlov lq d vljqdo surfhvvlqj dojrulwkp ru derxw wkh
vxlwdelolw| ri vlwwlqj lq d fdqrh zkhq fkdvhg e| d klssrsrwdpxv1
L zrxog dovr olnh wr vshfldoo| wkdqn Mrkdq Krnihow1 Zh vwduwhg dw wkh gh0
sduwphqw dw wkh vdph wlph dqg lw kdv uhdoo| ehhq d sohdvxuh kdylqj vxfk d jrrg
froohdjxh ehlqj rq wkh vdph ohyho1 Lw lv qrw riwhq wkdw |rx kdyh wkh sulylohjh wr
zrun lq vxfk forvh frrshudwlrq zlwk d iulhqg dqg fodvvpdwh iurp mxqlru vfkrro1
L dp dovr judwhixo wr Wruohly Pdvhqj iru lqwurgxflqj ph wr wkh zruog ri
zluhohvv frppxqlfdwlrqv dqg iru vwduwlqj xs pdq| ri wkh surmhfwv suhvhqwhg lq
wkh wkhvlv1 Hyhu vlqfh L mrlqhg wkh ghsduwphqw kh kdv ehhq e| p| vlgh vkdulqj
klv nqrzohgjh dqg lqwxlwlrq1 Zlwkrxw klp/ wkhuh zrxog ghqlwho| qrw eh dq|
wkhvlv dw doo1
Wkhuh duh pdq| rwkhu shuvrqv zruwk phqwlrqlqj lq wkh udglr frppxqlfd0
wlrqv jurxs dqg dw wkh ghsduwphqw/ erwk iru wkhlu khos dqg iru frqwulexwlqj wr
vxfk d jrrg dwprvskhuh1 L zrxog hvshfldoo| olnh wr wkdqn Fduo Iuhguln Ohdqghu0
vrq/ Gdqlho Odqgvwup dqg Ehqjw0Duqh Prolq/ wkh odwwhu dovr iru surriuhdglqj
dqg surylglqj ydoxdeoh vxjjhvwlrqv uhjduglqj wkh pdqxvfulsw1
L zrxog dovr olnh wr wkdqn vrph shrsoh rxwvlgh wkh ghsduwphqw iru iuxlwixo
frrshudwlrq1 L kdyh kdg wkh sulylohjh wr zrun wrjhwkhu zlwk Plnh Idxonqhu iru d
orqj wlph1 Iluvw/ kh zdv d ylvlwlqj surihvvru dw wkh ghsduwphqw dqg odwhu L ylvlwhg
Ylfwruld Xqlyhuvlw|/ Phoerxuqh/ Dxvwudold/ iru kdoi d |hdu1 Lw lv uhdoo| lqvslulqj
wr zrun wrjhwkhu zlwk vxfk d shuvrq lqwhuhvwhg lq hyhu|wklqj uhodwhg wr zluhohvv
frppxqlfdwlrq/ qr pdwwhu li lw lv derxw olqhdul}lqj dpsolhuv ru rswlpl}lqj
SQ vhtxhqfhv1 L dovr uhdoo| dssuhfldwh wkh frrshudwlrq zlwk Whold Uhvhdufk/
Pdop/ dqg hvshfldoo| zrunlqj wrjhwkhu zlwk Shwhu Nduovvrq dqg Fkulvwldq
Ehujomxqj1 L kdyh uhdoo| hqmr|hg wkh rshq dwprvskhuh dw Whold Uhvhdufk dqg
wkhuh L ohduqhg d orw derxw vwdqgdugl}dwlrq dqg zluhohvv v|vwhpv lq sudfwlfh1 L
l{
{ DFNQRZOHGJPHQWV
kdyh dovr yhu| pxfk hqmr|hg zrunlqj wrjhwkhu zlwk Shwhu Krhkhu dw Xqlyhuvlw|
ri Nlho/ Jhupdq|1 Lw kdv uhdoo| ehhq d sohdvxuh kdylqj wkh rssruwxqlw| wr
frrshudwh zlwk vxfk d fohyhu dqg frpshwhqw shuvrq1
Ilqdoo|/ L dp yhu| judwhixo wr p| zlih/ Shwud/ zkr kdv hqfrxudjhg ph doo wkh
wlph dqg wr p| idplo| dqg iulhqgv zkr kdyh glvwudfwhg ph iurp zrun gxulqj
wkh vdph shulrg1
Wklv zrun kdv ehhq qdqfhg e| QXWHN +Wkh Vzhglvk Qdwlrqdo Erdug iru
Lqgxvwuldo dqg Whfkqlfdo Ghyhorsphqw,/ lq wkh whohfrppxqlfdwlrqv surjudp
+surmhfw qr1 ;7550<70394<8,1 L zrxog olnh wr wkdqn wkh iroorzlqj rujdql}dwlrqv
dqg irxqgdwlrqv/ surylglqj qdqfldo vxssruw iru frqihuhqfhv dqg ylvlwv deurdg=
Fodhv Dghovnogv plqqhvirqg/ Nxqjoljd i|vlrjudvnd voovndshw/ Whohirqdnwlher0
odjhw O P Hulfvvrqv vwliwhovh iu iupmdqgh dy hohnwurwhnqlvn iruvnqlqj dqg
Flylolqjhqmu Kdnrq Kdqvvrqv vwliwhovh1
Frqwhqwv
Devwudfw lll
Suhidfh y
Dfnqrzohgjphqwv l{
L Ryhuylhz ri wkh Uhvhdufk Duhd 4

4 Lqwurgxfwlrq 6
414 Gljlwdo Frppxqlfdwlrq V|vwhpv 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 7
415 Zluhohvv V|vwhpv 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 9
41514 Sdfnhw0Edvhg yv1 Flufxlw0Vzlwfkhg V|vwhpv 1 1 1 1 1 1 1 :
41515 Frkhuhqw yv1 Qrq0frkhuhqw V|vwhpv 1 1 1 1 1 1 1 1 1 1 1 1 ;
416 RIGP 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 ;
41614 Vlqjoh0Fduulhu yv1 Pxowl0Fduulhu V|vwhpv 1 1 1 1 1 1 1 1 1 44
5 Uhvhdufk Wrslfv 46
514 Fkdqqho Hvwlpdwlrq 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 46
51414 Slorw V|pero Dvvlvwhg Prgxodwlrq 1 1 1 1 1 1 1 1 1 1 1 1 48
51415 Suh0frpshqvdwlrq 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 4:
515 Pxowl0Fduulhu V|vwhp Ghvljq 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 4<
51514 Vxe0Fkdqqho Edqgzlgwk 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 53
51515 Suhdpeoh0Edvhg V|qfkurql}dwlrq 1 1 1 1 1 1 1 1 1 1 1 1 1 56
51516 Slorw Sdwwhuq Ghvljq 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 59
516 Vxshulpsrvhg Slorw Vhtxhqfhv 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 5<
51614 Wlph dqg Iuhtxhqf| V|qfkurql}dwlrq iru RIGP 1 1 1 1 63
51615 Fkdqqho Hvwlpdwlrq xvlqj Vxshulpsrvhg Slorw Vhtxhqfhv 64
6 Vxppdu| dqg Frqwulexwlrqv 68

614 Jhqhudo Uhvxowv 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 68
615 Sdshu Frqwulexwlrqv 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 69
{l
{ll FRQWHQWV
61514Slorw Dvvlvwhg Fkdqqho Hvwlpdwlrq iru RIGP lq Preloh

Fhooxodu V|vwhpv 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 69
61515 Rswlpl}dwlrq ri Vxe0Fkdqqho Edqgzlgwk iru Preloh RIGP
V|vwhpv 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 6:
61516 Suh0Frpshqvdwlrq iru Ud|ohljk Idglqj Fkdqqhov lq Wlph
Glylvlrq Gxsoh{ RIGP V|vwhpv 1 1 1 1 1 1 1 1 1 1 1 1 1 6;
61517 Suhdpeoh0Edvhg Wlph dqg Iuhtxhqf| V|qfkurql}dwlrq iru
RIGP V|vwhpv 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 6;
61518 RIGP Wlph dqg Iuhtxhqf| V|qfkurql}dwlrq e| Vsuhdg
Vshfwuxp Slorw Whfkqltxh 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 6<
61519 Fkdqqho Hvwlpdwlrq xvlqj Vxshulpsrvhg Slorw Vhtxhqfhv 73
6151: D Frpsdudwlyh Dqdo|vlv ehwzhhq Glhuhqw Ghwhfwlrq Vfkhphv
lq Iodw Ud|ohljk0idglqj Fkdqqhov 1 1 1 1 1 1 1 1 1 1 1 1 1 74
616 Frqfoxvlrqv dqg Glvfxvvlrq 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 74
Eleolrjudsk| 78
LL Lqfoxghg Sdshuv 86
Sdshu D/
Slorw dvvlvwhg fkdqqho hvwlpdwlrq iru RIGP lq preloh fhooxodu v|vwhpv1
Sdshu E/
Rswlpl}dwlrq ri vxe0fkdqqho edqgzlgwk iru preloh RIGP v|vwhpv1
Sdshu F/
Suh0frpshqvdwlrq iru Ud|ohljk idglqj fkdqqhov lq wlph glylvlrq gxsoh{ RIGP
v|vwhpv1
Sdshu G/
Suhdpeoh0edvhg wlph dqg iuhtxhqf| v|qfkurql}dwlrq iru RIGP v|vwhpv1
Sdshu H/
RIGP wlph dqg iuhtxhqf| v|qfkurql}dwlrq e| vsuhdg vshfwuxp slorw whfkqltxh1
Sdshu I/
Fkdqqho hvwlpdwlrq xvlqj vxshulpsrvhg slorw vhtxhqfhv1
Sdshu J/
D frpsdudwlyh dqdo|vlv ehwzhhq glhuhqw ghwhfwlrq vfkhphv lq dw Ud|ohljk0
idglqj fkdqqhov1
Sduw L
Ryhuylhz ri wkh Uhvhdufk

Duhd
4
Fkdswhu 4
Lqwurgxfwlrq
Udglr wudqvplvvlrq kdv doorzhg shrsoh wr frppxqlfdwh zlwkrxw dq| sk|vlfdo

frqqhfwlrq iru pruh wkdq kxqguhg |hduv1 Zkhq Pdufrql pdqdjhg wr ghprq0
vwudwh d whfkqltxh iru zluhohvv whohjudsk| pruh wkdq d fhqwxu| djr/ lw zdv d
pdmru euhdnwkurxjk dqg wkh vwduw ri d frpsohwho| qhz lqgxvwu|1 Pd|eh rqh
frxog qrw fdoo lw d preloh zluhohvv v|vwhp/ exw wkhuh zdv qr zluh$ Wrgd|/ wkh
surjuhvv lq vhplfrqgxfwru whfkqrorj| kdv pdgh lw srvvleoh/ qrw wr irujhw di0
irugdeoh/ iru ploolrqv ri shrsoh wr vlpxowdqhrxvo| frppxqlfdwh rq wkh pryh
doo ryhu wkh zruog1 Wkh uvw dxwrpdwlf preloh whohskrq| v|vwhp lq Vzhghq
zdv lqwurgxfhg lq 4<88 ^47`/ rhulqj yrlfh vhuylfhv lq wkuhh pdmru flwlhv1 Krz0
hyhu/ lw zdv rqo| diwhu wkh lqwurgxfwlrq ri wkh Qruglf Preloh Whohskrqh v|vwhp
+QPW, lq 4<;4 wkdw wkh qxpehu ri xvhuv lq Vzhghq h{fhhghg 53 3331 Wrgd|/
wkh qxpehu ri xvhuv lv lqfuhdvlqj gudpdwlfdoo| dqg wkhuh lv d pryhphqw wrzdugv
pruh gdwd frppxqlfdwlrq dqg pxowlphgld vhuylfhv1 Wkhvh vhuylfhv uhtxluh qhz
whfkqrorj| wr kdqgoh wkh gdwd udwhv dqg wkh dprxqw ri wud!f h{shfwhg1
Wkh preloh frppxqlfdwlrq v|vwhpv duh riwhq fdwhjrul}hg dv glhuhqw jhq0
hudwlrqv ghshqglqj rq wkh vhuylfhv rhuhg1 Wkh uvw jhqhudwlrq frpsulvhv wkh
dqdorj iuhtxhqf|0glylvlrq pxowlsoh{ dffhvv +IGPD, v|vwhpv vxfk dv wkh QPW
dqg DPSV +Dgydqfhg Preloh Skrqh Vhuylfhv,1 Wkh vhfrqg jhqhudwlrq frqvlvwv
ri wkh uvw preloh gljlwdo frppxqlfdwlrq v|vwhpv vxfk dv wkh wlph0glylvlrq
pxowlsoh dffhvv +WGPD, edvhg JVP +Joredo V|vwhp iru Preloh frppxqlfd0
wlrq,/ G0DPSV +Gljlwdo DPSV,/ SGF +Sdflf Gljlwdo Fhooxodu, dqg wkh frgh0
glylvlrq pxowlsoh dffhvv +FGPD, edvhg v|vwhp LV0<81 Wkhvh v|vwhpv rhu
pdlqo| vshhfk frppxqlfdwlrq/ exw dovr gdwd frppxqlfdwlrq olplwhg wr udwkhu
orz wudqvplvvlrq udwhv1 Wkh wklug jhqhudwlrq lv fxuuhqwo| xqghu ghyhorsphqw
dqg wkh uvw surwrw|sh v|vwhpv kdyh mxvw ehhq gholyhuhg1 Wklv jurxs lv fkdu0
dfwhul}hg e| hqkdqfhg gdwd frppxqlfdwlrq fdsdelolwlhv surylglqj gdwd udwhv ri
dw ohdvw 6;7 nelw2v ryhu d zlgh duhd dqg frqwdlqv wkh Zlghedqg0FGPD v|vwhp
+Z0FGPD, dqg wkh FGPD05333 v|vwhp1
Lq sdudooho wr wkh ghyhorsphqw ri wkh wklug jhqhudwlrq v|vwhpv/ wkhuh kdv
6
7 FKDSWHU 41 LQWURGXFWLRQ
ehhq dq lqfuhdvlqj lqwhuhvw lq kljk0vshhg zluhohvv orfdo duhd qhwzrunv +Z0

ODQv,1 Wkhvh v|vwhpv duh wkh irfxv ri wklv wkhvlv/ wkrxjk wkh whfkqltxhv
suhvhqwhg fdq eh dssolhg wr rwkhu v|vwhpv dv zhoo1 Wkh Z0ODQ v|vwhpv gr
qrw rhu wkh vdph zlgh duhd fryhudjh dv wkh wklug jhqhudwlrq preloh v|vwhpv
gr/ exw zlwklq wkhlu olplwhg fryhudjh duhd wkh| surylgh pxfk kljkhu gdwd udwhv1
Wkhuhiruh/ wkh Z0ODQ v|vwhpv fdq eh vhhq dv d jrrg frpsohphqw wr wkh preloh
v|vwhpv1 Zlwklq wkh fryhudjh duhd/ iru h{dpsoh wkh r!fh/ wkh zluhohvv v|vwhpv
rhu kljk elw udwhv zlwk shuirupdqfh forvh wr d zluhg frqqhfwlrq1 Rxwvlgh wkh
exloglqj/ wkh zlgh0duhd v|vwhpv rhu d vlplodu vhuylfh dw d orzhu gdwd udwh1
Krzhyhu/ wkhuh duh ri frxuvh dssolfdwlrqv rqo| vxlwdeoh iru wkh kljk gdwd udwhv1
Wkh Z0ODQ v|vwhpv duh/ iru h{dpsoh/ uhsuhvhqwhg e| wkh LHHH ;35144d ^6` dqg
LHHH ;35144e ^4` v|vwhpv lq wkh X1V1/ wkh Eurdgedqg Udglr Dffhvv Qhwzrun
+EUDQ, KlshuODQ25 v|vwhp lq Hxursh ^7` dqg wkh PPDF v|vwhp lq Mdsdq1
Wkhuh lv dovr dqrwkhu jurxs ri udglr v|vwhpv wkdw kdv dfklhyhg odujh lq0
whuhvw gxulqj wkh odvw |hduv1 Wkhvh duh wkh vr0fdoohg dg0krf qhwzrunv/ zklfk
iru h{dpsoh lqfoxgh wkh Eoxhwrrwk v|vwhp ^58`1 Lq wkhvh v|vwhpv/ wkhuh lv qr
suh0ghwhuplqhg dffhvv srlqw/ exw wkh whuplqdov hvwdeolvk frqqhfwlrqv ehwzhhq
hdfk rwkhu dqg hyhu| whuplqdo fdq kdyh wkh uroh dv d pdvwhu ru edvh vwdwlrq1
Wklv fkdqjhv wkh frqglwlrqv wr hvwdeolvk dqg pdlqwdlq wkh frqqhfwlrq/ vlqfh
wkhuh lv qr ghglfdwhg edvh vwdwlrq wr wdon wr1 Qhyhuwkhohvv/ wkh sureohpv ri
v|qfkurql}dwlrq dqg srvvleo| fkdqqho hvwlpdwlrq duh vwloo wkhuh1 Dovr lq wkhvh
v|vwhpv wkh uhfhlyhu qhhgv wr qg wkh vwduw ri d sdfnhw/ wkh iuhtxhqf| rvhw/
dqg vr rq/ wr eh deoh wr ghprgxodwh wkh gdwd1
414 Gljlwdo Frppxqlfdwlrq V|vwhpv

D gljlwdo frppxqlfdwlrq v|vwhp lv riwhq glylghg lqwr vhyhudo ixqfwlrqdo xqlwv
dv vkrzq lq Iljxuh 414 ^89`1 Wkh wdvn ri wkh vrxufh hqfrghu lv wr uhsuhvhqw wkh
gljlwdo ru dqdorj lqirupdwlrq e| elwv lq dq h!flhqw zd|1 Wkh elwv duh wkhq ihg
wr wkh fkdqqho hqfrghu/ zklfk dggv elwv lq d vwuxfwxuhg zd| wr hqdeoh ghwhfwlrq
dqg fruuhfwlrq ri wudqvplvvlrq huuruv1 Wkh elwv iurp wkh hqfrghu duh jurxshg
dqg wudqviruphg wr fhuwdlq v|perov/ ru zdyhirupv/ e| wkh prgxodwru/ dqg wkh
zdyhirupv duh pl{hg zlwk d fduulhu wr jhw d vljqdo vxlwdeoh wr eh wudqvplwwhg
wkurxjk wkh fkdqqho1 Dw wkh uhfhlyhu wkh uhyhuvh ixqfwlrqv wdnh sodfh1 Wkh
uhfhlyhg v|perov duh ghprgxodwhg dqg vriw ru kdug ydoxhv ri wkh fruuhvsrqglqj
elwv duh sdvvhg wr wkh ghfrghu1 Wkh ghfrghu dqdo|}hv wkh vwuxfwxuh ri wkh
uhfhlyhg elw sdwwhuq dqg wulhv wr fruuhfw ru ghwhfw huuruv1 Ilqdoo|/ wkh fruuhfwhg
elwv duh ihg wr wkh vrxufh ghfrghu wkdw lv xvhg wr uhfrqvwuxfw wkh dqdorj vshhfk
vljqdo ru gljlwdo gdwd lqsxw1
Wklv wkhvlv ghdov zlwk wkh wkuhh eorfnv wr wkh uljkw lq Iljxuh 414= wkh prg0
xodwru/ wkh fkdqqho/ dqg wkh ghprgxodwru1 Wkh pdlq txhvwlrq lv krz wr ghvljq
fhuwdlq sduwv ri wkh prgxodwru dqg ghprgxodwru wr dfklhyh h!flhqw dqg urexvw
wudqvplvvlrq wkurxjk d preloh zluhohvv fkdqqho1 D zluhohvv fkdqqho kdv vrph
4141 GLJLWDO FRPPXQLFDWLRQ V\VWHPV 8
source channel
source modulator
encoder encoder
channel
source channel de-

destination decoder decoder modulator
Iljxuh 414= Ixqfwlrqdo eorfnv lq d frppxqlfdwlrq v|vwhp1
surshuwlhv wkdw pdnh wkh ghvljq hvshfldoo| fkdoohqjlqj= lw lqwurgxfhv wlph ydu|0
lqj hfkrhv dqg skdvh vkliwv dv zhoo dv d wlph ydu|lqj dwwhqxdwlrq ri wkh dpsol0
wxgh1 Wkh wkhvlv irfxvhv rq wkh iroorzlqj sduwv lq wkh prgxodwru0ghprgxodwru
fkdlq=
RIGP v|vwhp ghvljq1 Ruwkrjrqdo iuhtxhqf| glylvlrq pxowlsoh{ +RIGP,

kdv suryhq wr eh d prgxodwlrq whfkqltxh zhoo vxlwhg iru kljk elw udwhv
rq wlph glvshuvlyh fkdqqhov ^85`1 Wkhuh duh/ krzhyhu/ vrph vshflf uh0
txluhphqwv zkhq ghvljqlqj zluhohvv RIGP v|vwhpv/ iru h{dpsoh krz wr
fkrrvh wkh edqgzlgwk ri wkh vxe0fkdqqhov xvhg iru wudqvplvvlrq dqg krz
wr dfklhyh uholdeoh v|qfkurql}dwlrq1 Wkh odwwhu lv hvshfldoo| lpsruwdqw lq
sdfnhw0edvhg v|vwhpv vlqfh v|qfkurql}dwlrq kdv wr eh dfklhyhg zlwklq d
ihz v|perov1
Fkdqqho hvwlpdwlrq1 Lq rughu wr dfklhyh jrrg shuirupdqfh wkh uhfhlyhu

kdv wr nqrz wkh lpsdfw ri wkh fkdqqho1 Wkh sureohp lv krz wr h{wudfw
wklv lqirupdwlrq lq dq h!flhqw zd|1 Frqyhqwlrqdoo|/ nqrzq elwv duh pxo0
wlsoh{hg lqwr wkh gdwd vhtxhqfh lq rughu wr hvwlpdwh wkh fkdqqho1 Lq
RIGP v|vwhpv/ wkhuh duh vrph vshflf rssruwxqlwlhv iru fkdqqho hvwl0
pdwlrq vlqfh vhyhudo dgmdfhqw fduulhuv duh xvhg iru wkh wudqvplvvlrq dv
rssrvhg wr vlqjoh0fduulhu v|vwhpv1
V|qfkurql}dwlrq1 Wkh uhfhlyhu kdv wr ghwhuplqh zkhq wkhuh lv d vljqdo wr

ghprgxodwh dqg zkhuh wkh sdfnhwv dqg v|perov vwduw lq rughu wr lqwhu0
suhw wkh uhfhlyhg vljqdo1 Zkhq wkhuh lv qr wudqvplvvlrq/ lw lv lpsruwdqw
wkdw wkh uhfhlyhu lv deoh wr hqwhu vrph nlqg ri vohhs prgh wr vdyh srzhu1
Krzhyhu/ dv vrrq dv wudqvplvvlrq vwduwv wkh uhfhlyhu kdv wr dfklhyh v|q0
fkurql}dwlrq lq d yhu| vkruw wlph lq wr suhyhqw wkh orvv ri gdwd1
Dv phqwlrqhg ehiruh/ wkh wudqvplwwhg lqirupdwlrq lv uhsuhvhqwhg e| glhuhqw

zdyhirupv wkdw prgxodwh d fduulhu ri d fhuwdlq iuhtxhqf|1 Wkh wudqvplwwhg
vljqdo fdq eh uhsuhvhqwhg dv ^89`
v+w, @ Uhi{+w,hm 5if w j> +414,
zkhuh Uhij ghqrwhv wkh uhdo ydoxhg sduw/ if lv wkh fduulhu iuhtxhqf| dqg {+w, lv
wkh orz0sdvv htxlydohqw/ wkh vr fdoohg frpsoh{ hqyhorsh/ ri wkh wudqvplwwhg vlj0
qdo1 Lq gljlwdo frppxqlfdwlrq v|vwhpv wkh frpsoh{ hqyhorsh lv riwhq ghvfulehg
e| d gdwd sduw dqg d sxovh vkdsh1 Iru sxovh dpsolwxgh prgxodwhg +SDP,
vfkhphv ru skdvh prgxodwhg/ skdvh vkliw nh|lqj +SVN,/ vfkhphv wkh frpsoh{
hqyhorsh fdq eh ghvfulehg e|
[
{+w, @ fn z+w nWv ,> +415,
n
zkhuh z+w, lv wkh sxovh vkdsh/ fn lv wkh frpsoh{0ydoxhg gdwd sduw dqg Wv lv wkh
v|pero wlph1
Rqh ri wkh fkdoohqjlqj sureohpv lq wkh ghvljq ri gljlwdo frppxqlfdwlrq v|v0
whpv lv krz wr fkrrvh wkh prgxodwlrq vfkhph lq rughu wr jhw urexvw dqg h!flhqw
v|vwhpv1 Wklv fkrlfh lv juhdwo| dhfwhg e| wkh hqylurqphqw lq zklfk wkh v|vwhp
lv vxssrvhg wr zrun1 Wkh lq xhqfh ri wkh fkdqqho fdq eh ghvfulehg e| lwv lp0
sxovh uhvsrqvh dqg riwhq/ wkhuh lv dovr dgglwlyh zklwh Jdxvvldq qrlvh +DZJQ,
uhsuhvhqwlqj glhuhqw glvwxuedqfhv lq wkh v|vwhp1 Iru preloh ru zluhohvv ds0
solfdwlrqv/ wkh fkdqqho lv riwhq ghvfulehg dv d vhw ri lqghshqghqw pxowlsdwk
frpsrqhqwv1 Wkh wlph ydu|lqj lpsxovh uhvsrqvh fdq eh ghvfulehg e| ^8;`
L 4
[
k+w> , @ dl +w,+ l +w,,> +416,
l@3
zkhuh dl +w, ghqrwhv wkh frpsoh{0ydoxhg wds jdlq iru sdwk qxpehu l/ L lv wkh
qxpehu ri wdsv/ l +w, lv wkh ghod| ri wds l/ dqg lv wkh Gludf ghowd ixqfwlrq1
Dprqj wkh prvw lpsruwdqw sdudphwhuv zkhq fkrrvlqj wkh prgxodwlrq vfkhph
duh wkh ghod| dqg wkh h{shfwhg uhfhlyhg srzhu iru glhuhqw ghod|v1 Odujh ghod|v
iru vwurqj sdwkv phdq wkdw wkh lqwhuihuhqfh ehwzhhq wkh glhuhqw uhfhlyhg vljqdo
sduwv fdq eh vhyhuh/ hvshfldoo| zkhq wkh v|pero udwh lv kljk vr wkdw wkh ghod|
h{fhhgv vhyhudo v|perov1 Lq wkdw fdvh rqh kdyh wr lqwurgxfh dq htxdol}hu wr
plwljdwh wkh hhfw ri lqwhuv|pero lqwhuihuhqfh +LVL,1 Dqrwkhu dowhuqdwlyh lv wr
xvh pdq| sdudooho fkdqqhov vr wkdw wkh v|pero wlph rq hdfk ri wkh fkdqqhov lv
orqj1 Wklv phdqv wkdw rqo| d vpdoo sduw ri wkh v|pero lv dhfwhg e| LVL dqg
wklv lv wkh lghd ehklqg ruwkrjrqdo iuhtxhqf| glylvlrq pxowlsoh{/ RIGP1
415 Zluhohvv V|vwhpv

Zluhohvv v|vwhpv duh rshudwlqj lq dq hqylurqphqw zklfk kdv vrph vshflf surs0
huwlhv frpsduhg wr {hg zluholqh v|vwhpv dqg wkhvh fdoo iru vshfldo ghvljq frq0
4151 ZLUHOHVV V\VWHPV :
vlghudwlrqv1 Lq d zluhg qhwzrun/ wkhuh duh qr idvw pryhphqwv ri whuplqdov ru

uh hfwlrq srlqwv dqg wkh fkdqqho sdudphwhuv duh fkdqjlqj yhu| vorzo|1 Lq dg0
glwlrq/ wlph glvshuvlrq lv ohvv vhyhuh lq d zluhg v|vwhp/ wkrxjk lw pljkw vwloo
eh d kdug sureohp gxh wr kljk gdwd udwhv1 Lq d preloh v|vwhp wkh whuplqdov
duh prylqj durxqg> wkh uhfhlyhg vljqdo vwuhqjwk/ dv zhoo dv wkh skdvh ri wkh
uhfhlyhg vljqdo/ duh fkdqjlqj udslgo|1 Ixuwkhu/ wkh vljqdo wudqvplwwhg ryhu wkh
udglr fkdqqho lv uh hfwhg e| krxvhv/ fduv hwf1/ ohdglqj wr glhuhqw sursdjdwlrq
sdwkv wr wkh uhfhlyhu1 Li wkh ohqjwk ri wkh sdwkv glhu/ wkh uhfhlyhg vljqdo zloo
frqwdlq vhyhudo ghod|hg yhuvlrqv ri wkh wudqvplwwhg vljqdo dffruglqj wr wkh fkdq0
qho lpsxovh uhvsrqvh +416,1 Dv phqwlrqhg ehiruh/ wkh ghod|v pdnh lw qhfhvvdu|
wr xvh frpsoh{ uhfhlyhu vwuxfwxuhv1 Lq d preloh zluhohvv v|vwhp/ wkh whuplqdov
duh ri frxuvh lqwhqghg wr eh sruwdeoh1 Wklv phdqv wkdw srzhu frqvxpswlrq
lv lpsruwdqw vlqfh edwwhulhv vrphwlphv zloo srzhu wkh htxlsphqw1 Wkhuhiruh/
orz frpsoh{lw| dqg orz srzhu frqvxpswlrq duh surshuwlhv wkdw duh hyhq pruh
ghvludeoh lq d zluhohvv v|vwhp wkdq lq d zluhg v|vwhp1
41514 Sdfnhw0Edvhg yv1 Flufxlw0Vzlwfkhg V|vwhpv

D zluhohvv v|vwhp fdq hlwkhu eh sdfnhw0edvhg ru flufxlw0vzlwfkhg1 Lq d sdfnhw0
edvhg v|vwhp wkh lqirupdwlrq elwv duh jurxshg dqg wudqvplwwhg lq sdfnhwv/ dqg
wudqvplvvlrq rffxuv rqo| zkhq wkhuh lv d qhhg iru frppxqlfdwlrq1 Wkhuhiruh/
lq d sxuh sdfnhw0edvhg v|vwhp/ qr fdsdflw| lv doorfdwhg lq dgydqfh dqg wkhvh
v|vwhpv duh vxlwdeoh iru exuvw| wud!f frqglwlrqv/ vxfk dv gdwd frppxqlfdwlrq1
Lq flufxlw0vzlwfkhg v|vwhpv/ d sk|vlfdo ru yluwxdo frqqhfwlrq lv hvwdeolvkhg dqg
rffxslhg dv orqj dv frppxqlfdwlrq surfhhgv1 Wklv phdqv wkdw fdsdflw| lv
uhvhuyhg lq dgydqfh/ dqg wkdw qr rwkhu xvhu fdq xvh wklv fdsdflw| dv orqj dv wkh
frqqhfwlrq lv hvwdeolvkhg1 Flufxlw0vzlwfkhg v|vwhpv duh ehvw vxlwhg iru uhdo0wlph
vhuylfhv zkhq ghod| lv d olplwlqj idfwru1
Lq sdfnhw0edvhg v|vwhpv/ wkh uhfhlyhu kdv wr dfklhyh v|qfkurql}dwlrq lq d
yhu| vkruw wlph1 Lw lv kdug wr wudfn fkdqqho yduldwlrqv ehwzhhq wkh sdfnhwv/
dqg wkhuhiruh idvw dftxlvlwlrq dojrulwkpv duh uhtxluhg1 Diwhu dftxlvlwlrq/ lqlwldo
hvwlpdwhv ri wkh v|qfkurql}dwlrq sdudphwhuv duh dydlodeoh dqg wkh hvwlpdwru
hqwhuv wudfnlqj prgh1 Lq wklv prgh/ wkh wdvn lv wr iroorz fkdqjhv dqg frqwlq0
xrxvo| xsgdwh wkh fkdqqho hvwlpdwhv dqg v|qfkurql}dwlrq sdudphwhuv1 Vlqfh
wkh fkdqjhv lq jhqhudo duh uhodwlyho| vorz wkh uhtxluhphqwv rq wkh hvwlpdwlrq
wlph fdq eh vrphzkdw orzhu dqg wkhuhiruh rwkhu dojrulwkpv fdq eh xvhg gxu0
lqj wudfnlqj ^5:`1 Lq d flufxlw0vzlwfkhg v|vwhp/ wkh uhfhlyhu qhhgv wr hqwhu
dftxlvlwlrq prgh pruh vhogrp gxh wr wkh frqwlqxrxv wudqvplvvlrq1 Wkhuhiruh
wkh uhtxluhphqwv rq idvw dftxlvlwlrq fdq eh orrvhqhg lq wkhvh v|vwhpv1 Wrgd|
wkhuh lv d wuhqg wrzdugv pruh dqg pruh sdfnhw0edvhg v|vwhpv gxh wr wkh lq0
fuhdvhg gdwd wud!f1 Iru h{dpsoh/ erwk wkh wklug0jhqhudwlrq preloh v|vwhpv
edvhg rq Z0FGPD ^68` dqg wkh KlshuODQ25 v|vwhp ^7` edvhg rq RIGP xvh
sdfnhw0edvhg frppxqlfdwlrq iru gdwd wud!f1
; FKDSWHU 41 LQWURGXFWLRQ
41515 Frkhuhqw yv1 Qrq0frkhuhqw V|vwhpv

Lq jhqhudo frkhuhqw vfkhphv uhvxow lq ehwwhu ghwhfwlrq shuirupdqfh frpsduhg
wr glhuhqwldo v|vwhpv/ exw wkh| uhtxluh fkdqqho hvwlpdwlrq lq rughu wr irup d
uhihuhqfh iru wkh ghflvlrqv1 Glhuhqwldo vfkhphv/ rq wkh rwkhu kdqg/ uhtxluh qr
fkdqqho hvwlpdwlrq/ exw wkhuh lv d shuirupdqfh orvv frpsduhg wr frkhuhqw gh0
whfwlrq1 Glhuhqwldo vfkhphv dovr pdnh lw kdug wr dsso| pxowlohyho prgxodwlrq1
Lq frkhuhqw vfkhphv wkh fkdqqho hvwlpdwhv duh riwhq dfklhyhg e| pxowlsoh{0
lqj nqrzq/ vr fdoohg/ slorw v|perov lqwr wkh gdwd vhtxhqfh1 Wkh whfkqltxh lv
fdoohg Slorw V|pero Dvvlvwhg Prgxodwlrq +SVDP,1 Lw zdv ruljlqdoo| sursrvhg
e| Prkhu dqg Orgjh lq ^7<` dqg ixuwkhu dqdo|}hg e| Fdyhuv lq ^48`1 Wkh
uhfhlyhu revhuyhv wkh lq xhqfh ri wkh fkdqqho rq wkh slorw v|perov dqg xvhv
lqwhusrodwlrq wr jhw dq hvwlpdwh ri wkh fkdqqho lpsdfw rq wkh gdwd v|perov1
Wkh uhfhlyhu wkhq uhpryhv wklv lpsdfw lq rughu wr pdnh ghflvlrqv1 Wkh slorw
v|perov wudqvplw qr gdwd dqg wkhuhiruh wkhuh lv d vpdoo ryhukhdg fdxvlqj d
edqgzlgwk h{sdqvlrq dqg dq hqhuj| orvv1 Erwk wkhvh orvvhv duh ghshqghqw rq
wkh slorw0wr0gdwd v|pero udwlr1 Lq pxowl0fduulhu v|vwhpv lqwhusrodwlrq lq erwk
wlph dqg iuhtxhqf| fdq eh xvhg dqg wkhuhiruh wkh slorw0wr0gdwd v|pero udwlr
fdq eh ghfuhdvhg frpsduhg wr vlqjoh0fduulhu v|vwhpv ^D`1
Lq frqyhqwlrqdo glhuhqwldo v|vwhpv/ wkh odvw uhfhlyhg v|pero lv xvhg dv d
uhihuhqfh zkhq pdnlqj wkh ghflvlrqv1 Vlqfh wkhuh lv qr devroxwh skdvh uhihuhqfh/
glhuhqwldo hqfrglqj lv uhtxluhg1 Lq wkhvh v|vwhpv/ wkhuh lv qr qhhg iru fkdqqho
hvwlpdwlrq dqg wkhuhiruh wkh frpsoh{lw| ri wkh uhfhlyhu fdq eh ghfuhdvhg vljqli0
lfdqwo| frpsduhg wr frkhuhqw vfkhphv1 Wkhuh lv/ krzhyhu/ d shuirupdqfh orvv
frpsduhg wr frkhuhqw v|vwhpv vlqfh/ lq frqyhqwlrqdo glhuhqwldo v|vwhpv/ rqo|
rqh v|pero lv xvhg wr irup wkh uhihuhqfh1 Wklv phdqv wkdw wkh uhihuhqfh v|pero
dqg wkh uhfhlyhg gdwd v|pero duh glvwxuehg e| qrlvh vdpsohv erwk kdylqj wkh
vdph yduldqfh1 Rqh zd| wr jhw durxqg wklv sureohp lv wr xvh pxowlsoh0v|pero
glhuhqwldo ghwhfwlrq/ lqwurgxfhg e| Glyvdodu dqg Vlprq lq ^54`1 Lq wklv vfkhph/
wkh uhfhlyhu xvhv pruh wkdq wkh odvw uhfhlyhg v|pero wr irup d uhihuhqfh dqg lv
wkhuhiruh ohvv dhfwhg e| qrlvh1 Qdwxudoo|/ wklv vfkhph lv pruh frpsoh{ wkdq
frqyhqwlrqdo glhuhqwldo ghwhfwlrq1
416 RIGP
Wkh uvw RIGP vfkhph zlwk ryhuodsslqj vshfwud zdv lqwurgxfhg e| Fkdqj lq
4<99 ^4:`1 Dv rssrvhg wr wrgd|*v v|vwhpv/ d edqn ri vlqxvrlgdo jhqhudwruv zdv
xvhg wr surgxfh wkh ruwkrjrqdo fkdqqhov1 Lq 4<:4 Zhlqvwhlq dqg Hehuw ^;9`
sursrvhg wr xvh wkh Glvfuhwh Irxulhu Wudqvirup +GIW, iru wklv sxusrvh dqg
wkh uvw prghuq RIGP v|vwhp zdv eruq1 Wkhuh h{lvw pdq| ryhuylhzv dqg
ghwdlohg suhvhqwdwlrqv ri RIGP v|vwhpv lq wkh olwhudwxuh/ iru h{dpsoh ^:` ^45`
^55` ^59` ^85`1 Khuh L zloo uhylhz wkh prvw ixqgdphqwdo surshuwlhv dqg frqfhswv
qhhghg iru wkh uhvw ri wkh wkhvlv1
Lq RIGP v|vwhpv pdq| vxe0fkdqqhov duh xvhg lq sdudooho1 Wkh fkdqqhov duh
4161 RIGP <
ryhuodsslqj lq iuhtxhqf|/ exw wkh glvwdqfh ehwzhhq wkhp lv fkrvhq vr wkdw wkh
glhuhqw fkdqqhov dq|krz duh ruwkrjrqdo1 Lq lwv prvw vlpsoh irup wkh vdpsoh
dw wlph n ri wkh frpsoh{ hqyhorsh lv jlyhq e| ^:`
u Q 4
Hv [ n
{n @ fq hm 5q Q > +417,
Q q@3
zkhuh Q lv wkh vl}h ri wkh GIW dqg Hv lv wkh v|pero hqhuj|1 Wkh vl}h ri wkh
GIW lv htxdo wr wkh qxpehu ri vxe0fkdqqhov dydlodeoh iru wudqvplvvlrq/ exw doo
ri wkh fkdqqhov qhhg qrw eh dfwlyh1 Wkh vxe0fkdqqho edqgzlgwk lv jlyhq e|
4 ivdps
ivf @ @ > +418,
Wv Q
zkhuh ivdps lv wkh vdpsoh udwh dqg Wv lv wkh v|pero wlph1 Qrupdoo|/ d f|folf
h{whqvlrq ri wkh v|pero ^87` lv xvhg wr plwljdwh wkh hhfw ri LVL1 Wklv/ vr fdoohg/
f|folf suh{ lv d frs| ri wkh odvw J vdpsohv dqg wkhvh duh wudqvplwwhg sulru
wr wkh dfwxdo v|pero1 Dowhuqdwlyho|/ d suh{ ri ohqjwk J4 dqg d srvw{ ri
ohqjwk J5 fdq eh xvhg1 Wkh pdlq lghd ehklqg RIGP lv wr xvh orqj v|perov/
vxfk wkdw wkh sduw dhfwhg e| LVL +wkh suh{, fdq eh glvuhjdughg zlwk rqo| d
plqru hqhuj| orvv1 Wkh vljqdo lv riwhq dovr zlqgrzhg lq rughu wr lpsuryh wkh
shuirupdqfh1 Wklv lv hvshfldoo| lpsruwdqw iru uhgxflqj rxw0ri0edqg hplvvlrqv
dqg wr ghfuhdvh wkh lqwhufkdqqho lqwhuihuhqfh/ LFL/ gxh wr iuhtxhqf| rvhwv ^:`1
Glhuhqw dssurdfkhv iru wkh zlqgrzlqj ixqfwlrqv duh sursrvhg lq wkh olwhudwxuh1
Lq/ iru h{dpsoh/ wkh LHHH ;35144d v|vwhp ^6` d vprrwko| ghfd|lqj h{whqvlrq
ri wkh v|perov fdq eh xvhg/ zkloh lq ^:` d pxowlsolfdwlrq zlwk d udlvhg fr0
vlqh ixqfwlrq lq wlph lv sursrvhg1 Wkh frpsoh{ hqyhorsh lqfoxglqj erwk sxovh
vkdslqj2zlqgrzlqj dqg wkh suh2srvw{ fdq eh zulwwhq dv
u
[4 Q[ .J5 Q[
4 n
fn>o hm 5q Q z+w
Hv n
{+w, @ oWwrw ,> +419,
Q .J i
o@4 n@J4 q@q v
zkhuh fn>o ghqrwhv wkh gdwd v|pero rq wkh nwk vxe0fkdqqho gxulqj wkh owk
RIGP v|pero/ z+w, ghqrwhv wkh zlqgrzlqj ixqfwlrq/ dqg Wwrw @ +Q . J,@iv
ghqrwhv wkh wrwdo RIGP v|pero wlph lqfoxglqj wkh f|folf suh2srvw{1
Dq riwhq xvhg prgho iru zluhohvv v|vwhpv lv wkdw wkh wudqvplwwhg vljqdo lv
frqyroyhg zlwk wkh fkdqqho lpsxovh uhvsrqvh1 Wkhq/ qrlvh dqg lqwhuihuhqfh
duh dgghg wr wklv vljqdo1 Lq dq RIGP uhfhlyhu/ d GIW lv shuiruphg lq rughu
wr ghwhfw wkh gdwd v|perov rq hdfk ri wkh vxe0fkdqqhov1 Li zh dvvxph lghdo
v|qfkurql}dwlrq/ d f|folf h{whqvlrq h{fhhglqj wkh pd{lpxp h{fhvv ghod| +ghod|
ehwzhhq wkh uvw dqg odvw wds ri wkh fkdqqho, dqg qr iuhtxhqf| rvhw wkhq wkh
uhfhlyhg v|pero diwhu wkh GIW lv jlyhq e| ^:`
un @ Kn fn . qn +41:,
r0
H0 n0
r1
H1 n1
rN
HN nN
Iljxuh 415= Wkh gdwd v|perov duh pxowlsolhg e| wkh ydoxhv ri wkh wudqvihu
ixqfwlrq dw wkh vxe0fkdqqho iuhtxhqflhv dqg glvwxuehg e| qrlvh1
iru d vorzo| ydu|lqj wlph glvshuvlyh fkdqqho zlwk dgglwlyh Jdxvvldq qrlvh1 Khuh
qn ghqrwhv zklwh Jdxvvldq qrlvh dqg Kn ghqrwhv wkh frpsoh{ ydoxh ri wkh wudqv0
ihu ixqfwlrq dw vxe0fkdqqho n1 Lq wkh lghdo fdvh/ wkh v|pero lv rqo| glvwxuehg e|
qrlvh dqg skdvh vkliwhg dqg dwwhqxdwhg e| wkh iuhtxhqf| uhvsrqvh/ dffruglqj
wr Iljxuh 415/ diwhu ghprgxodwlrq dqg wudqvplvvlrq ryhu d pxowlsdwk fkdqqho1
Wklv lv lq frqwudvw wr vlqjoh0fduulhu v|vwhpv/ zkhuh wkh uhfhlyhg vljqdo jhqhudoo|
lv fruuxswhg e| LVL1
D eorfn ghvfulswlrq ri d frqyhqwlrqdo RIGP prgxodwru dqg ghprgxodwru
lv suhvhqwhg lq Iljxuh 4161 Iluvw wkh frghg elwv duh ihg wr wkh prgxodwru zkhuh
jurxsv ri orj5 P elwv duh pdsshg wr P 0du| v|perov/ zkhuh P lv wkh vl}h ri
wkh v|pero doskdehw rq hdfk fduulhu1 Wkhq Q ri wkhvh v|perov duh jurxshg
lqwr rqh RIGP v|pero +qrwh/ krzhyhu/ wkdw vrph ri wkh rxwhu fkdqqhov pd|
eh }hur0ydoxhg,1 Hdfk gdwd v|pero lv dvvljqhg wr d vshflf vxe0fkdqqho e|
wkh LGIW/ wkh lqyhuvh glvfuhwh Irxulhu wudqvirup1 Qrupdoo| wklv odwwhu rshud0
wlrq lv shuiruphg dv dq lqyhuvh idvw Irxulhu wudqvirup +LIIW, gxh wr lwv orzhu
frpsoh{lw|1 Wkhq wkh f|folf h{whqvlrq lv dgghg dqg qdoo| sxovh vkdslqj lv
shuiruphg ehiruh wkh vljqdo lv xs0frqyhuwhg dqg wudqvplwwhg ryhu wkh fkdqqho1
Dw wkh uhfhlyhu/ wkh uhfhlyhg vljqdo lv grzq0frqyhuwhg dqg vdpsohg1 Wkhq
wkh vdpsohv duh ihg lq sdudooho wr wkh fkdqqho hvwlpdwru/ wkh IIW dqg wkh
dxwrpdwlf jdlq frqwuro2frduvh v|qfkurql}hu/ uhvshfwlyho|1 Wkh dxwrpdwlf jdlq
frqwuro +DJF, wudfnv wkh lqfrplqj srzhu dqg dgmxvwv wkh ohyho vr wkdw wkh
ixoo g|qdplf udqjh ri wkh dqdorj0wr0gljlwdo frqyhuwhu lv xvhg1 Wkh wdvn ri wkh
v|qfkurql}hu/ ru pruh suhflvh wkh frduvh v|qfkurql}hu/ lv wr qg wkh vwduw ri hdfk
gdwd sdfnhw dqg hdfk RIGP v|pero/ dqg wr qg dq hvwlpdwh ri wkh iuhtxhqf|
rvhw vr wkdw frpshqvdwlrq fdq eh pdgh ehiruh wkh vdpsohv duh ihg wr wkh IIW
4161 RIGP 44
coded
bits c cyclic ext. x
modulation IFFT pulse shaping
channel
channel
estimation
soft
decisions demodu- r
decision FFT
lation
AGC/
synch.
Iljxuh 416= Eorfn ghvfulswlrq ri d frqyhqwlrqdo RIGP prgxodwru dqg ghprg0

xodwlrq fkdlq
eorfn1 Wkh RIGP ghprgxodwru lv yhu| vhqvlwlyh wr iuhtxhqf| rvhwv ^97` dqg
lq rughu wr dyrlg LFL iuhtxhqf| rvhw frpshqvdwlrq kdv wr eh pdgh ehiruh wkh
GIW1
Wkh fkdqqho hvwlpdwru lv xvhg wr hqdeoh frkhuhqw prgxodwlrq vfkhphv1 Wkh
wlph glvshuvlyh fkdqqho uhvxowv lq d skdvh vkliw dqg dq dpsolwxgh dwwhqxdwlrq
dffruglqj wr +41:, dqg wkhuhiruh wkh ydoxhv ri wkh wudqvihu ixqfwlrq dw wkh gli0
ihuhqw vxe0fkdqqho iuhtxhqflhv kdyh wr eh hvwlpdwhg lq rughu wr xvh frkhuhqw
ghwhfwlrq1 Lq dgglwlrq/ li wkhuh lv d vpdoo wlplqj rvhw/ wklv zloo uhvxow lq olq0
hduo| lqfuhdvlqj skdvh ehwzhhq wkh vxe0fkdqqhov1 Wkh fkdqqho hvwlpdwru fdq
hvwlpdwh dqg frpshqvdwh iru wkhvh skdvh vkliwv vlqfh wkh| fdq eh uhjdughg dv
d sduw ri wkh fkdqqho wudqvihu ixqfwlrq ^59`1
Wkh IIW eorfn shuirupv d GIW rq wkh uhfhlyhg RIGP v|pero lq rughu wr
uhwulhyh wkh gdwd v|perov1 Wkhvh v|perov duh wkhq sdvvhg wr wkh ghprgxodwlrq
vwdjh zkhuh wkh lq xhqfh ri wkh fkdqqho lv uhpryhg vr wkdw kdug ru vriw ghflvlrqv
fdq eh pdgh1 Wkhvh ghflvlrqv duh wkhq sdvvhg wr wkh fkdqqho ghfrghu dv lq
Iljxuh 4141
41614 Vlqjoh0Fduulhu yv1 Pxowl0Fduulhu V|vwhpv

Wkh pdlq glhuhqfh ehwzhhq d vlqjoh0fduulhu dqg d pxowl0fduulhu v|vwhp lv krz
wkh lq xhqfh ri wkh fkdqqho lv plwljdwhg1 Lq dq RIGP v|vwhp wkh uhfhlyhu kdv
wr shuirup dq IIW lq rughu wr ghprgxodwh wkh gdwd/ exw wkh htxdol}hu fdq rq
wkh rwkhu kdqg eh uhdoo| vlpsoh vlqfh d wlph glvshuvlyh fkdqqho rqo| uhvxowv lq
d pxowlsolfdwlrq e| wkh ydoxh ri wkh fkdqqho wudqvihu ixqfwlrq diwhu wkh IIW1
Lq d vlqjoh0fduulhu uhfhlyhu wkhuh lv qr IIW eorfn/ exw wkh htxdol}hu pxvw eh
ghvljqhg wr plwljdwh wkh LVL lqvwhdg1 Wkh frpsoh{lw| ri wkh htxdol}hu fdq eh
txlwh kljk li wkh LVL vsdqv ryhu vhyhudo v|perov/ exw li lw vsdqv ryhu rqo| d ihz
v|perov wkh vroxwlrq lv vwudljkwiruzdug1 Dq hvwlpdwh ri wkh frpsoh{lw| ri wkh
htxdol}hu lv irxqg e| frpsdulqj wkh pd{lpxp h{fhvv ghod| ri wkh fkdqqho wr
wkh v|pero wlph> li wkh LVL vsdqv ryhu vhyhudo v|perov/ wkhq RIGP lv wr suhihu
wr wkh xvh ri orqj +dqg wkhuhe| frpsoh{, htxdol}huv ^:`1
Dqrwkhu dgydqwdjh ri pxowl0fduulhu prgxodwlrq lv wkdw lw lv srvvleoh wr xvh
dgdswlyh prgxodwlrq rq d vxe0fkdqqho edvlv1 Wklv phdqv wkdw glhuhqw prg0
xodwlrq ohyhov fdq eh xvhg rq glhuhqw vxe0fkdqqhov1 Rq d edg vxe0fkdqqho
zkhuh wkh fkdqqho dwwhqxdwlrq lv kljk/ d orz0ohyho prgxodwlrq vfkhph/ vxfk dv
elqdu| SVN/ fdq eh xvhg dqg rq d jrrg vxe0fkdqqho d kljkhu rughu prgxodwlrq/
vxfk dv 490TDP/ fdq eh xvhg1 Wklv phdqv wkdw wkh fkdqqho lv pruh h!flhqwo|
xvhg dqg wkdw wkh elw huuru udwh +EHU,/ fdq eh ghfuhdvhg vljqlfdqwo| frp0
sduhg wr {hg prgxodwlrq vfkhphv ^59`1 Qrwh krzhyhu/ wkdw dgdswlyh vfkhphv
uhtxluh erwk h{wud kdugzduh dqg frqwuro vljqdolqj1
Wkh pdlq glvdgydqwdjhv ri xvlqj pxowl0fduulhu v|vwhpv/ ru RIGP v|vwhpv
lq sduwlfxodu/ duh wkh vhqvlwlylw| wr iuhtxhqf| rvhwv dqg wkh odujh g|qdplf
udqjh ri wkh vljqdo wr eh wudqvplwwhg ^:`1 Iuhtxhqf| rvhwv fdxvh erwk LFL dqg
dwwhqxdwlrq ri wkh ghvluhg vljqdo1 Wkh vhqvlwlylw| wr iuhtxhqf| rvhwv fdq eh
h{sodlqhg e| wkh odujh iuhtxhqf| vlgh orehv ri wkh vljqdo dqg wkh vhqvlwlylw|
fdq eh ghfuhdvhg e| xvlqj d zlqgrz ixqfwlrq zlwk d pruh dwwudfwlyh iuhtxhqf|
fkdudfwhulvwlfv frpsduhg wr wkh rqh uhvxowlqj iurp wkh uhfwdqjxodu wlph gr0
pdlq zlqgrz ^:`1 Wkh odujh g|qdplf udqjh fdq eh xqghuvwrrg e| orrnlqj dw
wkh RIGP vljqdo dv d vxp ri Q vlqxvrlgv1 Vrphwlphv wkh zdyhirupv dgg
frqvwuxfwlyho|/ zklfk phdqv wkdw d shdn lv surgxfhg1 Iru prvw ri wkh wlph/
krzhyhu/ wkh zdyhirupv kdyh glhuhqw vljqv vxfk wkdw wkh vxp dfklhyhv orz
dpsolwxgh1
Fkdswhu 5
Uhvhdufk Wrslfv
Lq wklv fkdswhu d ghwdlohg edfnjurxqg dqg vrph ri wkh pdlq sxeolfdwlrqv lq

hdfk ri wkh uhvhdufk duhdv duh suhvhqwhg1 Dovr/ wkh uhodwlrq ehwzhhq wkh sdshuv
suhvhqwhg lq wklv wkhvlv dqg rwkhu sdshuv lv glvfxvvhg1 Dv phqwlrqhg lq wkh
lqwurgxfwlrq wkh zrun kdv ehhq irfxvhg rq v|vwhp dvshfwv ri wkh prgxodwru0
ghprgxodwru fkdlq1 Lq sulqflsoh L kdyh ehhq dfwlyh lq irxu glhuhqw uhvhdufk
surmhfwv/ dqg wkhvh duh wkh duhdv wkdw L zloo suhvhqw lq pruh ghwdlo lq wklv
fkdswhu1 Wkh irxu surmhfwv duh= fkdqqho hvwlpdwlrq lq RIGP v|vwhpv/ zluhohvv
RIGP v|vwhp ghvljq/ suhdpeoh0edvhg v|qfkurql}dwlrq iru RIGP/ dqg fkdqqho
hvwlpdwlrq dqg v|qfkurql}dwlrq xvlqj vxshulpsrvhg slorw vhtxhqfhv1 Sdshuv D
dqg F ehorqj wr wkh uvw jurxs rq fkdqqho hvwlpdwlrq dqg sdshu G ehorqjv
wr suhdpeoh0edvhg v|qfkurql}dwlrq1 Sdshu E ehorqjv wr RIGP v|vwhp ghvljq/
exw ri frxuvh wkh sdshuv uhjduglqj fkdqqho hvwlpdwlrq/ suh0frpshqvdwlrq dqg
suhdpeoh0edvhg v|qfkurql}dwlrq fdq dovr eh dvvljqhg wr wklv jurxs1 Ilqdoo|/
sdshuv H/ I dqg J ehorqj wr wkh surmhfw rq vxshulpsrvhg slorw vhtxhqfhv1
Qh{w/ wkhuh lv d suhvhqwdwlrq ri fkdqqho hvwlpdwlrq lq jhqhudo vlqfh wklv lv d
ixqgdphqwdo wrslf iru wkh zkroh wkhvlv1 Diwhu wkdw/ ghwdlov ri wkh glhuhqw
uhvhdufk duhdv duh suhvhqwhg1
514 Fkdqqho Hvwlpdwlrq

Wkh wdvn ri wkh htxdol}hu lv wr frpshqvdwh iru wkh lq xhqfh ri wkh fkdqqho1
Wklv frpshqvdwlrq uhtxluhv/ krzhyhu/ wkdw dq hvwlpdwh ri wkh fkdqqho uhvsrqvh
lv dydlodeoh1 Wkhuh duh pdq| dowhuqdwlyhv lghqwli|lqj wkh fkdqqho uhvsrqvh dqg
ghshqglqj rq wkh fkdqqho vwdwlvwlfv dqg wkh prgxodwlrq vfkhph glhuhqw vr0
oxwlrqv duh suhihudeoh1 Riwhq wkh fkdqqho lpsxovh uhvsrqvh ru wkh iuhtxhqf|
uhvsrqvh lv ghulyhg iurp wudlqlqj vhtxhqfhv ru slorw v|perov/ exw lw lv dovr srv0
vleoh wr xvh qrq0slorw0dlghg dssurdfkhv/ vr fdoohg eolqg dojrulwkpv ^7;`1 Wkh
xvh ri wudlqlqj v|perov phdqv wkdw wkh frpsoh{lw| ri wkh uhfhlyhu fdq eh nhsw
udwkhu orz dqg/ lq fdvh wkdw wkh wudlqlqj v|perov dqg gdwd v|perov duh uh0
46
47 FKDSWHU 51 UHVHDUFK WRSLFV
fhlyhg zlwkrxw lq xhqfh iurp hdfk rwkhu/ wkh surfhvvhv ri fkdqqho hvwlpdwlrq
dqg ghwhfwlrq fdq eh vhsdudwhg1 Dovr/ wudlqlqj v|perov duh vxlwdeoh iru sdfnhw0
edvhg frppxqlfdwlrq vlqfh wkhvh uhvxow lq urexvw ehkdylru zlwk qr vljqlfdqw
wudqvlhqw hhfwv ^7;`1
Lq wkh jhqhudo fdvh wkh rswlpdo uhfhlyhu/ lq wkh vhqvh ri plqlpxp vh0
txhqfh huuru suredelolw| zkhq qr sulru fkdqqho hvwlpdwhv duh dydlodeoh/ lv wkh
pd{lpxp0olnholkrrg +PO, uhfhlyhu pdnlqj mrlqw ghwhfwlrq dqg fkdqqho hvwlpd0
wlrq ^7;`1 Wkh wdvn lv wr qg d fkdqqho hvwlpdwh dqg d gdwd vhtxhqfh wkdw pdnh
wkh dfwxdo uhfhlyhg vhtxhqfh dv suredeoh dv srvvleoh/ l1h1 wr qg
+a a
f> k,PO @ duj pd{ i+umf> k,
f>k
+514,
zkhuh u lv wkh uhfhlyhg vhtxhqfh/ f lv wkh gdwd vhtxhqfh/ k lv wkh vhtxhqfh ri wkh
fkdqqho uhvsrqvhv/ i+, ghqrwhv wkh suredelolw| ghqvlw| ixqfwlrq dqg duj pd{
ghqrwhv wkh dujxphqw pd{lpl}lqj wkh ixqfwlrq1 Wklv dssurdfk ohdgv/ krzhyhu/
wr frpsoh{ uhfhlyhu vwuxfwxuhv dqg wkhuhiruh rwkhu ohvv frpsoh{ dowhuqdwlyhv duh
ri lqwhuhvw1 Lq fdvh fkdqqho hvwlpdwlrq dqg gdwd ghwhfwlrq fdq eh shuiruphg
vhsdudwho|/ wkh vwuxfwxuh fdq eh vlpsolhg e| ghulylqj wkh fkdqqho hvwlpdwhv
ehiruh ghwhfwlrq wdnhv sodfh1 Jlyhq d fhuwdlq fkdqqho hvwlpdwh k/ a wkh rswlpdo
ghwhfwru/ lq wkh vhqvh ri plqlpdo vhtxhqfh huuru suredelolw|/ lv wkh Ylwhuel
ghwhfwru ^89`1 Wklv htxdol}hu jlyhv wkh pd{lpxp0olnholkrrg vhtxhqfh/ zklfk
phdqv wkdw lw pdnhv d ghflvlrq lq idyru ri wkh vhtxhqfh wkdw pd{lpl}hv wkh
suredelolw| ri uhfhlylqj wkh dfwxdo uhfhlyhg vhtxhqfh/ l1h1
f @ duj pd{ i +umf> k,=

a a +515,
f
Lq RIGP v|vwhpv wkhuh lv qr LVL diwhu ghprgxodwlrq +xqghu lghdo frqgl0

wlrqv, dqg/ dv qrw dozd|v wkh fdvh iru vlqjoh0fduulhu v|vwhpv/ fkdqqho hvwlpdwlrq
dqg ghwhfwlrq fdq eh shuiruphg vhsdudwho| lq d wlph glvshuvlyh fkdqqho1 Dq0
rwkhu dgydqwdjh lv wkdw d yhu| vlpsoh htxdol}hu vwuxfwxuh fdq eh xvhg/ vlqfh rqo|
d frpsoh{ glylvlrq e| wkh ydoxh ri wkh fkdqqho wudqvihu ixqfwlrq lv uhtxluhg1 D
qhfhvvlw| iru wklv lv wkdw wkh uhfhlyhu nqrzv/ ru dw ohdvw pdnhv hvwlpdwhv ri/ wkh
wudqvihu ixqfwlrqv dw doo wkh vxe0fkdqqho iuhtxhqflhv1
Rqh whfkqltxh vxlwdeoh lq RIGP v|vwhpv iru hvwlpdwlqj wkh ydoxhv ri wkh
wudqvihu ixqfwlrq lv wr xvh slorw v|pero dvvlvwhg prgxodwlrq +SVDP, rshudwlqj
lq wkh iuhtxhqf| grpdlq1 Rwkhu whfkqltxhv iru fkdqqho hvwlpdwlrq dqg fdofx0
odwlrq ri wkh wudqvihu ixqfwlrqv lq RIGP v|vwhpv lqfoxgh/ iru h{dpsoh/ wkh
xvh ri fruuhodwlrq0edvhg hvwlpdwruv zrunlqj lq wkh wlph grpdlq ^4;` dqg fkdq0
qho hvwlpdwlrq xvlqj vlqjxodu ydoxh ghfrpsrvlwlrq ^56` ^77`1 Wkh odwwhu fdq eh
edvhg rq slorw v|perov/ exw wr fdofxodwh wkh fkdqqho hvwlpdwhv wkh vwdwlvwlfdo
surshuwlhv ri wkh fkdqqho duh xvhg lq d glhuhqw zd| lq rughu wr ghfuhdvh wkh
frpsoh{lw|1
5141 FKDQQHO HVWLPDWLRQ 48
51414 Slorw V|pero Dvvlvwhg Prgxodwlrq

SVDP zdv/ dv phqwlrqhg hduolhu/ lqwurgxfhg e| Prkhu dqg Orgjh lq ^7<`/ iru
wkh xvh lq d wuhoolv frghg prgxodwhg +WFP, vfkhph1 Wkh fkdqqho hvwlpdwhv
zhuh ghulyhg xvlqj d orz0sdvv owhu/ dqg wkh hvwlpdwhv zhuh xvhg erwk iru gh0
whfwlrq ri wkh SVN prgxodwhg gdwd v|perov dqg dv hvwlpdwhv ri wkh uhfhlyhg
v|pero hqhuj| lq wkh ghfrghu1 Djkdprkdppdgl hw do1 lqghshqghqwo| sur0
srvhg d vlplodu slorw0edvhg vfkhph ^8` lqwhqghg iru xqfrghg SVN dqg TDP
vfkhphv1 Wkh| xvhg d Ndopdq vprrwkhu wr ghulyh wkh fkdqqho hvwlpdwhv dqg
wkh| dovr fdph wr wkh frqfoxvlrq wkdw slorw0edvhg whfkqltxhv ohdg wr urexvw
ghwhfwlrq zlwk d shuirupdqfh ehwwhu wkdq wkdw ri qrq0frkhuhqw vfkhphv1 Erwk
lq ^8` dqg ^7<` lw zdv frqfoxghg wkdw wkh pd{lpxp vsdflqj ehwzhhq wkh slorw
v|perov lv ghwhuplqhg e| wkh Grssohu iuhtxhqf| vlqfh wkh slorw whfkqltxh fdq
eh vhhq dv d vdpsolqj ri wkh orz0sdvv idglqj surfhvv1 Erwk wkhvh sdshuv uholhg
rq vlpxodwlrqv iru wkhlu dqdo|vhv/ exw Fdyhuv pdgh d wkrurxjk dqdo|wlfdo dqdo0
|vlv ri wkh whfkqltxh lq ^48`1 Khuh/ kh xvhg dq rswlpdo Zlhqhu owhu iru wkh
fkdqqho hvwlpdwhv dqg kh ghulyhg wkh uhvxowlqj elw huuru udwh erwk xqghu lghdo
dqg plvpdwfkhg frqglwlrqv1 Rqh ri wkh pdlq frqfoxvlrqv zdv wkdw SVDP rxw0
shuirupv glhuhqwldo ghwhfwlrq iru doo ydoxhv ri wkh vljqdo0wr0qrlvh udwlr +VQU,
dqg Grssohu iuhtxhqf|1 Dqrwkhu lpsruwdqw frqfoxvlrq zdv wkdw li wkh fkdqqho
hvwlpdwru lv ghvljqhg iru wkh zruvw0fdvh Grssohu iuhtxhqf|/ wkhq wkh shuiru0
pdqfh zloo lq jhqhudo qrw eh zruvh iru rwkhu orzhu Grssohu iuhtxhqflhv1 Wklv
phdqv wkdw d zruvw0fdvh ghvljq lv dssursuldwh iru wkh fkdqqho hvwlpdwlrq owhu1
Fdyhuv dovr vwxglhg wkh lpsdfw ri wkh slorw vsdflqj dqg vwdwhg wkdw lq rughu
wr ixooo wkh vdpsolqj wkhruhp wkh vsdflqj lq wlph ehwzhhq wkh slorw v|perov/
QW / vkrxog qrw h{fhhg ^48`
4
QW > +516,
5iG Wv
zkhuh iG lv wkh pd{lpxp Grssohu iuhtxhqf|1 Wklv phdqv wkdw wkh pd{lpxp
slorw glvwdqfh lv sursruwlrqdo wr wkh frkhuhqfh wlph/ qrupdol}hg wr wkh qxpehu
ri v|perov1
SVDP iru RIGP v|vwhpv zdv sursrvhg e| Krhkhu lq ^63`1 Wkh wzr0
glphqvlrqdo fkdqqho hvwlpdwlrq vfkhph xvhg lq RIGP v|vwhpv fdq eh vhhq dv
d jhqhudol}dwlrq ri wkh rqh0glphqvlrqdo SVDP vfkhph zrunlqj rqo| lq wkh wlph
gluhfwlrq1 Lq RIGP v|vwhpv d wzr0glphqvlrqdo hvwlpdwlrq dojrulwkp fdq eh
hpsor|hg vlqfh wkh fkdqqho wudqvihu ixqfwlrqv duh fruuhodwhg ehwzhhq dgmdfhqw
v|perov erwk lq wkh wlph dqg lq wkh iuhtxhqf| gluhfwlrq1 Wklv phdqv wkdw wkh
ryhudoo slorw0wr0gdwd v|pero udwlr fdq eh ghfuhdvhg frpsduhg wr vlqjoh0fduulhu
v|vwhpv ^63`1
Wkh rswlpdo olqhdu fkdqqho hvwlpdwru/ lq wkh vhqvh ri plqlpl}lqj wkh phdq
vtxduh huuru/ lv wkh wzr0glphqvlrqdo Zlhqhu owhu1 Wklv owhu lv edvhg rq nqrzo0
hgjh ri wkh wlph2iuhtxhqf| fruuhodwlrq ixqfwlrq dqg zhljkv wkh uhfhlyhg slorw
ydoxhv zlwk uhvshfw wr wkhlu fruuhodwlrq dqg qrlvh surshuwlhv lq rughu wr ghulyh
wkh fkdqqho hvwlpdwhv1 Iru wkh slorw srvlwlrq dw srvlwlrq Sl d whqwdwlyh fkdqqho
l ,/ iru wkh wudqvihu ixqfwlrq fdq eh rewdlqhg e| edfn0urwdwlqj wkh

hvwlpdwh/ K+S
uhfhlyhg vdpsohv dv
l , @ uSl >
K+S +517,
sSl
zkhuh S @ ^S3 > S4 > = = = > SO4 ` lv wkh vhw ri slorw v|pero srvlwlrqv xvhg iru wkh
hvwlpdwh/ sSl lv wkh slorw v|pero dw srvlwlrq Sl dqg uSl lv wkh fruuhvsrqglqj
uhfhlyhg ydoxh diwhu wkh GIW1 Wkh fkdqqho hvwlpdwh/ K a n>o / ri vxe0fkdqqho n dw
wlph o fdq wkhq eh frpsxwhg e| olqhdu owhulqj
O
[ 4
a n>o @
K l ,>
zn>o +Sl , K+S +518,
l@3
zkhuh O lv wkh qxpehu ri slorw v|perov xvhg iru wkh hvwlpdwh dqg zn>o +Sl , lv wkh
fkdqqho hvwlpdwlrq frh!flhqw xvlqj wkh slorw v|pero dw srvlwlrq Sl iru wkh gdwd
v|pero dw srvlwlrq +n> o,1 Li wkh fkdqqho lv zlgh0vhqvh vwdwlrqdu|/ wkh rswlpdo
O 4 yhfwru ri owhu frh!flhqwv lv wkh vroxwlrq wr wkh Zlhqhu0Krsi htxdwlrq
^64`
zn>o +S, @ 4 n>o > +519,
zkhuh n>o ghqrwhv wkh O 4 furvv0fruuhodwlrq yhfwru zlwk hohphqwv n>l +l, @
+Sl ,` dqg ghqrwhv wkh O O dxwrfruuhodwlrq pdwul{ zlwk hohphqwv
H^Kn>o K
l, K
l>m @ H^K+S +Sm ,`1 Wkh phdq vtxduh huuru +PVH, fdq eh xvhg dv d shu0
irupdqfh phdvxuh ri wkh hvwlpdwlrq owhu/ exw fdq dovr eh xvhg wr fdofxodwh wkh
uhvxowlqj EHU1 Wkh PVH xqghu lghdo frqglwlrqv +nqrzq fruuhodwlrq pdwulfhv,
lv jlyhq e| ^64`
Mn>o @ 5K Wn>o 4 n>o > +51:,
zkhuh 5K lv wkh yduldqfh ri wkh frpsoh{ frpsrqhqwv ri wkh fkdqqho1 Ilqdoo|/
diwhu whglrxv ghulydwlrqv/ wkh EHU iru T0SVN fdq eh fdofxodwhg dv ^89`
4 n>o
Sn>o +huuru, @ t > +51;,
5 5 5 5
n>o
zkhuh
5K
n>o @ Hv s 5 +51<,
+ K . 5q ,+ 5K . Mn>o ,
dqg 5q ghqrwhv wkh yduldqfh ri wkh qrlvh1

5141 FKDQQHO HVWLPDWLRQ 4:
pre-compensate
f estimate
uplink downlink
observation time prediction time
pilot symbol data symbol
Iljxuh 514= Wkh sulqflsoh ri suh0frpshqvdwlrq vkrzlqj wzr RIGP eorfnv1
51415 Suh0frpshqvdwlrq
Suh0frpshqvdwlrq/ vrphwlphv fdoohg suhfrglqj/ lv d whfkqltxh zkhuh wkh vljqdo
lv frpshqvdwhg ehiruh lw lv wudqvplwwhg lq rughu wr lpsuryh wkh shuirupdqfh lq
vrph vhqvh1 Wkh whfkqltxh fdq dovr eh xvhg wr pryh frpsoh{ ixqfwlrqv iurp
wkh uhfhlyhu wr wkh wudqvplwwhu/ wkhuhe| hqdeolqj ohvv frpsoh{ whuplqdo vwuxf0
wxuhv zlwk shuirupdqfh vlplodu wr pruh frpsoh{ rqhv1 Frpshqvdwlrq fdq/ iru
h{dpsoh/ eh xvhg wr plwljdwh LVL lq vlqjoh0fduulhu v|vwhpv wr dyrlg wkh qhhg
iru htxdol}dwlrq dw wkh uhfhlyhu ^78` ^;<` ^<3` ru iru dqwhqqd glyhuvlw| xvlqj rqo|
rqh uhfhlyh dqwhqqd ^44` ^5;` ^84` ^9;`1 Wkh whfkqltxh fdq dovr eh xvhg wr frp0
shqvdwh wkh idglqj fkdqqho lq rughu wr lpsuryh wkh shuirupdqfh ^F` ^7:` ru
wr pdlqwdlq ruwkrjrqdolw| ehwzhhq wkh glhuhqw xvhu vljqdov lq pxowl0fduulhu
vsuhdg0vshfwuxp +PF0VV, v|vwhpv ^6;` ^;8` ^;;`1 Gxh wr wkh vlpsolflw| dqg ur0
exvwqhvv d yduldqw iru dqwhqqd glyhuvlw| lv xvhg lq wkh GHFW v|vwhp ^5` zkhuh
wkh dqwhqqd uhfhlylqj wkh prvw srzhu gxulqj xsolqn lv xvhg iru grzqolqn wudqv0
plvvlrq1 Dovr lq wkh XPWV2Z0FGPD v|vwhp d vlpsoh skdvh frpshqvdwlrq
v|vwhp lv xvhg lq wlph glylvlrq gxsoh{ +WGG, prgh wr doljq wkh skdvhv ri
glhuhqw sdwkv ^68`1
Wkh whfkqltxh uholhv rq wkdw wkh fkdqqho fkdudfwhulvwlfv duh uhflsurfdo dqg
frqvwdqw gxulqj rqh uhfhlyh0wudqvplw0f|foh1 Wkhuhiruh suh0frpshqvdwlrq lv vxlw0
deoh zkhq WGG lv xvhg dqg wkh wlph ehwzhhq uhfhlylqj prgh dqg wudqvplvvlrq
prgh lv vkruw frpsduhg wr wkh frkhuhqfh wlph ri wkh fkdqqho1 Wkh fkdqqho
fkdudfwhulvwlfv duh lghqwlhg lq rqh gluhfwlrq dqg wkhq frpshqvdwlrq lv pdgh
sulru wr wudqvplvvlrq lq wkh rwkhu gluhfwlrq/ dv ghvfulehg lq Iljxuh 5141 Vlqfh
frpshqvdwlrq lv pdgh lq dgydqfh/ wkh pd{lpxp eorfn ohqjwkv duh ghwhuplqhg
e| wkh wlph0fruuhodwlrq ixqfwlrq ri wkh fkdqqho1 Wkh vkdsh ri wklv fruuhodwlrq
ixqfwlrq lv/ dprqj rwkhu wklqjv/ ghwhuplqhg e| wkh Grssohu iuhtxhqf|1 Kljk
4; FKDSWHU 51 UHVHDUFK WRSLFV
Grssohu iuhtxhqflhv qdwxudoo| fdoo iru vkruw eorfnv dqg wkhuhiruh wkh whfkqltxh
lv prvw vxlwdeoh iru v|vwhpv rshudwlqj dw shghvwuldq vshhgv1
Zkhq ixoo| frpshqvdwlqj wkh skdvh dqg dpsolwxgh ri d dw0idglqj fkdqqho
wkh frpsoh{ hqyhorsh ri wkh wudqvplwwhg vljqdo lv jlyhq e|
{+w, dujiKa +w,j

v+w, @ h > +5143,
a
K+w,
zkhuh {+w, lv wkh frpsoh{ hqyhorsh ri wkh xqfrpshqvdwhg vljqdo dqg K+w, a lv
wkh fkdqqho hvwlpdwh1 Li wkh fkdqqho lv Ud|ohljk idglqj wkh vljqdo xfwxdwlrqv
ehfrph vhyhuh dqg wkhuhiruh wkh dpsolwxgh kdv wr eh olplwhg/ hlwkhu e| olplwlqj
wkh rxwsxw srzhu ru e| lqklelwlqj wkh wudqvplwwhu zkhq wkh fkdqqho dwwhqxdwlrq
lv odujh ^F` ^7:`1 Lq d Ud|ohljk idglqj fkdqqho wkh idglqj glsv fdxvh prvw ri
wkh elw huuruv1 Li lw lv srvvleoh wr frpshqvdwh iru wkhvh wkhuh zloo/ iru xqfrghg
v|vwhpv/ eh d odujh lpsuryhphqw lq shuirupdqfh1 Lq ^7:` lw zdv vkrzq wkdw
lw lv srvvleoh wr ghfuhdvh wkh phdq wudqvplwwhu srzhu e| : gE +iru d {hg
EHU ri 4306 , zkhq lqwurgxflqj srzhu frpshqvdwlrq lq d qrq0frkhuhqw IVN
v|vwhp1 Lq ^F` lw lv vkrzq wkdw wkh EHU fdq eh ghfuhdvhg 53 wlphv frpsduhg
wr dq xqfrpshqvdwhg v|vwhp iru frkhuhqw TSVN dw He @Q3 @ 43 gE1 Wklv/
krzhyhu/ uhvxowv lq dq lqfuhdvh ri wkh shdn0wr0dyhudjh srzhu udwlr e| 43 gE vr
wklv vwudwhj| pljkw eh ri olplwhg lqwhuhvw iru zluhohvv dssolfdwlrqv1
Srzhu frpshqvdwlrq kdv dovr ehhq sursrvhg wr pdlqwdlq ruwkrjrqdolw| eh0
wzhhq xvhuv lq PF0VV v|vwhpv ^;8`1 Lq dq PF0VV v|vwhp/ ru pruh vshflfdoo|
lq dq PF0FGPD v|vwhp/ wkh gdwd v|pero iurp rqh xvhu lv vsuhdg wr pdq|
vxe0fkdqqhov dqg wkhq gdwd v|perov iurp vhyhudo xvhuv duh wudqvplwwhg rq wkh
vdph vxe0fkdqqhov vlpxowdqhrxvo| ^6<`1 Vlqfh hdfk gdwd v|pero lv wudqvplw0
whg ryhu vhyhudo vxe0fkdqqhov/ d iuhtxhqf| glyhuvlw| jdlq fdq eh dfklhyhg zkhq
dgglqj wkh frpsrqhqwv iurp wkh glhuhqw vxe0fkdqqhov1 Li frkhuhqw ghwhfwlrq
lv xvhg wkhq htxdol}dwlrq lv uhtxluhg lq wkh vdph zd| dv lq frqyhqwlrqdo RIGP
v|vwhpv1 Krzhyhu/ wkh ruwkrjrqdolw| ehwzhhq xvhuv fdq qrw eh suhvhuyhg lq
d pxowlxvhu v|vwhp zkhq htxdol}lqj wkh xsolqn vljqdo lq wkh uhfhlyhu1 Wkhuh0
iruh pxowlxvhu ghwhfwlrq ru suh0frpshqvdwlrq lv uhtxluhg lq rughu wr dfklhyh
jrrg shuirupdqfh ^;8`1 Wkh txhvwlrq iru suh0frpshqvdwlrq lq wklv dssolfdwlrq
lv zkhwkhu rqh fdq drug wkh uhvxowlqj lqfuhdvh lq shdn0wr0dyhudjh srzhu udwlr
uhsruwhg lq ^F`1
Suh0frpshqvdwlrq fdq dovr eh xvhg iru dqwhqqd glyhuvlw|1 Wkh whfkqltxh
kdv ehhq xvhg wr fkrrvh wudqvplw dqwhqqd lq wkh GHFW v|vwhp ^5`/ wr dfklhyh
frqvwuxfwlyh dgglwlrq iurp wzr dqwhqqdv ^44` ^5;` ru iurp dq duud| ri dqwhq0
qdv ^84`1 Lq ^44` d glyhuvlw| jdlq ri 9043 gE zdv uhsruwhg +dw EHU 4306 , iru d
v|vwhp xvlqj wkh dqwhqqd glyhuvlw| duudqjhphqw dqg dq htxdol}hu +wr plwljdwh
LVL,/ frpsduhg wr d v|vwhp xvlqj rqo| wkh htxdol}hu1 Wklv jdlq lv ri wkh vdph
rughu dv wkh 43 gE jdlq wkdw fdq eh dfklhyhg +iru TSVN dw EHU 4306 , xvlqj wzr
uhfhlyh dqwhqqdv lq d Ud|ohljk idglqj fkdqqho ^89`1 Dqwhqqd suh0frpshqvdwlrq
lv wkhuhiruh dq h!flhqw whfkqltxh wr jhw forvh wr rswlpdo dqwhqqd glyhuvlw| xvlqj
5151 PXOWL0FDUULHU V\VWHP GHVLJQ 4<
rqo| rqh uhfhlyh dqwhqqd1 Lq ^9;` wkh vzlwfkhg dqwhqqd glyhuvlw| v|vwhp zdv
jhqhudol}hg wr RIGP v|vwhpv1 Lq wklv v|vwhp/ vrph ri wkh vxe0fkdqqhov zhuh
wudqvplwwhg iurp rqh dqwhqqd dqg wkh uhpdlqlqj vxe0fkdqqhov zhuh wudqvplw0
whg iurp wkh rwkhu dqwhqqd1 Wklv v|vwhp dovr uhvxowhg lq d forvh wr rswlpdo
glyhuvlw| jdlq/ dvvxplqj lqghshqghqw vxe0fkdqqhov dqg lghdo v|qfkurql}dwlrq
dqg fkdqqho hvwlpdwlrq1 Wkh txhvwlrq lv/ krzhyhu/ krz wr ghulyh wkh qhfhvvdu|
fkdqqho hvwlpdwhv lq dq h!flhqw zd|1 Lw lv srvvleoh wr xvh vrph nlqg ri slorw0
edvhg fkdqqho hvwlpdwlrq rq d shu vxe0fkdqqho edvlv/ exw lw lv kdug wr xvh wkh
fruuhodwlrq ehwzhhq wkh vxe0fkdqqhov iru hvwlpdwlrq lq wkh uhfhlyhu vlqfh glhu0
hqw dqwhqqdv duh xvhg1 Dowhuqdwlyho|/ skdvh0rqo| suh0frpshqvdwlrq ^F` fdq eh
xvhg lq rughu wr doljq wkh skdvhv dqg dyrlg fkdqqho hvwlpdwlrq dw wkh uhfhlyhu1
D jhqhudo sureohp zkhq xvlqj suh0frpshqvdwlrq lq dq| irup lv wkdw wkh
fkdqqho qhhgv wr eh frqvwdqw/ ru doprvw frqvwdqw/ gxulqj hdfk uhfhlyh0wudqvplw0
f|foh1 Lq ^F` wkh pd{lpxp eorfn ohqjwk lv fdofxodwhg zkhuh wkh lqvwdqwdqhrxv
EHU jhwv zruvh wkdq wkdw ri glhuhqwldo ghwhfwlrq1 Lw lv vkrzq wkdw iru skdvh0
rqo| suh0frpshqvdwlrq d eorfn ohqjwk ri iG Weorfn 3=59 fdq eh xvhg ehiruh wkh
shuirupdqfh ghwhulrudwhv1 Wr frpsduh wklv qxpehu wr dq h{lvwlqj WGG v|vwhp
wklv fruuhvsrqgv wr d eorfn ohqjwk 73 wlphv orqjhu wkdq wkh rqh lq wkh GHFW
v|vwhp dw 4<33 PK} dqg d yhorflw| ri 43 np2k1 Wkh frqfoxvlrq dovr krogv zkhq
d zruvw0fdvh ghvljq lv pdgh iru wkh fkdqqho hvwlpdwruv ^F`1 Vlplodu eorfn ohqjwkv
fdq suredeo| eh h{shfwhg iru wkh rwkhu suh0frpshqvdwhg v|vwhpv/ exw wklv lv ri
frxuvh ghshqghqw rq wkh vhqvlwlylw| ri wkh uhfhlyhu wr glhuhqw nlqgv ri huuruv1
Wkh EHU lv ghshqghqw rq wkh delolw| ri pdnlqj fkdqqho suhglfwlrqv1 Li rqo| wkh
odvw hvwlpdwhv lq wkh xsolqn eorfn duh xvhg iru frpshqvdwlrq ri wkh grzqolqn
eorfn/ wkh shuirupdqfh lv ghjudghg frpsduhg wr wkh xvh ri rswlpdo suhglfwlrq1
Iru wkh skdvh0rqo| frpshqvdwlrq v|vwhp lq ^F` wkh pd{lpxp eorfn ohqjwk lv
dssur{lpdwho| iG Weorfn 3=38 zkhq wkh odvw hvwlpdwh ri wkh xsolqn eorfn lv
xvhg iru skdvh frpshqvdwlrq1 Lq ^6;`/ zkhuh wkh srzhu lv suh0frpshqvdwhg lq
dq PF0FGPD v|vwhp/ wkh pd{lpxp eorfn ohqjwk iru d wzr0irog ghjudgdwlrq lq
wkh EHU lv iG Weorfn 3=41 Erwk wkhvh qxpehuv fdq/ iru h{dpsoh/ eh frpsduhg
wr wkh dyhudjh gxudwlrq ri d 43 gE idglqj gls zklfk lv iG Wgls 3=46 iru d
Ud|ohljk idglqj fkdqqho zlwk lvrwurslf vfdwwhulqj ^;:`1
515 Pxowl0Fduulhu V|vwhp Ghvljq
Lq wklv vhfwlrq wkh edfnjurxqg ri wkh uhvhdufk duhdv uhodwhg wr RIGP v|vwhp
ghvljq lv suhvhqwhg1 Wkh wrslfv suhvhqwhg fryhu rswlpl}dwlrq ri wkh vxe0fkdqqho
edqgzlgwk/ slorw sdwwhuq ghvljq dqg suhdpeoh0edvhg v|qfkurql}dwlrq1 V|vwhp
ghvljq ri wkh sk|vlfdo od|hu lqfoxghv pruh wrslfv wkdq suhvhqwhg khuh/ exw wkhvh
duh rxw ri vfrsh iru wkh wkhvlv1 Rwkhu wrslfv lqfoxgh/ iru h{dpsoh/ fkrlfh ri wkh
prgxodwlrq irupdw/ ghvljq ri wkh frglqj vfkhph dqg pxowlsoh dffhvv vfkhph1
51514 Vxe0Fkdqqho Edqgzlgwk

Zkhq fkrrvlqj d vxlwdeoh vxe0fkdqqho edqgzlgwk rqh kdv wr pdnh d frp0
surplvh ehwzhhq wkh ryhukhdg fdxvhg e| wkh f|folf suh{ dqg lpsdluphqwv
lqwurgxfhg e| wkh kdugzduh dqg wkh wudqvplvvlrq fkdqqho1 Wkh f|folf suh{ lv
xvhg wr plwljdwh wkh hhfw ri wlph glvshuvlrq dqg lwv gxudwlrq kdv wr h{fhhg
wkh pd{lpxp h{fhvv ghod| ri wkh fkdqqho lq rughu wr dyrlg LVL1 Vlqfh wkh f|folf
suh{ lv d uhshwlwlrq ri wkh vdpsohv/ wkh vdpsoh udwh kdv wr eh lqfuhdvhg li wkh
gdwd udwh lv wr eh frqvwdqw diwhu lqwurgxflqj wkh f|folf suh{1 Wkhuhiruh/ wklv
ohdgv wr dq lqfuhdvh lq edqgzlgwk1 Lq dgglwlrq/ wkh uhfhlyhu xvhv rqo| vdpsohv
fruuhvsrqglqj wr wkh dfwxdo v|pero iru ghprgxodwlrq/ wkh vdpsohv xvhg iru wkh
f|folf suh{ duh glvuhjdughg/ dqg wkhuhiruh wkhuh lv dq hqhuj| orvv1 Wklv hqhuj|
orvv lv pruh hylghqw zkhq wkh v|pero wlph lv vkruw vlqfh wkh uhodwlyh gxudwlrq
ri wkh f|folf suh{ wkhq lv orqjhu1 Wkhuhiruh/ wkh f|folf suh{ fdoov iru d orqj
v|pero wlph/ zlwk lwv fruuhvsrqglqj qduurzedqg vxe0fkdqqhov1 Wkh v|pero
ohqjwk fdq qrw/ krzhyhu/ eh duelwudulo| orqj ehfdxvh lq rughu wr suhvhuyh ru0
wkrjrqdo vxe0fkdqqhov wkh lpsdfw ri wkh fkdqqho rq wkh wudqvplwwhg vljqdo kdv
wr eh frqvwdqw gxulqj hdfk RIGP v|pero1 Wklv phdqv wkdw wkh fkdqqho lv qrw
doorzhg wr fkdqjh/ wkhuh pxvw eh qr rvfloodwru rvhwv dqg qr rvfloodwru skdvh
qrlvh/ wkh vdpsolqj wlph kdv wr eh fruuhfw dqg doo kdugzduh frpsrqhqwv pxvw
eh lghdo1 Ri frxuvh/ wklv lv qrw srvvleoh lq sudfwlfh dqg wkhuhiruh rqh kdv wr
doorz d fhuwdlq ohyho ri LFL1
Wkh LFL dulvlqj iurp fkdqqho ru kdugzduh lpsdluphqwv fdq riwhq eh gh0
vfulehg dv h{wud dgglwlyh Jdxvvldq qrlvh dqg wkh frqwulexwlrqv iurp glhuhqw
lqwhuihuhqfh vrxufhv fdq eh dgghg wr jhw wkh ryhudoo lpsdfw rq wkh v|vwhp ^97`1
Wkh ohyho ri wkh lqwhuihuhqfh lv/ dprqj rwkhu wklqjv/ ghshqghqw rq wkh Grssohu
iuhtxhqf| dqg iuhtxhqf| rvhw lq uhodwlrq wr wkh vxe0fkdqqho edqgzlgwk1 Vlqfh
wkh vxe0fkdqqho edqgzlgwk lv lqyhuvho| sursruwlrqdo wr wkh qxpehu ri vxe0
fkdqqhov xvhg/ wkh lqwhuihuhqfh ohyho lv wkhuhiruh dovr ghshqghqw rq wklv qxpehu1
Lq prvw fdvhv/ wkh LFL lqfuhdvhv dv wkh vxe0fkdqqho edqgzlgwk lv fkrvhq wr eh
vpdoohu dqg wkhuhiruh lw lv lpsruwdqw wr lghqwli| wkh grplqdqw vrxufhv ri lqwhu0
ihuhqfh dqg fkrrvh d vxe0fkdqqho edqgzlgwk vxfk wkdw wkh LFL grhv qrw ehfrph
wkh olplwlqj idfwru lq wkh v|vwhp shuirupdqfh1 Jrrg ryhuylhzv ri wkh hhfwv
frqwulexwlqj wr wkh LFL/ lqfoxglqj h{suhvvlrqv ri wkh uhvxowlqj LFL ohyhov/ fdq
eh irxqg lq ^97` dqg ^99`1 Wkhuh duh dovr pdq| rwkhu sdshuv lq wkh olwhudwxuh
vshfldol}hg lq rqh ru d ihz vrxufhv ri lqwhuihuhqfh1 Lq ^99` wkh srzhu dw wkh nwk
rxwsxw ri wkh GIW lv ghfrpsrvhg lqwr irxu jurxsv
Hv Q Q3
S ^n` @ +SX ^n` . SLFL ^n` . SLVL ^n`, . > +5144,
Wv Q . J4 . J5 Wv
zkhuh SX ^n` ghqrwhv wkh +qrupdol}hg, xvhixo srzhu iurp wkh v|pero wr eh gh0
whfwhg/ SLFL ^n` ghqrwhv wkh lqwhuihulqj srzhu iurp rwkhu vxe0fkdqqhov/ SLVL ^n`
ghqrwhv frqwulexwlrqv iurp rwkhu RIGP v|perov dqg Q3 ghqrwhv wkh srzhu
vshfwudo ghqvlw| ri wkh qrlvh1 Lw fdq eh vkrzq ^99` wkdw wkh vxp ri wkh xvhixo
srzhu dqg wkh lqwhuihulqj srzhu lv frqvwdqw/ SX ^n`.SLFL ^n`.SLVL ^n` @ 4/ vr li
5151 PXOWL0FDUULHU V\VWHP GHVLJQ 54
wkh LFL lqfuhdvhv wkh xvhixo srzhu dxwrpdwlfdoo| ghfuhdvhv e| wkh vdph dprxqw1
Lq wkh iroorzlqj/ zh wuhdw LFL lq pruh ghwdlo/ vlqfh wklv lv d yhu| lpsruwdqw lvvxh
lq RIGP v|vwhp ghvljq1
Dq lpsruwdqw hhfw wr frqvlghu lq wkh v|vwhp ghvljq lv lqwhuihuhqfh gxh wr
fkdqqho fkdqjhv gxulqj dq RIGP v|pero1 Wkh duhd kdv/ iru h{dpsoh/ ehhq
vwxglhg lq ^76` ^94`1 Wklv lqwhuihuhqfh lv fdxvhg e| pryhphqwv ri hlwkhu wkh
wudqvplwwhu/ wkh uhfhlyhu ru uh hfwlrq srlqwv dqg lw lv kdug/ li qrw lpsrvvleoh/
wr frpshqvdwh iru ^94`1 Wudqvplwwhu ru uhfhlyhu pryhphqwv uhvxow lq wkh vdph
nlqg ri lqwhuihuhqfh dv rvfloodwru rvhwv ^97`/ exw lq d zluhohvv hqylurqphqw wkh|
fdxvh d Grssohu vsuhdg udwkhu wkdq d iuhtxhqf| vkliw dqg wkhuhiruh wkh| duh
pxfk kdughu wr frpshqvdwh1 Wkh srzhu ri wkh LFL gxh wr wkh wlph0vhohfwlyh
idglqj lv sursruwlrqdo wr wkh vtxduhg qrupdol}hg Grssohu iuhtxhqf| +iG Wv ,5 /
dqg wkhuhiruh lw lv lqyhuvho| sursruwlrqdo wr wkh vtxduh ri wkh vxe0fkdqqho edqg0
zlgwk ivf5 1 Wkh yduldqfh ri wkh htxlydohqw qrlvh fdxvhg e| wkh wlph0vhohfwlyh
fkdqqho fdq eh fdofxodwhg dv ^97`
5
5wv @ +iG Wv ,5 > +5145,
9
zkhq wkh uhfhlyhg vljqdo srzhu lv qrupdol}hg wr 41
Lq Ê`/ wkh vxe0fkdqqho edqgzlgwk lv rswlpl}hg zlwk uhvshfw wr wkh sk|vlfdo
skhqrphqd ghvfulehg deryh/ wkh hqhuj| orvv dqg wkh LFL gxh wr wkh wlph0
vhohfwlyh fkdqqho1 Wkh rswlpdo vxe0fkdqqho edqgzlgwk lv ghshqghqw rq wkh
vljqdo0wr0qrlvh udwlr vlqfh wkhuh lv d edodqfh ehwzhhq wkh LFL dqg wkh qrlvh>
wkhuhiruh lw lv kdug wr jlyh dq h{dfw h{suhvvlrq iru d vxlwdeoh vxe0fkdqqho edqg0
zlgwk1 Krzhyhu/ iru preloh dssolfdwlrqv zkhuh He @Q3 @ 43 58 gE d qru0
pdol}hg Grssohu iuhtxhqf| ri iG Wv 3=34 3=35 vhhpv vxlwdeoh Ê` ^99`1 Wklv
fruuhvsrqgv wr d ghjudgdwlrq lq VQU ri ohvv wkdq 4 gE ^97` dqg uhvxowv lq d
vljqdo0wr0lqwhuihuhqfh udwlr h{fhhglqj 63 gE ^76`1
LFL fdq dovr eh fdxvhg e| iuhtxhqf| rvhwv dqg iru frqvxphu surgxfwv/
zkhuh orz0frvw lpsohphqwdwlrqv duh lpsruwdqw/ wklv pljkw eh d vljqlfdqw
vrxufh ri lqwhuihuhqfh1 Wkh lqwhuihulqj qrlvh ohyho fdq eh dssur{lpdwhg e|
^97`
5
5ir +i Wv ,5 > +5146,
6
zkhuh i ghqrwhv wkh iuhtxhqf| rvhw1 Dv vhhq lq +5146,/ wkh yduldqfh lv djdlq
sursruwlrqdo wr wkh vtxduh ri wkh qrupdol}hg iuhtxhqf| rvhw/ %5f @ +i Wv ,5 /
dqg wkhuhiruh lqyhuvho| sursruwlrqdo wr wkh vtxduh ri wkh vxe0fkdqqho edqg0
5 1 Qrupdoo|/ iuhtxhqf| rvhw frpshqvdwlrq lv pdgh sulru wr wkh GIW
zlgwk ivf
dqg wkh lqwhuhvwlqj sdudphwhu lv wkhuhiruh wkh uhpdlqlqj iuhtxhqf| rvhw diwhu
frpshqvdwlrq1 Wkh h{suhvvlrq iru wkh yduldqfh ri wkh htxlydohqw qrlvh fdxvhg
e| iuhtxhqf| rvhw lv vlplodu wr wkh qrlvh fdxvhg e| wkh wlph0vhohfwlyh fkdqqho1
Wkhuhiruh/ iru wkh vdph ohyho ri lqwhuihuhqfh wkh pd{lpxp iuhtxhqf| rvhw lv
ri wkh vdph rughu dv wkh Grssohu iuhtxhqf|/ w|slfdoo| %f 3=34 3=35/ exw wklv
vkrxog eh vhhq rqo| dv d urxjk jxlgholqh1
Rwkhu hhfwv wkdw qhhg frqvlghudwlrq duh wkh lq xhqfh ri skdvh qrlvh ri wkh
rvfloodwru dqg wkh hhfw ri vdpsolqj huuruv1 Skdvh qrlvh kdv ehhq vwxglhg lq/ iru
h{dpsoh/ ^9` ^88` ^93` dqg lw ohdgv wr wzr nlqgv ri glvwruwlrq1 Iluvw/ wkhuh lv dq
LFL whup/ zkrvh yduldqfh lqfuhdvhv zlwk wkh qxpehu ri vxe0fduulhuv1 Lw lv kdug
wr jhw qxphulfdo h{suhvvlrqv iru wkh yduldqfh vlqfh wklv lv ghshqghqw rq wkh
prgho ri wkh skdvh qrlvh lqfoxglqj srvvleoh ihhgedfn lq wkh flufxlw1 Iru Zlhqhu
skdvh qrlvh/ wkh fruuhvsrqglqj VQU orvv gxh wr wkh LFL lv sursruwlrqdo wr wkh
rvfloodwru olqhzlgwk dqg lqyhuvho| sursruwlrqdo wr wkh vxe0fkdqqho edqgzlgwk
^88`1 Wkh skdvh qrlvh dovr ohdgv wr d skdvh huuru frpprq wr doo vxe0fkdqqhov
zlwklq dq RIGP v|pero1 Wkh yduldqfh ri wklv glvwruwlrq ghfuhdvhv dv wkh vxe0
fkdqqho edqgzlgwk ghfuhdvhv ^93`1 Wkh fruuhodwlrq ri wkh frpprq skdvh huuru
lv lq jhqhudo orz ehwzhhq vxffhvvlyh RIGP v|perov/ vr frpshqvdwlrq rq d
v|pero0e|0v|pero edvlv lv uhtxluhg li wkh yduldwlrqv duh vljqlfdqw1
Vdpsolqj rvhwv ohdg wr d skdvh urwdwlrq ri wkh v|perov1 Ixuwkhu/ li vrph
sdwkv ri wkh fkdqqho idoo rxwvlgh wkh jxdug lqwhuydo/ LFL dqg LVL rffxu1 Wkh
uhvxowlqj lqwhuihuhqfh fdq eh dssur{lpdwhg dv Jdxvvldq qrlvh zlwk yduldqfh ^97`

[ 5 %l q %l r5
5wr mdl +w,m 5 > +5147,
l
Q Q
zkhuh %l lv wkh qrupdol}hg hhfwlyh wlph rvhw ghqhg dv

; w
A l w A l
? Wvdps
%l @ l JW vdps w 3 ? w ? l JWvdps +5148,
A Wvdps
=
3 hovh
wkurxjk wkh wlph rvhw w dqg wkh vdpsoh wlph Wvdps @ 4@ivdps 1 Dv ehiruh/ Q
ghqrwhv wkh qxpehu ri vxe0fkdqqhov/ J ghqrwhv wkh ohqjwk ri wkh f|folf suh{/
l ghqrwhv wkh ghod| ri sdwk l/ dqg dl +w, ghqrwhv wkh fkdqqho wds ydoxh ri sdwk
l1 Wkh yduldqfh ri wkh LFL dqg LVL gxh wr vdpsolqj rvhwv lv kljko| ghshqghqw
rq wkh srzhu ghod| suroh ri wkh fkdqqho ^97` dqg wkh glvwxuedqfh lv lq sulqflsoh
sursruwlrqdo wr wkh vxe0fkdqqho edqgzlgwk ^99`1 Li pdq| vxe0fkdqqhov duh xvhg/
wkh uhodwlyh lpsruwdqfh ri vdpsohv fruuhvsrqglqj wr sdwkv idoolqj rxwvlgh wkh
f|folf suh{ ehfrphv vpdoohu dqg wkhuhiruh wkh ryhudoo glvwruwlrq ghfuhdvhv1
Ilqdoo| vrph uhpdunv derxw wkh uhodwlrq ehwzhhq wkh vxe0fkdqqho edqg0
zlgwk dqg wkh ohqjwk ri sdfnhwv lq zluhohvv v|vwhpv1 Wkh qxpehu ri vxe0
fkdqqhov xvhg ghwhuplqhv wkh plqlpxp qxpehu ri v|perov lq d sdfnhw dqg
wkhuhiruh dovr wkh judqxodulw| zkhq fkrrvlqj wkh sdfnhw ohqjwkv1 Iru odujh gdwd
sdfnhwv wklv pljkw qrw eh d sureohp/ exw li wkhuh duh pdq| vkruw sdfnhwv/ ru
wkhuh duh orz0udwh frqqhfwlrqv zlwk frqvwudlqwv rq wkh ghod|/ wklv hhfw kdv wr
eh frqvlghuhg zkhq ghwhuplqlqj wkh vxe0fkdqqho edqgzlgwk1
51515 Suhdpeoh0Edvhg V|qfkurql}dwlrq

Wkhuh duh pdq| v|qfkurql}dwlrq sdudphwhuv wr eh hvwlpdwhg ehiruh v|pero gh0
flvlrqv fdq eh pdgh1 Wkh ghprgxodwru kdv wr nqrz zkhuh wkh vwduw ri d sdfnhw
lv/ zkhuh wkh RIGP v|perov vwduw/ krz odujh wkh iuhtxhqf| rvhw lv hwf1 Lq
sdfnhw0edvhg v|vwhpv lw lv ghvludeoh wr plqlpl}h wkh dftxlvlwlrq wlph dqg wkhuh0
iruh slorw0dlghg dssurdfkhv iru v|qfkurql}dwlrq duh vxlwdeoh/ wkrxjk wkhvh ohdg
wr d wkurxjksxw ghjudgdwlrq1 Riwhq d suhdpeoh lv xvhg/ zklfk lv wudqvplwwhg
ehiruh hdfk qhz sdfnhw1 Vxfk dq dssurdfk lv xvhg lq wkh KlshuODQ25 v|vwhp
^7`1
Wkh lq xhqfhv ri wlph dqg iuhtxhqf| rvhwv kdyh ehhq ghvfulehg lq Vhfwlrq
515141 Wkh yduldqfh ri wkh lqwhuihuhqfh fdxvhg e| iuhtxhqf| rvhwv lv jlyhq
lq +5146, dqg lw lv hylghqw wkdw wkh RIGP ghprgxodwru lv yhu| vhqvlwlyh wr
iuhtxhqf| rvhwv1 Lq rughu wr dfklhyh d wrohudeoh ghjudgdwlrq lq v|vwhp shu0
irupdqfh wkh uhpdlqlqj iuhtxhqf| rvhw pxvw qrw h{fhhg d ihz shufhqw ri wkh
vxe0fkdqqho edqgzlgwk1 Wkh uhtxluhphqwv rq wkh wlplqj hvwlpdwru duh vrph0
zkdw orzhu gxh wr wkh urexvwqhvv lqwurgxfhg e| wkh f|folf suh{1 Li doo ghod|hg
sdwkv ri wkh fkdqqho idoo lqwr wkh f|folf suh{/ ruwkrjrqdolw| lv suhvhuyhg dqg
qhlwkhu LVL qru LFL dulvh gxh wr wkh wlph rvhw> rwkhuzlvh wkh ohyho ri lqwhuihu0
hqfh lv jlyhq e| +5147,1
V|qfkurql}dwlrq lq RIGP v|vwhpv lv riwhq edvhg rq wkh wudqvplvvlrq ri d
uhshdwhg v|pero1 Wkh v|qfkurql}dwlrq sdudphwhuv duh wkhq ghulyhg e| fruuh0
odwlqj fxuuhqwo| uhfhlyhg vdpsohv zlwk hduolhu vdpsohv1 Wlph v|qfkurql}dwlrq lv
dfklhyhg zkhq wkh uhvxow ri wkh fruuhodwlrq h{fhhgv d fhuwdlq wkuhvkrog/ zklfk
phdqv wkdw wkh fxuuhqw vdpsohv fruuhvsrqg wr wkh hduolhu uhfhlyhg rqhv1 Zkhq
wkh fruuhfw vwduw ri wkh iudph kdv ehhq irxqg wkh iuhtxhqf| rvhw lv hvwlpdwhg
e| wkh skdvh vkliw ehwzhhq wkh uvw dqg vhfrqg sduw ri wkh uhshdwhg v|pero1
Qrupdoo| fruuhodwlrq lv shuiruphg xvlqj wkh wlph grpdlq vdpsohv ^96`/ exw lw
fdq dovr eh shuiruphg xvlqj wkh vljqdo diwhu wkh IIW ^83`1 Lq wkh odwwhu fdvh/
wkhuh lv d shuirupdqfh orvv iru odujh iuhtxhqf| rvhwv/ exw wkh phwkrg lv vwloo
srvvleoh wr xvh1
Wkh vshflf vwuxfwxuh ri wkh suhdpeoh ydulhv/ exw wkh sulqflsoh lv wkh vdph
iru doo wkhvh phwkrgv1 Lq ^83` dqg ^95` d uhshdwhg ixoo0ohqjwk RIGP v|pero
lv xvhg1 Lq ^79` dqg ^96` wkh uhshwlwlrq lv dffrpsolvkhg zlwklq rqh v|pero e|
rqo| prgxodwlqj hyhu| qwk vxe0fkdqqho1 Wkh vkruwhu wlph ehiruh uhshdwlqj wkh
vljqdo phdqv dq lqfuhdvhg iuhtxhqf| rvhw fdswxuh udqjh ri wkh v|qfkurql}hu
dw wkh frvw ri d fruuhvsrqglqj lqfuhdvh lq yduldqfh ri wkh hvwlpdwh1 Rwkhu ds0
surdfkhv xvh wkh uhshwlwlrq gxh wr wkh f|folf suh{ ^6:` ^;`1 Li wkh h{lvwlqj
f|folf suh{ lv xvhg/ wkhq wkhuh lv qr ryhukhdg lqwurgxfhg dqg wklv lv ri frxuvh
ghvludeoh1 Lq vrph fdvhv/ wkh f|folf suh{ lv wrr vkruw wr jlyh vx!flhqw hvwl0
pdwlrq dffxudf|/ dqg wkhq dyhudjlqj ryhu vhyhudo v|perov lv uhtxluhg/ zlwk d
fruuhvsrqglqj ghod|1 Lq ^74` d vshfldo v|qfkurql}dwlrq vhtxhqfh/ d FD]DF0
vhtxhqfh +frqvwdqw0dpsolwxgh }hur dxwrfruuhodwlrq vhtxhqfh,/ lv lqvhuwhg rq
hdfk vlgh ri dq RIGP v|pero iru v|qfkurql}dwlrq sxusrvhv1 Wkh vhtxhqfh lv
uhshdwhg erwk ehwzhhq wkh v|perov dqg zlwklq wkh eorfn/ zklfk phdqv wkdw
frduvh v|qfkurql}dwlrq fdq eh shuiruphg zlwklq wkh eorfn dqg qh iuhtxhqf|
v|qfkurql}dwlrq fdq eh shuiruphg e| xvlqj dgmdfhqw eorfnv1 Ilqdoo|/ lq ^G`
d uhshdwhg p0vhtxhqfh +pd{lpdo0ohqjwk vhtxhqfh, lv xvhg1 Wklv dssurdfk lv
prwlydwhg e| wkh dwwudfwlyh fruuhodwlrq surshuwlhv iru wkhvh vhtxhqfhv1
Wkhuh duh dovr phwkrgv edvhg rq sulqflsohv rwkhu wkdq uhshdwhg v|perov1
Lq ^75` wkh slorw v|perov lqwhqghg iru fkdqqho hvwlpdwlrq duh xvhg wrjhwkhu
zlwk wkh f|folf suh{ wr ghulyh dq lpsuryhg wlph v|qfkurql}dwlrq vljqdo1 Lq
^46` wkh suhdpeoh lv jhqhudwhg xvlqj rqo| wkh kljkhvw ru orzhvw kdoi ri wkh vxe0
fkdqqhov1 Lq wkh v|vwhp/ wlph v|qfkurql}dwlrq lv dfklhyhg e| frpsdulqj wkh
vshfwudo frqwhqw lq wkh wzr kdoyhv ri wkh qrupdo wudqvplvvlrq edqgzlgwk1 Lq
^98` d vshfldo orz0udwh vljqdolqj fkdqqho xvlqj iuhtxhqf| vkliw nh|lqj +IVN, lv
xvhg iru v|qfkurql}dwlrq sxusrvhv1 Wklv dssurdfk zdv prwlydwhg e| frpsoh{lw|
dujxphqwv vlqfh wkhq d vlpsoh IVN uhfhlyhu fdq eh xvhg gxulqj vohhs prgh1 Lq
^8<` frqwlqxrxv slorw fduulhuv zhuh xvhg wr ghulyh d iuhtxhqf| rvhw hvwlpdwh1
Dqrwkhu dssurdfk lv edvhg rq vkruw srzhu yduldwlrqv lq wkh wudqvplwwhg vljqdo
^95`1 E| xvlqj vr fdoohg qxoo v|perov wkh uhfhlyhu rqo| kdv wr orrn iru fhuwdlq
yduldwlrqv lq wkh uhfhlyhg srzhu wr qg d frduvh wlph hvwlpdwh/ zklfk fdq eh
shuiruphg xvlqj d orz0frpsoh{lw| uhfhlyhu zlwkrxw shuiruplqj dq IIW1
Wkh pdlq glhuhqfh ehwzhhq wkh dssurdfk lq ^G` dqg wkh rwkhu fruuhodwlrq0
edvhg phwkrgv lv wkdw lq ^G` fruuhodwlrq zlwk wkh nqrzq vhtxhqfh lv pdgh sulru
wr pxowlsolfdwlrq zlwk wkh vhfrqg sduw ri wkh uhshdwhg vhtxhqfh zkhq fdofx0
odwlqj wkh wlplqj v|qfkurql}dwlrq1 Wklv v|qfkurql}hu lv fdoohg d suh0fruuhodwlqj
v|qfkurql}hu1 Wkh v|qfkurql}dwlrq vljqdo fdq iru erwk v|qfkurql}huv eh fdofx0
odwhg dv
OS4 NS4
^n> d` @ s ^n q oU dù^n q oU`
o@3 q@3
N 4 & +5149,
S
s ^n q +o . T,U dù^n q +o . T,U` >
q@3
zkhuh d lv wkh odj ehwzhhq wkh nqrzq v|qfkurql}dwlrq vhtxhqfh dqg wkh uh0
fhlyhg vhtxhqfh/ O lv wkh qxpehu ri surgxfwv xvhg iru wkh v|qfkurql}dwlrq
vljqdo/ s^n` lv wkh nqrzq vhtxhqfh/ T lv dq h{wud ghod| zklfk fdq eh xvhg wr
lpsuryh wkh shuirupdqfh ri wkh iuhtxhqf| rvhw hvwlpdwh/ N lv wkh fruuhodwru
ohqjwk dqg U lv wkh uhshwlwlrq ohqjwk ri wkh vhtxhqfh1 Wkh rqo| glhuhqfh
ehwzhhq wkh frqyhqwlrqdo fruuhodwlrq0edvhg v|qfkurql}hu dqg wkh v|qfkurql}hu
lq ^G` lv krz wkhvh sdudphwhuv duh fkrvhq1 Wkh frqyhqwlrqdo v|qfkurql}hu lv
vkrzq lq Iljxuh 515 dqg wklv lv rewdlqhg e| fkrrvlqj wkh fruuhodwlrq ohqjwk wr
N @ 4> wkh qxpehu ri surgxfwv xvhg iru wkh hvwlpdwh wr O @ U/ wkh nqrzq
vhtxhqfh wr s^n` @ 4 dqg T @ 41 Wkh suh0fruuhodwlqj v|qfkurql}hu lv vkrzq lq
Iljxuh 5161 Wkh v|qfkurql}dwlrq vljqdo lv rewdlqhg e| fkrrvlqj wkh qxpehu ri
surgxfwv wr O @ 4> wkh fruuhodwlrq ohqjwk wr wkh uhshwlwlrq ohqjwk N @ U dqg
T @ 4= Wkh suh0fruuhodwlrq zlwk wkh nqrzq vhtxhqfh phdqv wkdw wkh qrlvh lv
received samples, r
timing
( )*
freq.
R
offset

arg L
Iljxuh 515= Vwuxfwxuh ri wkh frqyhqwlrqdo v|qfkurql}hu1 Wklv fruuhvsrqgv wr

N @ 4/ T @ 4 dqg O @ U lq +5149,1
K
received samples, r
timing
( )*
freq. R
offset
arg
Iljxuh 516= Vwuxfwxuh ri wkh suh0fruuhodwlqj v|qfkurql}hu1 Wklv fruuhvsrqgv wr

N @ U/ T @ 4 dqg O @ 4 lq +5149,1
dyhudjhg ehiruh pxowlsolfdwlrq lv shuiruphg1 Wkhuhiruh/ wkh phdq dpsolwxgh ri

wkh v|qfkurql}dwlrq vljqdo zkhq wkh wlplqj lv zurqj lv ghfuhdvhg zkhq xvlqj
wklv v|qfkurql}hu lqvwhdg ri wkh frqyhqwlrqdo v|qfkurql}hu/ zklfk lq wxuq uh0
vxowv lq d shuirupdqfh hqkdqfhphqw1 Lq ^74` d vlplodu dssurdfk zdv phqwlrqhg
iru wkh vshfldo wudlqlqj vhtxhqfh xvhg/ exw wklv zdv qrw wuhdwhg ixuwkhu1
Wkh frpsoh{lw| ri wkh v|qfkurql}hu lv dq lpsruwdqw lvvxh/ hvshfldoo| lq vohhs
prgh zkhuh orz srzhu frqvxpswlrq lv lpsruwdqw1 Wkh v|qfkurql}dwlrq phwk0
rgv edvhg rq wkh srzhu frpsdulvrq dqg wkh vhsdudwh vljqdolqj fkdqqho duh zhoo
vxlwhg iru orz0srzhu lpsohphqwdwlrqv/ exw wkhuh lv d odujh shuirupdqfh ghjud0
gdwlrq frpsduhg wr wkh fruuhodwlrq0edvhg phwkrgv1 Orz0frpsoh{lw| yduldqwv ri
wkh fruuhodwlrq0edvhg phwkrgv xvlqj rqo| rqh0elw uhvroxwlrq duh sursrvhg lq ^<`
dqg ^6:` iru v|qfkurql}dwlrq xvlqj wkh f|folf suh{ dqg lq ^G` iru wkh fdvh ri d
SQ0vhtxhqfh suhdpeoh1 Lq ^6:` wkh v|qfkurql}dwlrq vljqdo lv xvhg erwk iru wlph
dqg iuhtxhqf| v|qfkurql}dwlrq dqg lw lv vkrzq wkdw lw lv srvvleoh wr dfklhyh
vdwlvidfwru| uhvxowv iru wkh iuhtxhqf| rvhw hvwlpdwh xvlqj rqo| rqh0elw uhvr0
oxwlrq ri wkh uhfhlyhg vdpsohv1 Krzhyhu/ wkh f|folf suh{ zdv txlwh orqj dqg
dyhudjlqj ryhu vhyhudo v|perov zdv uhtxluhg lq rughu wr ghfuhdvh wkh yduldqfh1
Wkh SQ0edvhg suhdpeoh lv hvshfldoo| vxlwhg iru wkh rqh0elw uhvroxwlrq v|qfkur0
ql}hu vlqfh wkh wudqvplwwhg vhtxhqfh kdv frqvwdqw dpsolwxgh dqg fruuhodwlrq
lv shuiruphg ehiruh pxowlsolfdwlrq1 Lq ^G` lw lv vkrzq wkdw wkh orz0uhvroxwlrq
SQ0edvhg v|qfkurql}hu lq idfw fdq dfklhyh orzhu idovh ghwhfwlrq suredelolw|
dqg orzhu suredelolw| ri plvvlqj wkh v|qfkurql}dwlrq vljqdo frpsduhg wr d ixoo0
uhvroxwlrq frqyhqwlrqdo v|qfkurql}hu edvhg rq d uhshdwhg RIGP v|pero1 Wkh
vkruw suhdpeoh xvhg lv/ krzhyhu/ qrw orqj hqrxjk wr jlyh vdwlvidfwru| uhvxowv
iru orz0uhvroxwlrq iuhtxhqf| rvhw hvwlpdwlrq1
Wkh yduldqfh ri wkh iuhtxhqf| rvhw hvwlpdwh lv vlplodu iru doo wkh fruuhodwlrq0
edvhg phwkrgv1 Wkh shuirupdqfh lv lq sulqflsoh ghwhuplqhg e| wkh qxpehu ri
revhuyhg vdpsohv dqg wkh ghod| iru wkh fruuhodwlrq1 Wkh RIGP0edvhg phwk0
rgv zkhuh vdpsoh0e|0vdpsoh pxowlsolfdwlrq lv xvhg vkrz d yhu| urexvw ehkdylru1
Wkh skdvh ri wkh v|qfkurql}dwlrq vljqdo kdv d dw uhjlrq durxqg wkh fruuhfw wlp0
lqj dqg wkh vhqvlwlylw| ri wlplqj huuruv lv wkhuhiruh orz iru iuhtxhqf| hvwlpdwlrq
xvlqj wkhvh phwkrgv1 Wkh SQ0edvhg phwkrg kdv d yhu| vkdus v|qfkurql}dwlrq
shdn1 Wklv pdnhv lw yhu| hdv| wr lghqwli| wkh shdn/ exw li wkhuh lv d wlplqj huuru
wkh shuirupdqfh ri wkh iuhtxhqf| rvhw hvwlpdwh zloo ghwhulrudwh vljqlfdqwo|1
51516 Slorw Sdwwhuq Ghvljq

Zkhq ghvljqlqj wkh slorw sdwwhuq wkh wdvn lv wr xvh vx!flhqwo| pdq| slorwv
lq rughu wr vxssuhvv wkh qrlvh dqg wr eh deoh wr iroorz wkh fkdqjhv lq wkh
wudqvihu ixqfwlrq zlwkrxw lqfuhdvlqj wkh hqhuj| orvv dqg edqgzlgwk h{sdqvlrq
wrr pxfk1 Wkh uhfhlyhu kdv wr hvwlpdwh wkh ydoxhv ri wkh wudqvihu ixqfwlrq lq
d wzr0glphqvlrqdo julg/ dqg wkhuhiruh lw lv dovr lpsruwdqw zkhuh wkh slorwv duh
orfdwhg1 Wkh slorw sdwwhuq ghvfulehv zkhuh lq wkh wlph dqg iuhtxhqf| sodqh
wr wudqvplw wkh slorw v|perov1 Wkhuh duh rqo| d ihz sdshuv lq wkh olwhudwxuh
5151 PXOWL0FDUULHU V\VWHP GHVLJQ 5:
uhjduglqj wkh ghvljq ri wklv sdwwhuq1 Ndlvhu jlyhv d jrrg ryhuylhz ri fkdqqho
hvwlpdwlrq dqg uhodwhg wrslfv iru pxowl0fduulhu v|vwhpv lq ^6<`/ exw lw lv kdug
wr qg dqdo|wlfdo wuhdwphqwv1 Lq ^D` lw zdv vkrzq wkdw wkh slorw sdwwhuq kdv
d odujh lq xhqfh rq wkh uhvxowlqj shuirupdqfh1 E| ghvljqlqj wkh sdwwhuq lq
dq h!flhqw zd| lw lv srvvleoh wr lqfuhdvh wkh Grssohu udqjh ryhu zklfk wkh
v|vwhp fdq rshudwh e| xs wr whq wlphv ^D` frpsduhg wr d v|vwhp xvlqj d slorw
sdwwhuq vlplodu wr d vlqjoh0fduulhu fdvh1 Lq wkh sdshu/ d Ndopdq owhu zdv xvhg
wr ghvljq wkh slorw sdwwhuq e| vhdufklqj iru wkh ehvw sodfh wr sxw wkh qh{w
slorw v|pero1 Pxfk zrun fdq eh sxw lqwr wkh ghvljq ri wkh slorw sdwwhuqv/ dqg
rswlpdo fkdqqho hvwlpdwlrq owhuv fdq eh xvhg iru wkh dqdo|vhv1 Lq uhdo0wlph
rshudwlrq ohvv frpsoh{ owhuv fdq qdwxudoo| eh xvhg1 Lq ^86` d Zlhqhu owhu zdv
xvhg iru fkdqqho hvwlpdwlrq dqg wkh wkhruhwlfdo v|pero huuru udwh zdv ghulyhg
edvhg rq wkh PVH ri wkh hvwlpdwlrq owhu1 Wkh lq xhqfh ri wkh slorw vsdflqj lq
wlph dqg iuhtxhqf| zdv wkhq lqyhvwljdwhg xvlqj d uhfwdqjxodu sdwwhuq1 Xqghu
rswlpdo frqglwlrqv/ l1h1 zkhq wkh fruuhodwlrq ixqfwlrqv duh nqrzq/ wkh dxwkruv
lq ^86` frqfoxghg wkdw d slorw vsdflqj forvh wr wkh vdpsoh wkhruhp zdv vx!flhqw1
Krzhyhu/ iru zruvw0fdvh ghvljqv/ ru zkhq ohvv frpsoh{ qrq0rswlpdo fkdqqho
hvwlpdwlrq owhuv duh xvhg/ wkh glvwdqfh ehwzhhq wkh slorw v|perov kdv wr eh
ghfuhdvhg e| dw ohdvw d idfwru wzr1 Wkh pd{lpxp slorw glvwdqfh lq iuhtxhqf|
+phdvxuhg lq qxpehu ri vxe0fkdqqhov, gxh wr wkh vdpsolqj wkhruhp lv ^4<`
4
QI > +514:,
pd{ ivf
zkhuh pd{ ghqrwhv wkh pd{lpxp h{fhvv ghod| ri wkh fkdqqho1 Wkh pd{lpxp
slorw glvwdqfh lq wlph lv vlploduo| jlyhq e| +516,1 Wkh pd{lpxp slorw glvwdqfh
lq wlph lv wkhuhiruh sursruwlrqdo wr wkh frkhuhqfh edqgzlgwk/ qrupdol}hg e|
wkh vxe0fkdqqho edqgzlgwk1 Lq ^64` wkh whup edodqfhg ghvljq zdv lqwurgxfhg/
phdqlqj wkdw wkh hvwlpdwlrq xqfhuwdlqw| lq wkh wlph gluhfwlrq dqg lq wkh iuh0
txhqf| gluhfwlrq vkrxog eh wkh vdph1 Dv d uxoh ri wkxpe wkh dxwkruv sursrvhg
wzr wlphv ryhuvdpsolqj/ uhvxowlqj lq wkh iroorzlqj jxlgholqh4 =
5iG Wv QW pd{ ivf QI 4@5> +514;,
zkhuh QI lv wkh slorw vsdflqj lq wkh iuhtxhqf| gluhfwlrq1 Wklv jxlgholqh frlq0
flghv zlwk wkh uhvxowv lq ^86`1
Wkh sxeolfdwlrqv phqwlrqhg deryh ghdo pdlqo| zlwk uhfwdqjxodu slorw sdw0
whuqv dffruglqj wr Iljxuh 5171 Dqrwkhu dowhuqdwlyh lv wr xvh d kh{djrqdo sdwwhuq
dqg wklv dssurdfk zdv lqyhvwljdwhg lq ^57`1 Wkh dxwkruv frpsduhg wkh shuiru0
pdqfh ri glhuhqw sdwwhuqv dqg frqfoxghg wkdw d kh{djrqdo julg jlyhv ehwwhu
shuirupdqfh vlqfh lw ohdgv wr d pruh h!flhqw vdpsolqj ri wzr0glphqvlrqdo vlj0
qdov1 Krzhyhu/ wr pdnh d pruh jhqhudo frqfoxvlrq/ lqyhvwljdwlrqv iru pruh
fkdqqhov wkdq wkh KI fkdqqho dgguhvvhg lq ^57` duh uhtxluhg1
4 Lq wkh ruljlqdo sxeolfdwlrq ^64` edodqfhg ghvljq zdv ghqhg dv s( Ar u A E 4@ srS u
8 E *ec exw wklv zdv fkdqjhg wr +514;, lq ^6<`1
5; FKDSWHU 51 UHVHDUFK WRSLFV
f
NF
NT
data symbol pilot symbol
Iljxuh 517= H{dpsoh ri d uhfwdqjxodu slorw sdwwhuq1 QI lv wkh slorw glvwdqfh lq

iuhtxhqf| zkhuhdv QW lv wkh glvwdqfh lq wlph1
Wkh lpsruwdqw frqfoxvlrq lq doo wkh sdshuv lv wr vsuhdg wkh slorw v|perov
lq erwk wlph dqg iuhtxhqf|/ rwkhuzlvh wkhuh zloo eh d shuirupdqfh ghjudgdwlrq
gxh wr ryhukhdg ru lqdffxudwh fkdqqho hvwlpdwhv1 Lq wkh KlshuODQ25 v|vwhp
^7` wkh slorw v|perov duh frqwlqxrxvo| wudqvplwwhg dw wkh vdph vxe0fkdqqhov
gxulqj wkh zkroh eorfn dqg wklv surshuw| lv wkhuhiruh qrw ixooohg1 Hdfk gdwd
eorfn lv suhfhghg e| d suhdpeoh/ exw iru d orqj eorfn lw lv kdug wr pdnh ixoo
xvh ri wkh wlph fruuhodwlrq ri wkh fkdqqho zkhq ghulylqj wkh fkdqqho hvwlpdwhv1
Lq ^;3` lw zdv vkrzq wkdw lw zdv srvvleoh wr jdlq 40418 gE mxvw e| uhduudqjlqj
wkh slorw v|perov vr wkh| duh orfdwhg dw glhuhqw vxe0fkdqqhov1 Wklv vwxg|
zdv shuiruphg iru dq xqfrghg v|vwhp zlwkrxw xvlqj wkh suhdpeoh iru fkdqqho
hvwlpdwlrq/ exw vlpxodwlrqv vkrzhg ^53` wkdw wkh vdph jdlq frxog eh dfklhyhg
lq d frghg v|vwhp lqfoxglqj wkh suhdpeoh iru fkdqqho hvwlpdwlrq1
Wkh fkdqqho prgho iru wkh wlph fruuhodwlrq xvhg lq ^D` lv dq dxwr0uhjuhvvlyh
prgho/ zklfk dovr kdv ehhq xvhg e| rwkhu uhvhdufkhuv ^69` ^;7`1 Lw lv kdug wr
frpsduh wkh uhvxowlqj slorw sdwwhuqv lq ^D` gluhfwo| wr wkh jxlgholqhv deryh
vlqfh wkh vkdsh ri wkh fruuhodwlrq ixqfwlrqv lv glhuhqw1 Lq dgglwlrq/ wkhuh
lv d glhuhqfh lq wkdw rqo| rqh slorw v|pero dw d wlph lv xvhg lq ^D`/ zkloh
pdq| slorw v|perov duh wudqvplwwhg vlpxowdqhrxvo| lq wkh rwkhu lqyhvwljdwhg
v|vwhpv1 Krzhyhu/ wkh jhqhudo uhvxow vwloo krogv wkdw wkh slorw v|perov vkdoo
eh vsuhdg erwk lq wkh wlph dqg lq wkh iuhtxhqf| gluhfwlrq vr wkdw wkh wlph
fruuhodwlrq dqg iuhtxhqf| fruuhodwlrq ri wkh fkdqqhov fdq eh xvhg wr lpsuryh
wkh fkdqqho hvwlpdwhv1 Lq ^4<` wkh lq xhqfh ri wkh hgjh hhfwv zdv phqwlrqhg
5161 VXSHULPSRVHG SLORW VHTXHQFHV 5<
data xk
pilot {1,-1} pk
Iljxuh 518= Jhqhudwlrq ri wkh vljqdo zkhq xvlqj vxshulpsrvhg slorw v|perov1
dqg lw zdv vxjjhvwhg wr kdyh pruh slorw v|perov dw wkh rxwhu v|perov lq d
sdfnhw wr plwljdwh wklv hhfw1 Wklv fruuhvsrqgv zhoo wr wkh uhvxow ri wkh slorw
sdwwhuq vhdufk lq ^D`/ zkhuh wkh rxwhu fkdqqhov duh xvhg iru slorw v|perov pruh
iuhtxhqwo|1
516 Vxshulpsrvhg Slorw Vhtxhqfhv

Vxshulpsrvhg slorw vhtxhqfhv phdqv wkdw wkh nqrzq slorw vhtxhqfh lv ryhuodlg
rq wkh gdwd vhtxhqfh dqg wudqvplwwhg lq wkh vdph iuhtxhqf| edqg dqg dw wkh
vdph wlph dv wkh gdwd vhtxhqfh1 Wkh wudqvplwwhg vljqdo frqwdlqv erwk d nqrzq
slorw sduw dqg dq xqnqrzq gdwd sduw1 Lw lv jhqhudwhg dffruglqj wr Iljxuh 518
dqg fdq eh ghvfulehg dv
{n @ xn . sn > +514<,
zkhuh sn lv wkh nqrzq slorw vhtxhqfh zlwk hqhuj| Hv @ H^msn m5 `Wv dqg xn lv
wkh xqnqrzq gdwd sduw zlwk hqhuj| +4 , Hv @ H^mxn m5 `Wv 1 Wkh srzhu ri wkh
slorw vhtxhqfh lv rqo| d vpdoo iudfwlrq ri wkh wudqvplwwhg srzhu/ exw wkh slorw
vhtxhqfh fdq dq|krz eh xvhg iru fkdqqho hvwlpdwlrq ru wlph dqg iuhtxhqf|
v|qfkurql}dwlrq1 Wkh slorw vhtxhqfh fdq eh wuhdwhg hlwkhu dv h{wud qrlvh ru dv
d nqrzq glvsodfhphqw ri wkh frqvwhoodwlrq1 Lq wkh odwwhu fdvh wkh glvwxuedqfh
rq wkh gdwd vhtxhqfh fdq eh pdgh qhjoljleoh/ exw rwkhuzlvh wkh glvwruwlrq rq
wkh gdwd sduw lv dq|krz vpdoo1
Wkh lghd ri vxshulpsrvlqj d slorw vhtxhqfh kdv ehhq sursrvhg e| ydulrxv
dxwkruv iru glhuhqw dssolfdwlrqv1 Kroghq dqg Ihkhu sursrvhg lq ^67` d v|v0
whp iru fduulhu uhfryhu| edvhg rq vxshulpsrvhg slorw vhtxhqfhv1 Wkh| fdoohg
wkh whfkqltxh vsuhdg0vshfwuxp slorw whfkqltxh +VVSW, vlqfh wkh slorw vhtxhqfh
xvhg zdv d SQ0vhtxhqfh/ zklfk zdv olqhduo| dgghg wr wkh lqskdvh gdwd fkdq0
qho1 E| pxowlso|lqj wkh lqfrplqj vljqdo zlwk wkh nqrzq SQ0vhtxhqfh lw lv
srvvleoh wr uhfryhu wkh fduulhu vljqdo lq dq hdv| zd|1 Diwhu gh0vsuhdglqj/ wkh
fduulhu vljqdo lv qduurzedqg dqg wkhuhiruh pxfk ri wkh glvwruwlrq iurp wkh gdwd
vljqdo fdq eh owhuhg rxw1 Frpsduh wklv wr wkh surfhvvlqj jdlq dfklhyhg lq
FGPD v|vwhpv1 Wkh phwkrg vkrzhg wr eh urexvw iru orz VQUv/ wkh v|vwhp
zdv lpsohphqwhg lq kdugzduh dqg wkh fduulhu uhfryhu| xqlw zrunhg zhoo grzq
wr He @Q3 @ 3 gE1 Ehmmdql dqg Ehoruh sursrvhg d v|vwhp xvlqj d yduldqw ri vx0
shulpsrvhg slorw vhtxhqfhv ^43` iru xqghuzdwhu frppxqlfdwlrq1 Wkh jrdo wkhuh
zdv wr hqdeoh wudfnlqj ri d udslgo| fkdqjlqj fkdqqho e| lqwurgxflqj d frqwlq0
xrxv slorw vhtxhqfh1 Lq wkh v|vwhp/ wzr glhuhqw ruwkrjrqdo sxovh vkdshv zhuh
xvhg iru wkh gdwd dqg slorw vhtxhqfh1 Wkhuhiruh/ wkhuh zdv d orvv lq vshfwudo
h!flhqf|/ exw wkh v|vwhp vkrzhg wr eh urexvw iru kljk Grssohu iuhtxhqflhv1
Wklv whfkqltxh lv vlplodu wr wkh XPWV2Z0FGPD v|vwhp lq xsolqn/ zkhuh wkh
gdwd dqg slorw vhtxhqfhv duh wudqvplwwhg vlpxowdqhrxvo| ^68`/ exw xvlqj wkh
ruwkrjrqdo lqskdvh dqg txdgudwxuh fkdqqhov/ uhvshfwlyho|1 Erwk wkhvh phwk0
rgv phdq wkdw wkh vshfwudo h!flhqf| lv ghfuhdvhg vlqfh qr gdwd lv wudqvplwwhg
rq wkh slorw fkdqqho1 Vxshulpsrvhg slorw vhtxhqfhv kdyh dovr ehhq xvhg iru
iudph v|qfkurql}dwlrq ^9:`1 Wkh dxwkruv xvhg d qrq0ruwkrjrqdo slorw vhtxhqfh
wr ghwhuplqh wkh vwduw ri iudphv lq d zluhohvv v|vwhp1 Wkh slorw vhtxhqfh zdv
qrw xvhg iru elw v|qfkurql}dwlrq dqg wkhuhiruh wkh lq xhqfh ri wkh gdwd frxog
eh uhpryhg ehiruh fruuhodwlqj wkh uhfhlyhg slorw vhtxhqfh zlwk wkh nqrzq slorw
vhtxhqfh1 Wkhq/ zkhq iudph v|qfkurql}dwlrq zdv dfklhyhg/ wkh lq xhqfh rq
wkh gdwd v|perov zdv uhpryhg e| vxewudfwlqj wkh slorw vhtxhqfh1 Wklv whfk0
qltxh dovr vkrzhg jrrg v|qfkurql}dwlrq shuirupdqfh frpsduhg wr frqyhqwlrqdo
v|vwhpv/ hvshfldoo| lq idvw Ud|ohljk idglqj fkdqqhov zkhuh wkh orqj frqwlqxrxv
slorw vhtxhqfh phdqv wkdw wlph glyhuvlw| lv dfklhyhg1
Lq sdshu ^H` vxshulpsrvhg slorw vhtxhqfhv duh xvhg iru wlph dqg iuhtxhqf|
v|qfkurql}dwlrq lq RIGP v|vwhpv dqg lq sdshu Î` wkh whfkqltxh lv xvhg iru
mrlqw fkdqqho hvwlpdwlrq dqg ghwhfwlrq lq dw Ud|ohljk idglqj fkdqqhov1 Wkh
odwwhu sdshu lv qrw uhvwulfwhg wr pxowl0fduulhu v|vwhpv/ exw fdq eh xvhg iru dq|
v|vwhp zkhuh wkh wlph glvshuvlrq lv qhjoljleoh1 Wkh v|qfkurql}hu vwuxfwxuh lq
^H` kdv orz frpsoh{lw| dqg lv vxlwdeoh iru lpsohphqwdwlrq exw wkh frpsoh{lw|
ri wkh fkdqqho hvwlpdwru lq Î` lv txlwh kljk dqg wkhuhiruh orz0frpsoh{lw| ds0
sur{lpdwlrqv duh ri lqwhuhvw1 Wkh vfrsh ri Î` kdv ehhq wr vkrz wkh srwhqwldo
ri xvlqj vxshulpsrvhg slorw vhtxhqfhv iru fkdqqho hvwlpdwlrq dqg wr ghulyh dq
rswlpdo uhfhlyhu vwuxfwxuh1 Wkh sdshu krzhyhu dovr dgguhvvhv frpsoh{lw| lvvxhv
dqg srvvleoh vlpsolhg uhfhlyhu vwuxfwxuhv1
51614 Wlph dqg Iuhtxhqf| V|qfkurql}dwlrq iru RIGP

Vxshulpsrvhg slorw vhtxhqfhv duh xvhg lq ^H` iru wlph dqg iuhtxhqf| v|qfkur0
ql}dwlrq lq RIGP v|vwhpv1 Wkh dgydqwdjh ri xvlqj vxshulpsrvhg slorwv iru
wklv sxusrvh lv wkdw d frqwlqxrxv vhtxhqfh lv suhvhqw wkdw fdq eh xvhg iru
v|qfkurql}dwlrq sxusrvhv1 Wklv phdqv wkdw wkh iuhtxhqf|0rvhw udqjh fdq eh
lqfuhdvhg frpsduhg wr phwkrgv zkhuh wkh f|folf suh{ lv xvhg iru v|qfkurql}d0
wlrq1 Ixuwkhu/ wkh wlph v|qfkurql}dwlrq vljqdo vkrzv d pruh dwwudfwlyh ehkdylru
vlqfh lw kdv d vkdushu shdn frpsduhg wr wkh phwkrgv lqfrusrudwlqj d f|folf suh0
5161 VXSHULPSRVHG SLORW VHTXHQFHV 64
{1 Wkh whfkqltxh lv vxlwdeoh iru erwk wudfnlqj dqg dftxlvlwlrq dqg lw lv yhu|
h{leoh lq wkdw wkh srzhu ri wkh slorw vhtxhqfh fdq eh dgmxvwhg wr wkh vshflf
qhhgv1 Lq dftxlvlwlrq prgh doo wkh wudqvplwwhg srzhu fdq eh ghyrwhg wr wkh
slorw vhtxhqfh dqg wkhq wkh wudqvplwwhg vljqdo lv htxlydohqw wr wkh suhdpeoh
ghvfulehg lq ^G`1 E| xvlqj doo wkh srzhu iru wkh slorw vhtxhqfh gxulqj dftxlvl0
wlrq/ wkh dftxlvlwlrq wlph fdq eh pdgh vpdoo1 Gxulqj wudfnlqj/ wkh slorw srzhu
fdq eh orz zlwkrxw orrvlqj shuirupdqfh vlqfh wkh sdudphwhuv gr qrw fkdqjh
wkdw idvw dqg dyhudjlqj ryhu vhyhudo v|perov fdq eh xvhg1
Wkh v|qfkurql}dwlrq vljqdo lv fdofxodwhg lq wkh vdph zd| dv iru wkh suh0
fruuhodwlqj v|qfkurql}hu dffruglqj wr +5149, dqg Iljxuh 5161 Wkh uhfhlyhg vljqdo
lv uvw fruuhodwhg zlwk wkh nqrzq vhtxhqfh ri ohqjwk U= Wkhq wkh fruuhodwhg
ydoxhv duh pxowlsolhg lq sdluv dqg qdoo| O ri wkhvh surgxfwv duh vxpphg lq
rughu wr jhw wkh v|qfkurql}dwlrq vljqdo1 Wkh rqo| glhuhqfh frpsduhg wr wkh
suhdpeoh v|qfkurql}dwlrq vfkhph lv wkdw wkh gdwd vhtxhqfh lv wudqvplwwhg dw
wkh vdph wlph dv wkh v|qfkurql}dwlrq vhtxhqfh dqg wkhuhiruh dfwv olnh dq dggl0
wlrqdo glvwxuedqfh1 Wkhuhiruh/ wkh fruuhodwru ohqjwk ru wkh qxpehu ri surgxfwv
xvhg iru wkh v|qfkurql}dwlrq vljqdo kdv wr eh lqfuhdvhg frpsduhg wr wkh fdvh
zlwk wkh v|qfkurql}dwlrq vljqdo rqo|/ l1h1 wkh suhdpeoh1 Wkh dpsolwxgh ri wkh
v|qfkurql}dwlrq vljqdo lv xvhg iru wlph v|qfkurql}dwlrq dqg zkhq wkh fruuhfw
wlplqj lv irxqg wkh iuhtxhqf|0rvhw hvwlpdwh lv jlyhq e| wkh skdvh dqjoh ri wkh
v|qfkurql}dwlrq vljqdo1
Lq wkh olwhudwxuh qr vlplodu zrun fdq eh irxqg iru RIGP v|vwhpv/ dw ohvw dv
idu dv L nqrz1 Wkh phwkrg kdv vrph vlplodulwlhv wr v|qfkurql}dwlrq lq FGPD
v|vwhpv1 Riwhq lq FGPD v|vwhpv/ iru h{dpsoh lq wkh XPWV2Z0FGPD v|v0
whp/ wkhuh lv d vshfldo v|qfkurql}dwlrq fkdqqho wudqvplwwhg rq wrs ri wkh gdwd
fkdqqhov1 Vlqfh wkh v|qfkurql}dwlrq fkdqqho fdq eh vkduhg ehwzhhq wkh xvhuv/
wkh ryhukhdg lv vpdoo dq|zd|1 Krzhyhu/ dv rssrvhg wr wkh vxshulpsrvhg slorw
vhtxhqfhv/ wkh v|qfkurql}dwlrq fkdqqho lq wkh FGPD v|vwhp lv ruwkrjrqdo wr
wkh gdwd fkdqqhov e| phdqv ri glhuhqw frghv1 Wklv phdqv vrphzkdw ehwwhu
ghwhfwlrq surshuwlhv/ vlqfh wkh v|qfkurql}dwlrq vfkhph edvhg rq vxshulpsrvhg
slorw vhtxhqfhv uholhv rq vwdwlvwlfdo dyhudjlqj iru ghwhfwlrq ri wkh v|qfkurql}d0
wlrq vljqdo1
Lq wkh vsuhdg0vshfwuxp slorw v|vwhp lq ^67` vxshusrvlwlrq ri wkh slorw vh0
txhqfh dqg gdwd vhtxhqfh zdv pdgh lq wkh dqdorj grpdlq1 Wklv phdqv wkdw
iru wklv v|vwhp elw v|qfkurqrxv wudqvplvvlrq frxog qrw eh jxdudqwhhg1 Wklv lv
lq frqwudvw wr wkh vxshulpsrvhg slorw vhtxhqfhv lq ^H`/ zkhuh wklv surshuw| lv
hvvhqwldo iru wlph v|qfkurql}dwlrq dqg odwhu iru fkdqqho hvwlpdwlrq lq Î`1
51615 Fkdqqho Hvwlpdwlrq xvlqj Vxshulpsrvhg Slorw Vh0

txhqfhv
Wkh vxshulpsrvhg slorw vhtxhqfhv fdq dovr eh xvhg iru fkdqqho hvwlpdwlrq dqg
ghwhfwlrq1 Gxh wr wkh frqwlqxrxv wudqvplvvlrq/ wkh whfkqltxh lv dovr yhu| xvhixo
lq idvw idglqj frqglwlrqv1 Slorw0edvhg v|vwhpv idlo zkhq wkh vdpsolqj wkhruhp lv
qrw ixooohg/ dqg iru frqyhqwlrqdo v|vwhpv wklv ehfrphv dq lpsruwdqw olplwdwlrq
dw kljk Grssohu iuhtxhqflhv1 Glhuhqw dssurdfkhv fdq eh xvhg iru ghwhfwlrq
edvhg rq vxshulpsrvhg slorw vhtxhqfhv1 Wkh slorw vhtxhqfh fdq eh vhhq dv d
slorw wrqh wkdw lv vsuhdg ryhu wkh zkroh wudqvplvvlrq edqgzlgwk ru dv d vpdoo
vkliw ri wkh vljqdo frqvwhoodwlrq1 Lq Î`/ d v|vwhp edvhg rq mrlqw pd{lpxp
olnholkrrg ghwhfwlrq ri wkh gdwd dqg wkh fkdqqho lv ghvfulehg1 Wklv phdqv wkdw
wkh slorw vhtxhqfh dqg wkh gdwd vhtxhqfh erwk duh xvhg iru fkdqqho hvwlpdwlrq
dqg ghwhfwlrq1 Rwkhu srvvleoh dssurdfkhv zrxog eh wr uvw hvwlpdwh wkh fkdqqho
e| khos ri wkh slorw vhtxhqfh dqg wkhq ghwhfw wkh gdwd edvhg rq wklv hvwlpdwh1
Lq ^67` wkh v|qfkurql}dwlrq vljqdo zdv h{wudfwhg dqg diwhu wkdw wkh glvwxuedqfh
ri wkh gdwd vhtxhqfh zdv ghfuhdvhg e| fkdqjlqj wkh ghwhfwlrq wkuhvkrog zlwk
uhjdugv wr wkh slorw vhtxhqfh1 Lq ^9:` wkh gdwd vljqdo zdv uvw ghwhfwhg dqg lwv
lq xhqfh rq wkh v|qfkurql}dwlrq vljqdo zdv uhpryhg ehiruh h{wudfwlrq ri wkh
v|qfkurql}dwlrq vljqdo1 Krzhyhu/ wkh odwwhu lv qrw srvvleoh iru gdwd ghwhfwlrq
dqg fkdqqho hvwlpdwlrq vlqfh wkh lq xhqfh ri wkh gdwd fdq qrw eh uhpryhg
ehiruh h{wudfwlrq ri wkh v|qfkurql}dwlrq sdudphwhuv dqg ylfh yhuvd1
Mrlqw fkdqqho hvwlpdwlrq dqg ghwhfwlrq lv zhoo ghvfulehg lq ^7;`1 Wkh dlp lv
wr qg wkh gdwd vhtxhqfh dqg fkdqqho hvwlpdwh wkdw pd{lpl}hv wkh suredelolw|
ghqvlw| ixqfwlrq ri wkh uhfhlyhg vhtxhqfh dffruglqj wr +514,1 Dvvxph wkdw wkh
uhfhlyhg vljqdo vdpsoh dw wlph n fdq eh ghvfulehg e|
un @ kn fn . qn > +5153,
zkhuh wkh qrlvh qn dqg pxowlsolfdwlyh fkdqqho fn duh mrlqwo| Jdxvvldq yduldeohv
dqg wkdw wkh gdwd v|perov duh lqghshqghqw udqgrp yduldeohv zlwk htxdo sure0
delolwlhv1 Lw fdq eh vkrzq wkdw ^7;` wkh rswlpdo mrlqw hvwlpdwlrq0ghwhfwlrq uxoh
fdq eh irupxodwhg dv
fPO > k,
+a a @ duj pd{ i +u> kmf, @ duj pd{ i +kmu> f,i+umf,= +5154,
f>k f>k
D vxlwdeoh vwudwhj| lv wkhuhiruh wr pd{lpl}h i +kmu> f, wr qg wkh PDS +pd{l0
pxp d srvwhulrul, fkdqqho hvwlpdwh iru hdfk srvvleoh gdwd vhtxhqfh/ f/ dqg xvh
wklv dv d wuxh sdudphwhu zkhq vhohfwlqj wkh vhtxhqfh zlwk wkh odujhvw olnholkrrg1
Vlqfh wkh fkdqqho dqg qrlvh yduldeohv duh dvvxphg wr eh Jdxvvldq/ wkh PDS
hvwlpdwh lv htxdo wr wkh plqlpxp phdq vtxduh huuru hvwlpdwh/ zklfk lq wxuq
fdq eh fdofxodwhg xvlqj d Zlhqhu owhu ^73`1 Wklv lv wkh dssurdfk lq Î`1 Iru
hdfk k|srwkhvlv ri wkh gdwd vhtxhqfh wkh fkdqqho hvwlpdwh lv fdofxodwhg xvlqj d
Zlhqhu owhu1 Wklv fkdqqho hvwlpdwh lv wkhq xvhg wr fdofxodwh wkh phwulf vxfk
wkdw wkh prvw suredeoh vhtxhqfh fdq eh irxqg1 Lq Î` erwk dq rswlpdo dqg dq
dssur{lpdwh phwulf lv ghulyhg1 Wkh xvh ri wkh rswlpdo phwulf uhtxluhv wkdw doo
srvvleoh gdwd vhtxhqfhv duh wdnhq lqwr dffrxqw zkhq orrnlqj iru wkh vhtxhqfh
uhvxowlqj lq wkh orzhvw phwulf/ zkhuhdv wkh dssur{lpdwh phwulf lv fdq eh xvhg
iru d ohvv frpsoh{ ghwhfwru e| xvlqj shu0vxuylyru surfhvvlqj ^8:`1
Lq ^J` wkhuh lv d frpsdulvrq ehwzhhq fkdqqho hvwlpdwlrq dqg ghwhfwlrq xvlqj
vxshulpsrvhg slorw vhtxhqfhv dqg rwkhu ghwhfwlrq dqg hvwlpdwlrq phwkrgv1 Wkh
5161 VXSHULPSRVHG SLORW VHTXHQFHV 66
rwkhu dqdo|}hg phwkrgv lqfoxgh SVDP/ pxowlsoh0v|pero glhuhqwldo ghwhfwlrq/

d qryho k|eulg ghwhfwlrq phwkrg lqwurgxfhg lq ^J` dqg glhuhqwldo ghwhfwlrq1
Wkh dqdo|vlv lv edvhg rq wkh fdofxodwlrq ri wkh sdluzlvh huuru suredelolw| wkdw
wkh ghflvlrq lv pdgh lq idyru ri d vshflf huurqhrxv gdwd vhtxhqfh lqvwhdg ri
wkh fruuhfw rqh1 Wkh sdluzlvh suredelolw| kdv suhylrxvo| ehhq xvhg wr dqdo|}h
WFP ^49` dqg pxowlsoh0v|pero glhuhqwldo ghwhfwlrq ^5<`1 Wkh phwkrg lv yhu|
xvhixo zkhq wkh ghflvlrqv duh ghshqghqw ehwzhhq vxffhvvlyh v|perov1 Vxfk vlw0
xdwlrqv rffxu/ iru h{dpsoh/ zkhq xvlqj WFP ru vxshulpsrvhg slorw vhtxhqfhv
iru fkdqqho hvwlpdwlrq dv ghvfulehg deryh1 Lq wkhvh vlwxdwlrqv/ lw lv lpsrvvleoh
ru wulfn| wr fdofxodwh h{dfw h{suhvvlrqv iru wkh EHU/ exw e| h{suhvvlqj wkh gh0
flvlrq yduldeoh lq d txdgudwlf irup ri Jdxvvldq yduldeohv lw lv srvvleoh wr ghulyh
h{suhvvlrqv iru wkh sdluzlvh huuru suredelolw| dqg dssur{lpdwh h{suhvvlrqv iru
wkh EHU1
Fkdswhu 6
Vxppdu| dqg
Frqwulexwlrqv
Lq wklv fkdswhu/ vrph jhqhudo uhvxowv duh suhvhqwhg dqg wkh frqwhqwv dqg pdlq
uhvxowv lq hdfk sdshu duh ghvfulehg1 Wkhuh lv dovr d suhvhqwdwlrq ri wkh frq0
wulexwlrqv lq hdfk sdshu wr wkh vshflf uhvhdufk hog1 Wkhuhdiwhu/ wkhuh lv d
glvfxvvlrq ri srvvleoh gluhfwlrqv iru ixwxuh zluhohvv frppxqlfdwlrq v|vwhpv1
Ilqdoo|/ vrph jhqhudo frqfoxvlrqv duh suhvhqwhg dqg srvvleoh ixuwkhu zrun lq
wkh uhvhdufk hogv lv glvfxvvhg1
614 Jhqhudo Uhvxowv

Ixwxuh frppxqlfdwlrq v|vwhpv kdyh wr surylgh kljk gdwd udwhv lq kljko| glv0
wxuehg hqylurqphqwv1 Wkh ghpdqg iru wkh vhuylfhv lv h{shfwhg wr eh kljk1
Wkhuhiruh qhz iuhtxhqf| edqgv kdyh wr eh xvhg dqg/ gxh wr wkh odfn ri edqg0
zlgwk/ wkhvh zloo eh orfdwhg dw kljkhu iuhtxhqflhv1 Wkhvh wklqjv ohdg wr lq0
fuhdvhg lpsruwdqfh ri v|qfkurql}dwlrq dqg fkdqqho hvwlpdwlrq lvvxhv lq wkh
ixwxuh1 Wkh xvh ri kljkhu iuhtxhqf| edqgv phdqv wkdw wkh Grssohu iuhtxhqf| lv
lqfuhdvhg iru d fhuwdlq yhorflw| ri wkh preloh whuplqdo1 Wkh pdq| xvhuv phdq
wkdw wkh ryhudoo lqwhuihuhqfh ohyho zloo lqfuhdvh1 Wklv pdnhv lw qhfhvvdu| wr eh
deoh wr rshudwh lq kljko| glvwxuehg hqylurqphqwv1 Lq dgglwlrq/ wkh uhtxluh0
phqwv ri vshfwudo h!flhqf| fdoo iru wkh xvh ri kljkhu rughu prgxodwlrq vfkhphv/
dqg wkhvh duh pruh vhqvlwlyh wr v|qfkurql}dwlrq dqg fkdqqho hvwlpdwlrq huuruv1
Lq prvw wh{werrnv rq gljlwdo frppxqlfdwlrqv/ wkh v|qfkurql}dwlrq lvvxhv duh/
krzhyhu/ qhjohfwhg dqg v|qfkurql}dwlrq sdudphwhuv duh riwhq surylghg e| d
jhqlh1 Lq uhdo v|vwhpv qr jhqlh h{lvwv dqg wkh v|qfkurql}huv dqg fkdqqho hvwl0
pdwruv kdyh d odujh lq xhqfh rq wkh shuirupdqfh1 Wkh jrdo ri wkh uhvhdufk kdv
ehhq wr vwxg| wkhvh nh| frpsrqhqwv dqg vxjjhvw lpsuryhphqwv vxfk wkdw wkh
uhtxluhphqwv rq ixwxuh v|vwhpv fdq eh phw1
68
69 FKDSWHU 61 VXPPDU\ DQG FRQWULEXWLRQV
Wkh irfxv ri wkh wkhvlv lv sureohpv uhodwhg wr wkh zluhohvv preloh fkdqqho1
Wklv fkdqqho kdv vrph lpsruwdqw surshuwlhv wkdw fdoo iru vshfldo dwwhqwlrq> wkh
fkdqqho fkdqjhv udslgo| dqg lw kdv odujh dpsolwxgh xfwxdwlrqv1 Wkhuhiruh/
wkh v|vwhp kdv wr eh rswlpl}hg dffruglqj wr wkhvh vshflf surshuwlhv1 Wkh
wdvn ri wkh v|qfkurql}hu dqg fkdqqho hvwlpdwru lv wr hqdeoh frppxqlfdwlrq
dovr xqghu vhyhuh frqglwlrqv1 Lq wkh wkhvlv/ zh kdyh wkhuhiruh vwxglhg urexvw
v|qfkurql}dwlrq dqg fkdqqho hvwlpdwlrq whfkqltxhv/ vxlwdeoh dovr iru orz VQUv
dqg idvw idglqj fkdqqhov1 Wkh zrun kdv ehhq irfxvhg rq ixwxuh v|vwhpv dqg
v|vwhpv xqghu vwdqgdugl}dwlrq1 Krzhyhu/ wkh sulqflsohv duh wkh vdph iru h{lvw0
lqj v|vwhpv dqg wkh dqdo|vhv fdq eh dssolhg wr wkhvh dv zhoo1 Lq wkh wkhvlv wkh
lq xhqfh ri wkh slorw sdwwhuq lq RIGP v|vwhpv lv vwxglhg dqg jxlgholqhv iru
wkh ghvljq duh jlyhq1 V|qfkurql}dwlrq lq RIGP v|vwhpv edvhg rq suhdpeohv lv
dovr dqdo|}hg dqg d qhz v|qfkurql}hu vwuxfwxuh lv sursrvhg1 Ixuwkhu/ wkh vxe0
fkdqqho edqgzlgwk lq zluhohvv RIGP v|vwhpv lv rswlpl}hg zlwk uhjdugv wr wkh
fkdudfwhulvwlfv ri wkh fkdqqho1 Wkh xvh ri suh0frpshqvdwlrq lv dovr lqyhvwljdwhg
e| ghulylqj h{suhvvlrqv iru wkh EHU dqg ghwhuplqlqj ihdvleoh eorfn ohqjwkv1
Ilqdoo|/ wkh xvh ri vxshulpsrvhg slorw vhtxhqfhv lv dqdo|}hg dqg sursrvhg iru
v|qfkurql}dwlrq dqg fkdqqho hvwlpdwlrq lvvxhv1
Wkh uhtxluhphqwv rq wkh v|qfkurql}hu dqg fkdqqho hvwlpdwru duh qrw vwdwlf
ryhu d orqjhu wlph1 Wkh| fkdqjh zkhq qhz whfkqrorjlhv dqg whfkqltxhv duh
lqwurgxfhg1 Lpsuryhg vhplfrqgxfwru whfkqrorj| kdv phdqw wkdw pruh frpsoh{
dojrulwkpv wkdq ehiruh fdq eh xvhg1 Wkh qhz dojrulwkpv/ xvhg dv dq h{dpsoh
iru ghwhfwlrq ru huuru fruuhfwlrq/ duh riwhq pruh dffxudwh wkdq wkh suhylrxvo|
xvhg rqhv/ exw dovr pruh vhqvlwlyh wr wkh dffxudf| ri wkh v|qfkurql}dwlrq sd0
udphwhuv1 Huuruv wkdw suhylrxvo| zhuh klgghq gxh wr d olplwhg uhvroxwlrq dqg
glg qrw dhfw wkh uhvxow fdq sod| dq lpsruwdqw uroh lq qhz v|vwhpv zlwk ehwwhu
uhvroxwlrq1 Wkh jrdo lv wkdw wkh vxssruwlqj ixqfwlrqv vkrxog qrw eh d olplwd0
wlrq ri wkh v|vwhp shuirupdqfh1 Frpsoh{lw| lv krzhyhu vwloo dq lpsruwdqw lvvxh
iru wkh dojrulwkpv/ vlqfh wkh frpsoh{lw| vkrxog eh vshqw rq ixqfwlrqv surylglqj
wkh odujhvw jdlq1
615 Sdshu Frqwulexwlrqv

61514 Slorw Dvvlvwhg Fkdqqho Hvwlpdwlrq iru RIGP lq
Preloh Fhooxodu V|vwhpv
Wklv sdshu frqwdlqv dq dqdo|vlv ri wkh lq xhqfh ri wkh slorw sdwwhuq rq wkh shu0
irupdqfh ri d preloh RIGP v|vwhp1 Wkh elw huuru shuirupdqfh lv fdofxodwhg
iru yh glhuhqw slorw sdwwhuqv edvhg rq wkh phdq vtxduh huuru ri wkh fkdqqho
hvwlpdwh1 Lq wkh sdshu dq dxwruhjuhvvlyh surfhvv lv dvvxphg iru wkh fkdqjhv ri
wkh Ud|ohljk idglqj fkdqqho dqg d Ndopdq owhu lv xvhg iru fkdqqho hvwlpdwlrq1
Wkh hvwlpdwlrq owhu lv dovr xvhg wr fdofxodwh wkh rswlpxp slorw sdwwhuq lq wkh
vhqvh ri plqlpxp xqfrghg EHU iru hdfk ri wkh RIGP v|perov1 Wkh shuiru0
pdqfh iurp wklv slorw sdwwhuq lv frpsduhg wr wkh shuirupdqfh zkhq xvlqj rwkhu
6151 SDSHU FRQWULEXWLRQV 6:
slorw sdwwhuqv irxqg lq wkh olwhudwxuh1 Lw lv vkrzq wkdw wkh slorw sdwwhuq kdv
d odujh lq xhqfh rq wkh uhvxowlqj elw huuru udwh1 D zhoo0ghvljqhg slorw sdwwhuq
phdqv wkdw wkh fkdqqho fkdqjhv fdq eh wudfnhg ehwwhu/ zlwk d fruuhvsrqglqj
shuirupdqfh lpsuryhphqw1 Lq d qxphulfdo h{dpsoh lw lv vkrzq wkdw wkh EHU
fdq eh ghfuhdvhg e| d idfwru ri 8 ru wkdw wkh v|vwhp fdq frsh zlwk 43 wlphv
kljkhu Grssohu iuhtxhqf| zkhq xvlqj d vxlwdeoh slorw sdwwhuq lqvwhdg ri rqh
vlplodu wr d vlqjoh0fduulhu fdvh1
Wkh pdlq frqwulexwlrq ri wkh sdshu lv wr vkrz wkh lpsruwdqfh ri d zhoo0
ghvljqhg slorw sdwwhuq dqg wkh odujh lpsuryhphqwv wkdw fdq eh dfklhyhg1 Wkh
sdshu lv rqh ri wkh uvw sdshuv xvlqj dq dqdo|wlfdo dssurdfk wr rswlpl}h wkh
slorw sdwwhuq1 Wkh dqdo|vlv lv jhqhudo dqg fdq eh xvhg wr ghulyh wkh dprxqw ri
slorw v|perov uhtxluhg dqg wr qg d vxlwdeoh slorw sdwwhuq iru dq| RIGP0edvhg
v|vwhp1
61515 Rswlpl}dwlrq ri Vxe0Fkdqqho Edqgzlgwk iru Pr0

eloh RIGP V|vwhpv
Lq wklv sdshu wkh edqgzlgwk ri wkh vxe0fkdqqhov lv rswlpl}hg zlwk uhvshfw

wr lqwhufkdqqho lqwhuihuhqfh +LFL, fdxvhg e| wkh wlph0vhohfwlyh fkdqqho dqg wkh
hqhuj| orvv fdxvhg e| wkh f|folf suh{1 Iru d jlyhq gxudwlrq ri wkh f|folf suh{/
wkh hqhuj| orvv lqfuhdvhv dv wkh vxe0fkdqqho edqgzlgwk lqfuhdvhv1 Rq wkh rwkhu
kdqg/ wkh LFL fdxvhg e| fkdqqho fkdqjhv gxulqj dq RIGP v|pero lqfuhdvhv dv
wkh vxe0fkdqqho edqgzlgwk ghfuhdvhv1 Dq h{suhvvlrq iru wkh uhvxowlqj xqfrghg
EHU dv d ixqfwlrq ri wkh vxe0fkdqqho edqgzlgwk lv wkhuhiruh ghulyhg/ lqfoxglqj
wkhvh hhfwv1 Wkh lq xhqfhv ri wkh Grssohu iuhtxhqf|/ VQU/ dqg wrwdo gdwd udwh
duh lqyhvwljdwhg1 Lq dgglwlrq/ wkh hhfwv ri lqwurgxflqj fkdqqho hvwlpdwlrq duh
dqdo|}hg zlwk uhvshfw wr wkh fkrlfh ri vxe0fkdqqho edqgzlgwk1 Lw lv frqfoxghg
wkdw wkh wrwdo gdwd udwh kdv yluwxdoo| qr hhfw rq wkh rswlpxp dqg wkdw wkh
rswlpl}dwlrq fdq eh shuiruphg lqghshqghqw ri fkdqqho hvwlpdwlrq1 Lw lv vkrzq
wkdw wkh rswlpdo vxe0fkdqqho edqgzlgwk ghshqgv rq wkh Grssohu iuhtxhqf|
dqg VQU1 Kljkhu Grssohu/ dv zhoo dv dq lqfuhdvhg VQU/ fdoo iru dq lqfuhdvhg
vxe0fkdqqho edqgzlgwk wr dyrlg wkdw wkh LFL ehfrphv wkh grplqdqw vrxufh ri
huuruv1 Ilqdoo|/ lw lv dujxhg wkdw d zruvw0fdvh ghvljq zlwk uhvshfw wr wkh Grssohu
iuhtxhqf| vkrxog eh shuiruphg1
Wkh pdlq frqwulexwlrq ri wkh sdshu lv wkh dqdo|wlfdo rswlpl}dwlrq ri wkh
vxe0fkdqqho edqgzlgwk1 Wkhuh duh qrw pdq| sdshuv dgguhvvlqj wkh fkrlfh ri
wkh qxpehu ri vxe0fkdqqhov lq RIGP v|vwhpv/ dqg dw wkh wlph ri sxeolfdwlrq
wkh fkrlfh kdg suhylrxvo| ehhq edvhg rqo| rq vlpxodwlrqv1 Lq dgglwlrq/ wkh
lq xhqfh ri fkdqqho hvwlpdwlrq lv lqyhvwljdwhg dqg lw lv frqfoxghg wkdw wkh
rswlpl}dwlrq fdq eh shuiruphg lqghshqghqw ri fkdqqho hvwlpdwlrq dqg rq d
zruvw0fdvh dvvxpswlrq ri wkh Grssohu iuhtxhqf|1
6; FKDSWHU 61 VXPPDU\ DQG FRQWULEXWLRQV
61516 Suh0Frpshqvdwlrq iru Ud|ohljk Idglqj Fkdqqhov lq

Wlph Glylvlrq Gxsoh{ RIGP V|vwhpv
Wklv sdshu lqyhvwljdwhv wkh srvvlelolw| ri xvlqj suh0frpshqvdwlrq iru frpshq0
vdwlqj wkh lq xhqfh ri wkh Ud|ohljk idglqj fkdqqho lq d zluhohvv RIGP wlph
glylvlrq gxsoh{ v|vwhp1 Wkuhh frpshqvdwlrq phwkrgv duh dqdo|}hg=
41 skdvh0rqo| frpshqvdwlrq
51 skdvh dqg dpsolwxgh frpshqvdwlrq xs wr d fhuwdlq srzhu ohyho/ deryh

zklfk wkh dpsolhu lv vdwxudwhg
61 skdvh dqg dpsolwxgh frpshqvdwlrq xs wr d fhuwdlq srzhu ohyho/ deryh

zklfk wkh dpsolhu lv vzlwfkhg r
Wkh EHUv dqg lqfuhdvh lq shdn0wr0dyhudjh srzhu udwlrv duh fdofxodwhg iru wkh
glhuhqw phwkrgv dvvxplqj shuihfw fkdqqho nqrzohgjh1 Wkh pd{lpxp eorfn
ohqjwkv duh fdofxodwhg iru wkh skdvh frpshqvdwhg v|vwhp/ vlqfh wklv lv dvvxphg
prvw dssursuldwh iru zluhohvv dssolfdwlrqv1 Lw lv frqfoxghg wkdw eorfn ohqjwkv
pxfk orqjhu wkdq wkh rqh lq wkh GHFW v|vwhp fdq eh xvhg ehiruh wkh lqvwdq0
wdqhrxv shuirupdqfh jhwv zruvh wkdq wkdw ri d glhuhqwldo v|vwhp1
Wkh pdlq frqwulexwlrqv duh wkh ghulydwlrq ri wkh uhvxowlqj xqfrghg EHU
iru wkh skdvh0rqo| frpshqvdwlrq phwkrg dqg wkh dqdo|vlv ri wkh lq xhqfh ri
erwk wkh xsolqn eorfn ohqjwk dqg wkh grzqolqn eorfn ohqjwk/ uhvshfwlyho|1 Wkh
fdofxodwlrqv duh shuiruphg erwk iru shuihfw nqrzohgjh ri wkh fkdqqho fruuhodwlrq
ixqfwlrqv dqg xqghu fruuhodwlrq plvpdwfk1 Wkh pd{lpxp eorfn ohqjwk lv d yhu|
lpsruwdqw ghvljq sdudphwhu lq wkh ghvljq ri suh0frpshqvdwhg wlph glylvlrq
gxsoh{ v|vwhpv dqg wr p| nqrzohgjh qr dqdo|wlfdo dqdo|vlv ri lwv lq xhqfh rq
wkh shuirupdqfh h{lvwv lq wkh olwhudwxuh1 Lq rwkhu sdshuv wkh fkdqqho lv dvvxphg
wr eh frqvwdqw ru vlpxodwlrqv duh xvhg wr lqyhvwljdwh wkh lq xhqfh ri eorfn
ohqjwk dqg Grssohu iuhtxhqf|1 Rwkhu frqwulexwlrqv lq wklv sdshu lqfoxgh wkh
ghulydwlrq ri wkh xqfrghg EHU iru wkh wkuhh phwkrgv zkhq dvvxplqj shuihfw
fkdqqho nqrzohgjh dqg wkh fruuhvsrqglqj lqfuhdvh lq shdn0wr0dyhudjh srzhu
udwlr1
61517 Suhdpeoh0Edvhg Wlph dqg Iuhtxhqf| V|qfkurql}d0

wlrq iru RIGP V|vwhpv
Wklv sdshu frpsduhv wkh v|qfkurql}dwlrq shuirupdqfhv dfklhyhg e| wzr gli0
ihuhqw suhdpeohv dqg wzr glhuhqw v|qfkurql}hu vwuxfwxuhv1 Wkh suhdpeohv
xqghu lqyhvwljdwlrq duh edvhg rq d uhshdwhg RIGP v|pero dqg d uhshdwhg
SQ0vhtxhqfh/ uhvshfwlyho|1 Wkh lqyhvwljdwhg v|qfkurql}huv duh d frqyhqwlrqdo
v|qfkurql}hu/ zkhuh pxowlsolfdwlrq lv shuiruphg rq d shu vdpsoh edvlv ehiruh
vxppdwlrq/ dqg d suh0fruuhodwlqj v|qfkurql}hu/ zkhuh fruuhodwlrq lv shuiruphg
zlwk wkh nqrzq vhtxhqfh sulru wr pxowlsolfdwlrq1 Lq wkh sdshu d orz0frpsoh{lw|
6151 SDSHU FRQWULEXWLRQV 6<
yhuvlrq ri wkh suh0fruuhodwlqj v|qfkurql}hu lv dovr lqwurgxfhg/ vxlwdeoh iru orz0

srzhu rshudwlrq gxulqj/ iru h{dpsoh/ vohhs prgh1 Wkh shuirupdqfh zkhq xvlqj
hdfk ri wkh wzr suhdpeohv wrjhwkhu zlwk hdfk ri wkh wzr v|qfkurql}huv lv dqd0
o|}hg dqdo|wlfdoo|1 Wkh suredelolw| ghqvlw| ixqfwlrq ri wkh wlph v|qfkurql}dwlrq
vljqdo lv ghulyhg iru wkh irxu glhuhqw frpelqdwlrqv1 Wkhq wkh suredelolw| ri
idovh ghwhfwlrq dqg wkh suredelolw| ri plvvlqj wkh v|qfkurql}dwlrq suhdpeoh duh
fdofxodwhg iru DZJQ fkdqqhov1 Wkh shuirupdqfh lq Ud|ohljk idglqj fkdqqhov
lv lqyhvwljdwhg e| vlpxodwlrqv1 Lq dgglwlrq/ wkh yduldqfh ri wkh iuhtxhqf|0rvhw
hvwlpdwh lv jlyhq1 Lw lv vkrzq wkdw wkh suh0fruuhodwlqj v|qfkurql}hu xvlqj wkh
SQ0edvhg suhdpeoh uhvxowv lq vljqlfdqwo| ehwwhu ghwhfwlrq shuirupdqfh frp0
sduhg wr wkh frqyhqwlrqdo v|qfkurql}hu xvlqj dq RIGP0edvhg suhdpeoh1 Wkh
orz0frpsoh{lw| v|qfkurql}hu lv dovr vkrzq wr kdyh dq dwwudfwlyh ghwhfwlrq shu0
irupdqfh1 Wkh suredelolw| ri idovh ghwhfwlrq dqg wkh suredelolw| ri plvvlqj wkh
v|qfkurql}dwlrq suhdpeoh lv hyhq orzhu wkdq wkh fruuhvsrqglqj suredelolwlhv
xvlqj wkh ixoo0uhvroxwlrq frqyhqwlrqdo v|qfkurql}hu1
Wkh pdlq frqwulexwlrqv duh wkh lqwurgxfwlrq dqg dqdo|vlv ri wkh suh0fruuh0
odwlqj v|qfkurql}hu iru suhdpeoh0edvhg v|qfkurql}dwlrq lq RIGP v|vwhpv1 Wkh
dqdo|vlv lv sxuho| dqdo|wlfdo dqg wkh vwdwlvwlfv ri wkh wlph v|qfkurql}dwlrq vlj0
qdov duh ghulyhg iru wkh fdvh ri fruuhfw dqg zurqj wlplqj lq DZJQ fkdqqhov1
Dq lpsruwdqw frqfoxvlrq ri wkh dqdo|vlv lv wkdw lw lv srvvleoh wr hqkdqfh wkh
ghwhfwlrq shuirupdqfh e| lqwurgxflqj wkh suh0fruuhodwru1 Dqrwkhu frqwulexwlrq
lv wkh lqwurgxfwlrq dqg dqdo|vlv ri wkh orz0frpsoh{lw| frduvh wlph v|qfkur0
ql}hu1 Wkh shuirupdqfh ri wklv v|qfkurql}hu lv dovr dqdo|}hg dqdo|wlfdoo| e|
fdofxodwlqj wkh suredelolw| glvwulexwlrq ri wkh v|qfkurql}dwlrq vljqdo1
61518 RIGP Wlph dqg Iuhtxhqf| V|qfkurql}dwlrq e| Vs0

uhdg Vshfwuxp Slorw Whfkqltxh
Lq wklv sdshu/ vxshulpsrvhg slorw vhtxhqfhv duh lqwurgxfhg iru wlph dqg iuh0
txhqf| v|qfkurql}dwlrq lq RIGP v|vwhpv1 Wkh sdshu vwduwv zlwk d ghvfuls0
wlrq ri wkh v|qfkurql}dwlrq v|vwhp edvhg rq vxshulpsrvhg slorw vhtxhqfhv1 Wkh
pdlq lghd ehklqg vxshulpsrvhg slorw vhtxhqfhv lv wr olqhduo| dgg d nqrzq slorw
vhtxhqfh wr wkh xqnqrzq gdwd vhtxhqfh1 Wkh wzr vhtxhqfhv duh wudqvplwwhg
vlpxowdqhrxvo| dqg lq wkh vdph iuhtxhqf| edqg1 Lq wkh sdshu/ d vxlwdeoh v|q0
fkurql}hu vwuxfwxuh lv dovr suhvhqwhg1 Wkh v|qfkurql}dwlrq vljqdo lv fdofxodwhg
lq wkh vdph zd| dv iru wkh suh0fruuhodwlqj suhdpeoh0edvhg v|qfkurql}hu gh0
vfulehg deryh1 Iluvw/ wkh uhfhlyhg vljqdo lv fruuhodwhg wr wkh nqrzq vhtxhqfh/
dqg wkhq wkh uhvxow lv pxowlsolhg e| dq hduolhu fruuhodwlrq ydoxh wr jhw d gli0
ihuhqwldo vljqdo1 Wkh ghod| ehwzhhq wkh whupv ghwhuplqhv wkh iuhtxhqf|0rvhw
udqjh> ghfuhdvlqj wkh ghod| lqfuhdvhv wkh iuhtxhqf|0rvhw udqjh1 Wkh pd{lpxp
iuhtxhqf| rvhw udqjh lv kdoi wkh RIGP edqgzlgwk/ zklfk vkrxog eh frpsduhg
wr kdoi wkh vxe0fkdqqho edqgzlgwk iru frqyhqwlrqdo v|qfkurql}dwlrq edvhg rq
wkh f|folf suh{1 Wkh dpsolwxgh ri wkh glhuhqwldo vljqdo/ ru d vxp ri wkhp/
lv wkh wlph v|qfkurql}dwlrq vljqdo dqg/ zkhq wkh fruuhfw wlplqj lv irxqg/ lwv
skdvh lv xvhg iru wkh iuhtxhqf|0rvhw hvwlpdwh1 Lq wkh sdshu/ wkh glvwulexwlrq
ri wkh v|qfkurql}dwlrq vljqdo lv ghulyhg iru DZJQ fkdqqhov/ dqg yhulhg e|
vlpxodwlrqv1 Wkh suhvhqwhg v|qfkurql}hu vwuxfwxuh lv ri orz frpsoh{lw| dqg lw
lv vkrzq wkdw wkh suhvhqwhg phwkrg zrunv zhoo dovr dw orz VQUv1 Wkh srzhu
ri wkh slorw vhtxhqfh fdq eh dgmxvwhg wr wkh vshflf uhtxluhphqwv/ exw lw fdq
lq jhqhudo eh orz gxh wr vxssuhvvlrq ri qrlvh dqg gdwd zkhq shuiruplqj wkh
fruuhodwlrq1
Wkh pdlq frqwulexwlrqv lqfoxgh wkh lqwurgxfwlrq ri vxshulpsrvhg slorw vh0
txhqfhv iru v|qfkurql}dwlrq sxusrvhv lq RIGP v|vwhpv1 Wkh sursrvhg phwkrg
lv lqghshqghqw ri wkh f|folf suh{ dqg lv yhu| h{leoh1 Wr p| nqrzohgjh/ wkh
ghulydwlrq ri wkh glvwulexwlrq ixqfwlrqv ri wkh wlph v|qfkurql}dwlrq vljqdo lv
dovr qhz/ hyhq wkrxjk vlplodu v|qfkurql}dwlrq vwuxfwxuhv fdq eh xvhg lq FGPD
v|vwhpv1 Dqrwkhu frqwulexwlrq lq wkh sdshu lv wkh jhqhudol}dwlrq ri rwkhu v|q0
fkurql}dwlrq phwkrgv vxfk wkdw wkh dqdo|vlv fdq eh dssolhg wr wkhvh dv zhoo1
61519 Fkdqqho Hvwlpdwlrq xvlqj Vxshulpsrvhg Slorw Vh0

txhqfhv
Lq wklv sdshu/ vxshulpsrvhg slorw vhtxhqfhv duh lqwurgxfhg iru wkh sxusrvh
ri fkdqqho hvwlpdwlrq1 Wkh v|vwhp dqg uhfhlyhu vwuxfwxuh duh wkrurxjko| gh0
vfulehg1 Wkh pdlq lghd lv wr olqhduo| dgg d nqrzq slorw vhtxhqfh wr wkh wudqv0
plwwhg gdwd vhtxhqfh dqg shuirup mrlqw fkdqqho hvwlpdwlrq dqg ghwhfwlrq lq wkh
uhfhlyhu1 Wkh ghwhfwlrq phwkrg lv edvhg rq k|srwkhvhv ri wkh wudqvplwwhg gdwd
vhtxhqfh dqg iru hdfk k|srwkhvlv/ d fkdqqho hvwlpdwh lv ghulyhg e| phdqv ri
shu0vxuylyru surfhvvlqj1 Wkh qdo ghflvlrq lv pdgh lq idyru ri wkh vhtxhqfh zlwk
wkh orzhvw glvwdqfh wr wkh uhfhlyhg vhtxhqfh1 Wkh sursrvhg v|vwhp lv dqdo|}hg
dqdo|wlfdoo| iru dw Ud|ohljk idglqj fkdqqhov1 Wkh sdluzlvh huuru suredelolw| lv
ghulyhg dv zhoo dv dssur{lpdwlrqv ri wkh EHU1 Wkh shuirupdqfh ri wkh v|vwhp
lv yhulhg e| vlpxodwlrqv dqg wkh lq xhqfhv ri glhuhqw sdudphwhuv vxfk dv slorw
srzhu/ owhu frh!flhqwv dqg owhu ohqjwkv duh lqyhvwljdwhg1 Lw lv vkrzq wkdw wkh
whfkqltxh lv urexvw> lw zrunv zhoo iru d zlgh udqjh ri VQU ydoxhv dqg dovr iru
kljk Grssohu iuhtxhqflhv1 Wkh vxshulpsrvhg slorw vhtxhqfhv duh wudqvplwwhg lq
wkh vdph edqg dv wkh gdwd v|perov dqg/ dv rssrvhg wr frqyhqwlrqdo slorw0edvhg
fkdqqho hvwlpdwlrq whfkqltxhv/ wkhuh lv qr edqgzlgwk h{sdqvlrq1 Wkh srzhu ri
wkh slorw vhtxhqfh fdq eh txlwh orz1 W|slfdoo| 8( ri wkh wrwdo srzhu lv hqrxjk1
Wkh whfkqltxh fdq eh xvhg erwk lq h{lvwlqj dqg qhz v|vwhpv dqg lw uhvxowv
lq ehwwhu srzhu h!flhqf| wkdq frqyhqwlrqdo slorw v|pero dvvlvwhg prgxodwlrq/
hvshfldoo| lq idvw idglqj hqylurqphqwv1
Wkh pdlq frqwulexwlrqv duh wkh lqwurgxfwlrq ri fkdqqho hvwlpdwlrq edvhg rq
vxshulpsrvhg slorw vhtxhqfhv dqg wkh fruuhvsrqglqj ghulydwlrq ri wkh rswlpdo
uhfhlyhu1 Wkh uhfhlyhu uholhv rq mrlqw ghwhfwlrq dqg fkdqqho hvwlpdwlrq dqg erwk
dq rswlpdo phwulf dqg d phwulf vxlwdeoh iru d uhfxuvlyh uhfhlyhu vwuxfwxuh duh
ghulyhg1 Wkh dqdo|vlv ri wkh v|vwhp lv dovr qhz/ hyhq wkrxjk vlplodu dqdo|vhv
fdq eh irxqg iru wuhoolv0frghg prgxodwlrq1 Ilqdoo|/ wkh v|vwhp lv yhulhg e|
6161 FRQFOXVLRQV DQG GLVFXVVLRQ 74
h{whqvlyh vlpxodwlrqv dqg wkh lq xhqfh ri owhu ohqjwkv/ qrq0rswlpdo owhuv/

slorw0wr0gdwd srzhu udwlr/ dqg Grssohu iuhtxhqf| kdyh ehhq lqyhvwljdwhg1
6151: D Frpsdudwlyh Dqdo|vlv ehwzhhq Glhuhqw Ghwhf0

wlrq Vfkhphv lq Iodw Ud|ohljk0idglqj Fkdqqhov
Wklv sdshu frqwdlqv d frpsdulvrq ehwzhhq glhuhqw frkhuhqw dqg qrq0frkhuhqw
ghwhfwlrq phwkrgv1 Wkh dqdo|vlv lv edvhg rq fdofxodwlrq ri wkh sdluzlvh vh0
txhqfh huuru suredelolw| dqg d frpprq iudphzrun lv xvhg iru wkh dqdo|vlv ri
doo wkh lqyhvwljdwhg phwkrgv1 Wkh dqdo|}hg phwkrgv lqfoxgh= slorw v|pero
dvvlvwhg prgxodwlrq +SVDP,/ pxowlsoh0v|pero glhuhqwldo ghwhfwlrq/ vxshulp0
srvhg slorw vhtxhqfhv/ glhuhqwldo ghwhfwlrq/ dqg d qryho k|eulg phwkrg/ vlplodu
wr SVDP/ zklfk xvhv erwk slorw v|perov dqg k|srwkhvhv ri wkh gdwd v|perov
iru fkdqqho hvwlpdwlrq1 Lq wkh sdshu/ lw lv vkrzq wkdw wkh ghflvlrq yduldeoh iru
doo phwkrgv fdq eh zulwwhq lq d txdgudwlf irup ri Jdxvvldq yduldeohv/ iru zklfk
wkh suredelolw| ghqvlw| ixqfwlrq lv ghulyhg1 Wkh sdluzlvh huuru suredelolw| lv
frpsduhg iru wkh huuru vhtxhqfhv grplqdwlqj wkh huuru udwh1 Wkh shuirupdqfh
lv dqdo|}hg iru glhuhqw VQU dqg Grssohu ydoxhv dqg lw lv frqfoxghg wkdw vxshu0
lpsrvhg slorw vhtxhqfhv/ pxowlsoh0v|pero glhuhqwldo ghwhfwlrq dqg wkh k|eulg
phwkrg ohdg wr urexvw ghwhfwlrq erwk iru d zlgh udqjh ri VQUv dqg iru kljk
Grssohu iuhtxhqflhv1 Frqyhqwlrqdo glhuhqwldo ghwhfwlrq vkrzv d shuirupdqfh
orvv gxh wr wkh odfn ri qrlvh dyhudjlqj zkhq xvlqj wkh suhylrxvo| uhfhlyhg v|pero
dv d uhihuhqfh iru ghwhfwlrq1 Wrjhwkhu zlwk SVDP wklv phwkrg dovr vxhu iurp
dq huuru rru iru kljk Grssohu ydoxhv gxh wr lqdffxudwh fkdqqho hvwlpdwhv1
Wkh pdlq frqwulexwlrq lq wkh sdshu lv wkh lqwurgxfwlrq ri wkh frpprq
iudphzrun xvhg iru wkh dqdo|vlv ri wkh glhuhqw frkhuhqw dqg qrq0frkhuhqw
ghwhfwlrq phwkrgv1 Wklv frpprq iudphzrun lv xvhg wr qg wkh vhtxhqfhv
grplqdwlqj wkh huuru udwh dqg dovr wr lqyhvwljdwh wkh lq xhqfh ri glhuhqw qrlvh
ohyhov dqg glhuhqw Grssohu iuhtxhqflhv1 Wkh suhvhqwhg dqdo|vlv lv jhqhudo dqg
fdq eh dssolhg wr rwkhu ghwhfwlrq phwkrgv dv zhoo1
616 Frqfoxvlrqv dqg Glvfxvvlrq

Zkloh ehlqj dw wkh ghsduwphqw/ L kdyh kdg wkh rssruwxqlw| wr iroorz wkh gh0
yhorsphqw ri zluhohvv whfkqltxhv dqg vwdqgdugv iru vrph |hduv1 Lq wkh wkhvlv
L kdyh suhvhqwhg uhvhdufk uhvxowv wkdw L kdyh surgxfhg gxulqj wkh odvw irxu
|hduv1 Wkh txhvwlrq lv qrz zkdw zloo kdsshq qh{wB Khuh L zloo jlyh vrph vkruw
shuvrqdo ylhzv/ exw dv dozd|v zkhq uhjduglqj wkh ixwxuh/ wkh| duh rqo| jxhvvhv1
Li zh vwduw dw d kljk ohyho/ wkhuh duh vrph fhuwdlq hyhqwv wkdw zloo fkdqjh
wkh dydlodelolw| ri dffhvv wr idvw zluhohvv ru preloh frppxqlfdwlrq1 Gxulqj
5334/ wkh uvw wklug0jhqhudwlrq v|vwhpv edvhg rq zlghedqg0FGPD zloo eh rq
wkh pdunhw1 Dw wkh vdph wlph wkhuh zloo eh odujh lqyhvwphqwv wr xsjudgh wkh
h{lvwlqj vhfrqg0jhqhudwlrq v|vwhpv vr wkdw wkhvh dovr fdq rhu kljkhu gdwd
udwhv dqg sdfnhw edvhg frppxqlfdwlrq1 Wklv phdqv wkdw gdwd udwhv ri xs wr
6;7 nelw2v fdq eh rhuhg ryhu zlgh duhdv1 Lq d ihz |hduv wkh qhz Z0ODQ
v|vwhpv zloo dovr eh rq wkh pdunhw1 Wkhvh v|vwhpv zloo zrun lq xqolfhqvhg
iuhtxhqf| edqgv dqg rhu gdwd udwhv ri durxqg 53 Pelw2v/ exw zlwk d olplwhg
fryhudjh1 Ilqdoo|/ wkh uvw frpphufldo Eoxhwrrwk ghylfhv zloo eh odxqfkhg vrrq1
Wklv phdqv wkdw vkruw glvwdqfh zluhohvv frppxqlfdwlrq zlwk gdwd udwhv ri xs
wr durxqg :33 nelw2v lv vrrq dydlodeoh iru d zlgh udqjh ri hohfwurqlf ghylfhv1
L wklqn wkdw wkh lpsruwdqfh ri wkh odwwhu wzr v|vwhpv fdq qrw eh ryhuhv0
wlpdwhg1 Wkhvh v|vwhpv zloo rhu idvw zluhohvv frppxqlfdwlrq iru iuhh dqg
fdq eh xvhg dv fkhds olqnv ehwzhhq dq| hohfwurqlf ghylfh dqg wkh dffhvv srlqw1
Wklv phdqv wkdw lw lv srvvleoh iru orz0frvw ghylfhv wr kdyh dffhvv wr idvw dqg
dgydqfhg frppxqlfdwlrq olqnv zkhuh wkh preloh whuplqdo dfwv pruh olnh d uhod|
vwdwlrq1 Suredeo| wkhuh zloo eh d zlgh udqjh ri orz0frvw wklug sduw| surgxfwv
rhulqj qhz vhuylfhv xvlqj lqgluhfw dffhvv wr wkh h{lvwlqj qhwzrun1
L dovr wklqn wkdw wkh Z0ODQ v|vwhpv zloo kdyh d yhu| lpsruwdqw uroh lq
wkh qhdu ixwxuh1 H{fhsw iru fryhudjh dqg pd|eh prelolw|/ wkhvh v|vwhpv duh
dfwxdoo| ixooolqj doo wkh uhtxluhphqwv vwdwhg e| vrph ri wkh ohdglqj dfwruv lq
wkh duhd iru d irxuwk jhqhudwlrq v|vwhp1 Wklv phdqv wkdw |rx fdq dfklhyh
gdwd udwhv kljk hqrxjk erwk iru ylghr dqg kljk0txdolw| dxglr dssolfdwlrqv1 Lw
lv dovr srvvleoh wr eh |rxu rzq rshudwru dqg rhu eurdgedqg vhuylfhv lq dq
r!fh ru d uhvlghqwldo duhd iuhh ri fkdujh1 Ixuwkhu/ |rx fdq vhw xs |rxu rzq
shuvrqdo duhd qhwzrun zkhuh hohfwurqlf ghylfhv lq |rxu krph fdq frppxqlfdwh
zlwk hdfk rwkhu1 Krshixoo| wkhvh v|vwhpv zloo eh xvhu0iulhqgo| hqrxjk qrw wr
suhyhqw wkhlu xvdjh1 Wkh xvh ri wkh xqolfhqvhg edqgv lv yhu| lpsruwdqw vlqfh
wklv phdqv wkdw lq |rxu rzq v|vwhp wkhuh zloo eh qr uhoxfwdqfh xvlqj lw gxh
wr kljk wudqvplvvlrq ihhv1
Lq wklv vfhqdulr wkhuh lv qr fhoo sodqqlqj dqg wkh lqwhuihuhqfh ohyho zloo vrph0
wlphv eh kljk1 Wkh kljk wudqvplvvlrq udwhv phdqv wkdw/ lq rughu wr olplw wkh
wudqvplwwhg srzhu wr dssursuldwh ohyhov/ wkh hqhuj| shu elw zloo eh ghfuhdvhg
frpsduhg wr wrgd|*v v|vwhpv1 Wkhuhiruh/ wkh v|vwhpv qhhg wr eh deoh wr dgdsw
wkh wudqvplvvlrq wr wkh suhvhqw vlwxdwlrq dqg rshudwh lq kljko| glvwxuehg hq0
ylurqphqwv1 Idvw dqg uholdeoh v|qfkurql}dwlrq dqg fkdqqho hvwlpdwlrq duh lp0
sruwdqw lq rughu wr suhyhqw wkh v|vwhpv iurp ehlqj olplwhg e| wkhvh ixqfwlrqv/
udwkhu wkdq e| qrlvh ru lqwhuihuhqfh1
Pd|eh wkh prvw lpsruwdqw wklqj wkdw L kdyh ohduqhg gxulqj wkhvh |hduv lv
wkdw zluhohvv frppxqlfdwlrq v|vwhpv duh frpsoh{ v|vwhpv dqg wkh wuxh fkdo0
ohqjh lv wr rswlpl}h wkh v|vwhp dqg jhw hyhu|wklqj wr zrun wrjhwkhu1 Lw lv qr
xvh wr rqo| lpsuryh fhuwdlq sduwv dv orqj dv wkh uhvw ri wkh v|vwhp fdq qrw
pdnh xvh ri wkhvh lpsuryhphqwv1 Wr ph lw vhhpv wkdw lq wkh kljko| rswlpl}hg
v|vwhpv wkh glhuhqw ixqfwlrqdo eorfnv ehfrph pruh ghshqghqw rq hdfk rwkhu/
wkh lqirupdwlrq h{fkdqjh ehwzhhq wkh eorfnv kdv ehfrph pruh lpsruwdqw1 Wkh
fkdqqho hvwlpdwru/ iru h{dpsoh/ kdv qrw rqo| wr gholyhu fkdqqho hvwlpdwhv wr
wkh htxdol}hu/ exw dovr wr wkh ghfrghu vxfk wkdw wkh ghfrglqj fdq eh rswlpl}hg>
dqg pd|eh wr wkh wudqvplwwhu vxfk wkdw wkh wudqvplwwhg vljqdo fdq eh dgmxvwhg
6161 FRQFOXVLRQV DQG GLVFXVVLRQ 76
wr wkh vshflf fkdqqho1 Lq rughu wr fkdoohqjh wrgd|*v olplwv wkh uhfhlyhu kdv wr
lqfrusrudwh pruh dqg pruh ri wkh dydlodeoh lqirupdwlrq/ riwhq wr wkh sulfh ri
lqfuhdvhg frpsoh{lw|1 Wkhuh lv dovr d txhvwlrq zkhwkhu wklv lqfuhdvh lq frp0
soh{lw| fdq eh drughg dqg krz odujh wkh uhvxowlqj jdlq lv1 Dv dozd|v/ lw lv d
txhvwlrq ri frvw yv1 shuirupdqfh1
Lq rughu wr ixooo wkh uhtxluhphqwv rq v|qfkurql}dwlrq dqg fkdqqho hvwlpd0
wlrq L wklqn wkdw lw lv zhoo zruwk vshqglqj d ihz shufhqw ri wkh wudqvplwwhg
srzhu dqg edqgzlgwk rq nqrzq slorw v|perov1 Wkh slorw v|perov phdq wkdw
idvwhu dqg ohvv frpsoh{ dojrulwkpv fdq eh xvhg/ riwhq uhvxowlqj lq pruh urexvw
ghwhfwlrq dqg ehwwhu shuirupdqfh1 Wkh orvv fdxvhg e| wkh slorw v|perov lv riwhq
vpdoo frpsduhg wr wkh jdlq wkdw fdq eh dfklhyhg/ vr slorw v|perov duh jhqhudoo|
d jrrg lqyhvwphqw
Lq RIGP v|vwhpv lw lv uhdoo| lpsruwdqw wr olplw wkh lq xhqfh ri wkh iuh0
txhqf| rvhw1 Wklv hhfw riwhq uhvxowv lq xqghvludeoh frqvwudlqwv lq wkh v|vwhp
ghvljq/ hvshfldoo| zkhq xvlqj orz frvw whuplqdov zkhuh/ lq jhqhudo/ wkh dffxudf|
ri wkh rvfloodwruv lv orz1 Wkhuhiruh/ L wklqn wkdw lw lv lpsruwdqw wr orrn iru idvw
dqg urexvw hvwlpdwlrq whfkqltxhv dv zhoo dv fkhds dqg dffxudwh rvfloodwruv1 Lq
rughu wr jhw h!flhqw v|vwhpv lw lv lpsruwdqw wkdw wkh shuirupdqfh lv qrw olp0
lwhg e| kdugzduh lpsdluphqwv/ exw udwkhu e| wkh fkdudfwhulvwlfv ri wkh sk|vlfdo
fkdqqho ru e| lqwhuihuhqfh iurp rwkhu xvhuv1
Uhjduglqj ixwxuh zrun rq wkh pdwhuldo suhvhqwhg lq wklv wkhvlv/ L wklqn wkdw
wkh sdshu rq suh0frpshqvdwlrq udwkhu hdvlo| fdq eh h{whqghg wr dffrxqw iru dq0
whqqd glyhuvlw| dv zhoo1 Wklv pdnhv wkh sdshu pruh frpsohwh e| dovr fryhulqj
v|vwhpv vlplodu wr wkh dqwhqqd suh0frpshqvdwlrq v|vwhp lq wkh XPWV2Z0
FGPD v|vwhp1 Wkh suhdpeoh0v|qfkurql}dwlrq sdshu fdq eh h{whqghg wr lq0
foxgh suhdpeoh0edvhg fkdqqho hvwlpdwlrq1 Lq lwv suhvhqw irup/ wkh irfxv lv rq
dftxlvlwlrq ri wlph dqg iuhtxhqf| v|qfkurql}dwlrq sdudphwhuv rqo|1 Lq sdfnhw
edvhg v|vwhpv wkhuh duh/ krzhyhu/ vrph lqlwldo wudqvlhqwv zkhq d sdfnhw duulyhv
dqg lw lv qrw phdqlqjixo wr pdnh d qdo fkdqqho hvwlpdwh ehiruh uhdfklqj vrph
nlqg ri vwhdg| vwdwh1 Dq|zd|/ wkh uhtxluhg lqirupdwlrq lv wkhuh/ diwhu fruuh0
odwlrq wkh fkdqqho lpsxovh uhvsrqvh lv dydlodeoh dv d frduvh fkdqqho hvwlpdwh1
Wkh xvh ri vxshulpsrvhg slorw vhtxhqfhv lv dovr zruwk ixuwkhu lqyhvwljdwlrqv1
Iluvw ri doo/ lw zrxog eh lqwhuhvwlqj wr orrn dw ohvv frpsoh{ uhfhlyhu vwuxfwxuhv1
Wkh whfkqltxh vhhpv surplvlqj/ hvshfldoo| iru idvw idglqj fkdqqhov/ exw wkh
frpsoh{lw| lv vwloo wrr kljk1 Lw zrxog eh yhu| lqwhuhvwlqj wr frpelqh v|qfkur0
ql}dwlrq dqg fkdqqho hvwlpdwlrq wr dfwxdoo| jhw d zrunlqj v|vwhp vxlwdeoh iru
lpsohphqwdwlrq1
Eleolrjudsk|
^4` ;35144e04<<< vxsssohphqw wr lqirupdwlrq whfkqrorj|whohfrppxqlfdwlrqv

dqg lqirupdwlrq h{fkdqjh ehwzhhq v|vwhpvorfdo dqg phwursrolwdq duhd
qhwzrunvvshflf uhtxluhphqwvsduw 44= Zluhohvv ODQ phglxp dffhvv
frqwuro +PDF, dqg sk|vlfdo od|hu +SK\, vshflfdwlrqv= Kljkhu vshhg
sk|vlfdo od|hu +SK\, h{whqvlrq lq wkh 517 JK} edqg1 Whfkqlfdo Uhsruw
S;35144e/ LHHH 0 Lqvwlwxwh ri Hohfwulfdo dqg Hohfwurqlfv Hqjlqhhuv/ Lqf1/
Qhz \run/ XVD/ 4<<<1
^5` Gljlwdo Hqkdqfhg Frugohvv Whohfrppxqlfdwlrqv +GHFW,> frpprq lqwhu0
idfh +FL,> sduw 5= Sk|vlfdo od|hu +SKO,1 HQ 633 4:805 Y41715/ HWVL
Hxurshdq Whohfrppxqlfdwlrqv Vwdqgdugv Lqvwlwxwh/ Vrskld Dqwlsrolv/
Iudqfh/ Mxqh 4<<<1
^6` Gudiw vxssohphqw wr lqirupdwlrq whfkqrorj|0whohfrppxqlfdwlrqv dqg lq0
irupdwlrq h{fkdqjh ehwzhhq v|vwhpv 0 orfdo dqg phwursrolwdq duhd qhw0
zrunv 0 vshflf uhtxluhphqwv 0 sduw 44= Zluhohvv ODQ phglxp dffhvv
frqwuro +PDF, dqg sk|vlfdo od|hu +SK\, vshflfdwlrqv= Kljk vshhg sk|v0
lfdo od|hu lq wkh 8 JK} edqg1 Whfkqlfdo Uhsruw S;35144d/ LHHH 0 Lqvwlwxwh
ri Hohfwulfdo dqg Hohfwurqlfv Hqjlqhhuv/ Lqf1/ Qhz \run/ XVD/ 4<<<1
^7` Eurdgedqg Udglr Dffhvv Qhwzrunv +EUDQ,> KLSHUODQ W|sh 5> Sk|vl0
fdo +SK\, od|hu1 WV 434 7:8 Y41414/ HWVL Hxurshdq Whohfrppxqlfdwlrqv
Vwdqgdugv Lqvwlwxwh/ Vrskld Dqwlsrolv/ Iudqfh/ Dsu1 53331
^8` D1 Djkdprkdppdgl/ K1 Ph|u/ dqg J1 Dvfkhlg1 D qhz phwkrg iru skdvh
v|qfkurql}dwlrq dqg dxwrpdwlf jdlq frqwuro ri olqhduo|0prgxodwhg vljqdov
rq iuhtxhqf|0 dw idglqj fkdqqhov1 LHHH Wudqv1 Frppxq1/ 6<+4,=585</
Mdq1 4<<41
^9` D1 J1 Dupdgd dqg P1 Fdoyl1 Skdvh qrlvh dqg vxe0fduulhu vsdflqj hhfwv
rq wkh shuirupdqfh ri dq RIGP frppxqlfdwlrq v|vwhp1 LHHH Frppx0
qlfdwlrqv Ohwwhuv/ 5+4,=4446/ Mdq1 4<<;1
^:` D1 U1 V1 Edkdl dqg E1 U1 Vdow}ehuj1 Pxowl0Fduulhu Gljlwdo Frppxqlfd0
wlrqv= Wkhru| dqg Dssolfdwlrqv ri RIGP1 Noxzhu Dfdghplf Sxeolvkhuv/
Gruguhfkw/ Wkh Qhwkhuodqgv/ 53331
78
79 ELEOLRJUDSK\
^;` M10M1 ydq gh Ehhn/ P1 Vdqghoo/ dqg S1 R1 Euhmhvvrq1 PO hvwlpdwlrq ri

wlph dqg iuhtxhqf| rvhw lq RIGP v|vwhpv1 LHHH Wudqv1 Vljqdo Surf1/
78+:,=4;334;38/ Mxo| 4<<:1
^<` M10M1 ydq gh Ehhn/ P1 Vdqghoo/ P1 Lvdnvvrq/ dqg S1 R1 Euhmhvvrq1 Orz0
frpsoh{ iudph v|qfkurql}dwlrq lq RIGP v|vwhpv1 Lq Lqw1 Frqi1 rq Xql0
yhuvdo Shuvrqdo Frppxqlfdwlrq/ sdjhv <;5<;9/ Wrn|r/ Mdsdq/ Qry1 4<<81
^43` H1 Ehmmdql dqg M10F1 Ehoruh1 Pxowlfduulhu frkhuhqw frppxqlfdwlrqv iru
wkh xqghuzdwhu dfrxvwlf fkdqqho1 Lq Surf1 Rfhdqv*<9/ yroxph 6/ sdjhv
44584463/ Iruw Odxghugdoh/ IO/ XVD/ 4<<91
^44` U1 Ehqmdplq/ L1 Nd|d/ dqg D1 Ql{1 Vpduw edvh vwdwlrqv iru gxpe wlph0
glylvlrq gxsoh{ whuplqdov1 LHHH Frppxq1 Pdj1/ sdjhv 457464/ Ihe1
4<<<1
^45` M1 D1 F1 Elqjkdp1 Pxowlfduulhu prgxodwlrq iru gdwd wudqvplvvlrq= Dq lghd
zkrvh wlph kdv frph1 LHHH Frppxq1 Pdj1/ 5;+8,=847/ Pd| 4<<31
^46` N1 Eu qlqjkdxv dqg P1 Udglpluvfk1 Frduvh iudph v|qfkurql}dwlrq iru
RIGP edvhg zluhohvv frppxqlfdwlrq v|vwhpv1 Lq Surf1 Lqw1 V|ps1 rq
Shuvrqdo/ Lqgrru dqg Preloh Udglr Frppxq1/ sdjhv ;39;43/ Ervwrq/
PD/ XVD/ Vhsw1 4<<;1
^47` V1 Fduovvrq1 Jhqhudwlrqvy{olqj iu prelowhohirq +lq Vzhglvk,1 WHOH/
;<+5,=6:/ 4<;61
^48` M1 N1 Fdyhuv1 Dq dqdo|vlv ri slorw0v|pero dvvlvwhg prgxodwlrq iru Ud|ohljk0
idglqj fkdqqhov1 LHHH Wudqv1 Yhklf1 Whfkqro1/ 73+7,=9;99<6/ Qry1 4<<41
^49` M1 N1 Fdyhuv dqg S1 Kr1 Dqdo|vlv ri wkh huuru shuirupdqfh ri wuhoolv0
frghg prgxodwlrq lq Ud|ohljk0idglqj fkdqqhov1 LHHH Wudqv1 Frppxq1/
73+4,=:7;6/ Mdq1 4<<51
^4:` U1 Z1 Fkdqj1 V|qwkhvlv ri edqg0olplwhg ruwkrjrqdo vljqdov iru pxowlfkdq0
qho gdwd wudqvplvvlrq1 Ehoo V|vwhp Whfk1 M1/ 78=4::84:<9/ Ghf1 4<991
^4;` O1 M1 Flplql1 Dqdo|vlv dqg vlpxodwlrq ri d gljlwdo preloh fkdqqho xv0
lqj ruwkrjrqdo iuhtxhqf|0glylvlrq pxowlsoh{lqj1 LHHH Wudqv1 Frppxq1/
FRP066+:,=9989:8/ Mxo| 4<;81
^4<` I1 Fodvvhq/ P1 Vshwk/ dqg K1 Ph|u1 Fkdqqho hvwlpdwlrq xqlwv iru dq
RIGP v|vwhp vxlwdeoh iru preloh frppxqlfdwlrq1 Lq LWJ Frqihuhqfh rq
Preloh Udglr/ Qhx0Xop/ Jhupdq|/ Vhsw1 4<<81
^53` P1 gh Frxuylooh/ E1 Pxtxhw/ dqg V1 Vlprhqv1 Qhz hohphqwv xqghuo|0
lqj wkh qhhg iru d plglhg slorw sdwwhuq1 HWVL HS EUDQ grf1 qr1
KO4618Prw4D/ Hxurshdq Whohfrppxqlfdwlrqv Vwdqgdugv Lqvwlwxwh/ Vrskld
Dqwlsrolv/ Iudqfh/ Dsu1 4<<<1
ELEOLRJUDSK\ 7:
^54` G1 Glyvdodu dqg P1 N1 Vlprq1 Pxowlsoh0v|pero glhuhqwldo ghwhfwlrq ri

PSVN1 LHHH Wudqv1 Frppxq1/ 6;+6,=63363;/ Pdu1 4<<31
^55` R1 Hgiruv1 Orz0frpsoh{lw| dojrulwkpv lq gljlwdo uhfhlyhuv1 Skg1 wkhvlv/

Oxoh Xqlyhuvlw| ri Whfkqrorj|/ Oxoh/ Vzhghq/ Vhsw1 4<<91
^56` R1 Hgiruv/ P1 Vdqghoo/ M10M1 ydq gh Ehhn/ V1 N1 Zlovrq/ dqg S1 R1 Eumhvvrq1

RIGP fkdqqho hvwlpdwlrq e| vlqjxodu ydoxh ghfrpsrvlwlrq1 LHHH Wudqv1
Frppxq1/ 79+:,=<64<6</ Mxo| 4<<;1
^57` P1 M1 Ihuqdqgh}0Jhwlqr Jdufld/ M1 P1 Sdh}0Eruudoor/ dqg V1 ]d}r1 H!flhqw

slorw sdwwhuqv iru fkdqqho hvwlpdwlrq lq RIGP v|vwhpv ryhu KI fkdqqhov1
Lq Surf1 LHHH Yhklf1 Whfkqro1 Frqi1/ yroxph 7/ sdjhv 54<654<:/ Dpv0
whugdp/ Wkh Qhwkhuodqgv/ Vhsw1 4<<<1
^58` M1 F1 Kdduwvhq1 Wkh Eoxhwrrwk udglr v|vwhp1 LHHH Shuv1 Frppxq1/

:+4,=5;69/ Ihe1 53331
^59` O1 Kdq}r/ Z1 Zhee/ dqg W1 Nhoohu1 Vlqjoh0 dqg Pxowl0fduulhu Txdgudwxuh

Dpsolwxgh Prgxodwlrq= Sulqflsohv dqg Dssolfdwlrqv iru Shuvrqdo Frp0
pxqlfdwlrqv/ ZODQv dqg eurdgfdvwlqj1 Mrkq Zloh| ) vrqv/ Fklfkhvwhu/
Hqjodqg/ 5qg hglwlrq/ 53331
^5:` V1 Kd|nlq1 Dgdswlyh Ilowhu Wkhru|1 Suhqwlfh Kdoo/ Xsshu Vdggoh Ulyhu/
QM/ XVD/ 6ug hglwlrq/ 4<<91
^5;` S1 V1 Khqu| dqg J1 E1 V1 D qhz dssurdfk wr kljk0fdsdflw| gljlwdo preloh

udglr1 Ehoo V|vwhp Whfk1 M1/ 93+;,=4;<44<37/ Rfw1 4<;41
^5<` S1 Kr dqg G1 Ixqj1 Huuru shuirupdqfh ri pxowlsoh0v|pero glhuhqwldo gh0

whfwlrq ri SVN vljqdov wudqvplwwhg ryhu fruuhodwhg Ud|ohljk idglqj fkdq0
qhov1 LHHH Wudqv1 Frppxq1/ 73+43,=4899489</ Rfw1 4<<51
^63` S1 Krhkhu1 WFP rq iuhtxhqf|0vhohfwlyh odqg0preloh idglqj fkdqqhov1

Lq H1 Eljolhul dqg P1 Oxlvh/ hglwruv/ Frghg Prgxodwlrq dqg Edqgzlgwk0
H!flhqw Wudqvplvvlrq/ sdjhv 64:65;1 Hovhylhu/ Dpvwhugdp/ Wkh Qhwkhu0
odqgv/ Vhsw1 4<<41
^64` S1 Krhkhu/ V1 Ndlvhu/ dqg S1 Urehuwvrq1 Slorw0v|pero0dlghg fkdqqho hv0

wlpdwlrq lq wlph dqg iuhtxhqf|1 Lq N1 Id}ho dqg J1 S1 Ihwwzhlv/ hglwruv/
Pxowl0Fduulhu Vsuhdg Vshfwuxp/ sdjhv 49<4:;1 Noxzhu Dfdghplf Sxeolvk0
huv/ Gruguhfkw/ Wkh Qhwkhuodqgv/ 4<<:1
^65` S1 Krhkhu dqg I1 Wxiyhvvrq1 Fkdqqho hvwlpdwlrq zlwk vxshulpsrvhg slorw

vhtxhqfh1 Lq Surf1 Jorehfrp/ sdjhv 54955499/ Ulr gh Mdqhlur/ Eud}lo/
Ghf1 4<<<1
7; ELEOLRJUDSK\

vhtxhqfh dssolhg wr pxowl0fduulhu v|vwhpv1 Lq N1 Id}ho dqg V1 Ndlvhu/
hglwruv/ Pxowl0Fduulhu Vsuhdg0Vshfwuxp dqg Uhodwhg Wrslfv/ sdjhv 5<8
6351 Noxzhu Dfdghplf Sxeolvkhuv/ Gruguhfkw/ Wkh Qhwkhuodqgv/ 4<<<1
^67` W1 Kroghq dqg N1 Ihkhu1 D vsuhdg vshfwuxp edvhg v|vwhp whfkqltxh iru
v|qfkurql}dwlrq ri gljlwdo preloh frppxqlfdwlrq v|vwhpv1 LHHH Wudqv1
Eurdgf1/ sdjhv 4;84<7/ Vhsw1 4<<31
^68` K1 Kropd dqg D1 Wrvndod/ hglwruv1 ZFGPD iru XPWV/ udglr dffhvv iru
wklug jhqhudwlrq preloh frppxqlfdwlrqv1 Mrkq Zloh| ) vrqv/ Fklfkhvwhu/
Hqjodqg/ 53331
^69` V1 M1 Krzdug dqg N1 Sdkodydq1 Dxwruhjuhvvlyh prghoolqj ri zlgh0edqg
lqgrru udglr sursdjdwlrq1 LHHH Wudqv1 Frppxq1/ <+73,=48734885/ Vhsw1
4<<51
^6:` P10K1 Kvlhk dqg F10K1 Zhl1 D orz0frpsoh{lw| iudph v|qfkurql}dwlrq
dqg iuhtxhqf| rvhw frpshqvdwlrq vfkhph iru RIGP v|vwhpv ryhu idglqj
fkdqqhov1 LHHH Wudqv1 Frppxq1/ 7;+6,=48<9493</ Vhsw1 4<<<1
^6;` L1 Mhrqj dqg P1 Qdndjdzd1 D wlph glylvlrq gxsoh{ FGPD v|vwhp xvlqj
dvv|phwulf prgxodwlrq lq gxsoh{ fkdqqho1 LHLFH Wudqv1 Frppxq/ H;50
E+45,=4<894<96/ Ghf1 4<<<1
^6<` V1 Ndlvhu1 Pxowl0fduulhu FGPD preloh udglr v|vwhpv 0 Dqdo|vlv dqg Rs0
wlpl}dwlrq ri Ghwhfwlrq/ Ghfrglqj/ dqg Fkdqqho Hvwlpdwlrq1 Sk1G1 wkhvlv/
Whfkqlvfkh Xqlyhuvlww P qfkhq/ Pxqlfk/ Jhupdq|/ Mdq1 4<<;1
^73` V1 P1 Nd|1 Ixqgdphqwdov ri Vwdwlvwlfdo Vljqdo Surfhvvlqj= Hvwlpdwlrq
Wkhru|1 Suhqwlfh0Kdoo/ Xsshu Vdggoh Ulyhu/ QM/ XVD/ 4<<61
^74` X1 Odpeuhwwh/ P1 Vshwk/ dqg K1 Ph|u1 RIGP exuvw iuhtxhqf| v|qfkur0

ql}dwlrq e| vlqjoh fduulhu wudlqlqj gdwd1 LHHH Frppxqlfdwlrqv ohwwhuv/
4+5,=797;/ Pdu1 4<<:1
^75` G1 Odqgvwup/ V1 N1 Zlovrq/ M10M1 ydq gh Ehhn/ S1 golqj/ dqg S1 R1

Eumhvvrq1 V|pero wlph rvhw hvwlpdwlrq lq frkhuhqw RIGP v|vwhpv1
Lq Surf1 Lqwhuq1 Frqi1 Frppxq1/ yroxph 4/ sdjhv 833838/ Ydqfryhu/
Fdqdgd/ Mxqh 4<<<1
^76` M1 Ol dqg P1 Ndyhkudg1 Hhfwv ri wkh wlph vhohfwlyh pxowlsdwk idglqj rq
RIGP v|vwhpv iru eurdgedqg preloh dssolfdwlrqv1 LHHH Frppxqlfdwlrq
Ohwwhuv/ 6+45,=665667/ Ghf1 4<<<1
^77` \1 J1 Ol/ O1 M1 Flplql/ dqg Q1 U1 Vroohqehujhu1 Urexvw fkdqqho hvwlpdwlrq
iru RIGP v|vwhpv zlwk udslg glvshuvlyh idglqj fkdqqhov1 LHHH Wudqv1
Frppxq1/ 79+:,=<35<48/ Mxo| 4<<;1
ELEOLRJUDSK\ 7<
^78` D1 Pdvrrp}dghk0Idug dqg V1 Sdvxsdwk|1 Frpelqgh htxdol}dwlrq dqg gli0

ihuhqwldo ghfrglqj xvlqj suhfrglqj1 LHHH Wudqv1 Frppxq1/ 78+6,=5:75:;/
Pdu1 4<<:1
^79` G1 Pdwlf/ Q1 Shwurfklorv/ W1 D1 Frhqhq/ I1 Vfkrxwh/ dqg U1 Sudvdg1 Df0

txlvlwlrq ri v|qfkurqlvdwlrq sdudphwhuv iru RIGP xvlqj d vlqjoh wudlqlqj
v|pero1 Lq N1 Id}ho dqg V1 Ndlvhu/ hglwruv/ PxowlFduulhu Vsuhdg Vshf0
wuxp dqg Uhodwhg Wrslfv1 Noxzhu Dfdghplf Sxeolvkhuv/ Gruguhfkw/ Wkh
Qhwkhuodqgv/ 4<<<1
^7:` K1 Pdwvxnl dqg K1 Wdndqdvkl1 Wkhruhwlfdo dqdo|vlv ri EHU lpsuryhphqw
e| dgdswlyh wudqvplwwhu srzhu frqwuro iru Ud|ohljk idglqj frpshqvdwlrq1
Hohfwurq1 Ohww1/ 5<+4:,=48534854/ Dxj1 4<<61
^7;` K1 Ph|u/ P1 Prhqhfodh|/ dqg V1 D1 Ihfkwho1 Gljlwdo Frppxqlfdwlrq Uh0

fhlyhuv1 Mrkq Zloh| ) vrqv/ Qhz \run/ XVD/ 4<<;1
^7<` P1 O1 Prkhu dqg M1 K1 Orgjh1 WFPS D prgxodwlrq dqg frglqj vwudwhj|
iru Ulfldq0idglqj fkdqqhov1 LHHH M1 Vhohfw1 Duhdv Frppxq1/ :+<,=467:
4688/ Ghf1 4<;<1
^83` S1 K1 Prrvh1 D whfkqltxh iru ruwkrjrqdo iuhtxhqf|0glylvlrq pxowlsoh{lqj
iuhtxhqf| rvhw fruuhfwlrq1 LHHH Wudqv1 Frppxq1/ 75+43,=5<3;5<47/
Rfw1 4<<71
^84` P1 P qvwhu/ W1 Nhoohu/ dqg O1 Kdq}r1 Fr0fkdqqho lqwhuihuhqfh vxssuhvvlrq
dvvlwhg dgdswlyh RIGP lq lqwhuihuhqfh olplwhg hqylurqphqwv1 Lq Surf1
LHHH Yhklf1 Whfkqro1 Frqi1/ sdjhv 5;75;;/ Dpvwhugdp/ Wkh Qhwkhu0
odqgv/ Vhsw1 4<<<1
^85` U1 ydq Qhh dqg U1 Sudvdg1 RIGP iru zluhohvv pxowlphgld frppxqlfd0
wlrqv1 Duwhfk Krxvh/ Ervwrq/ XVD/ 53331
^86` U1 Qlovvrq/ R1 Hgiruv/ P1 Vdqghoo/ dqg S1 R1 Eumhvvrq1 Dq dqdo|vlv ri
wzr0glphqvlrqdo slorw0v|pero dvvlvwhg prgxodwlrq iru RIGP1 Lq Surf1
Lqwhuq1 Frqi1 Shuv1 Zluhohvv Frppxq1/ sdjhv :4:7/ Erped|/ Lqgld/ Ghf1
4<<:1
^87` D1 Shohg dqg D1 Uxl}1 Iuhtxhqf| grpdlq gdwd wudqvplvvlrq xvlqj uhgxfhg
frpsxwdwlrqdo frpsoh{lw| dojrulwkpv1 Lq Surf1 LHHH Lqw1 Frqi1 Dfrxvw1/
Vshhfk/ Vljqdo Surfhvvlqj/ sdjhv <97<9:/ Ghqyhu/ FR/ XVD/ 4<;31
^88` W1 Sroohw/ P1 ydq Eodgho/ dqg P1 Prhqhfodh|1 EHU vhqvlwlylw| ri RIGP
v|vwhpv wr fduulhu iuhtxhqf| rvhw dqg Zlhqhu skdvh qrlvh1 LHHH Wudqv1
Frppxq1/ 76+52627,=4<44<6/ Ihe2Pdu2Dsu 4<<81
^89` M1 J1 Surdnlv1 Gljlwdo frppxqlfdwlrqv1 Suhqwlfh0Kdoo/ Qhz \run/ XVD/
6ug hglwlrq/ 4<<81
83 ELEOLRJUDSK\
^8:` U1 Udkhol/ D1 Sro|grurv/ dqg F10N1 W}rx1 Shu0vxuylyru surfhvvlqj= D

jhqhudo dssurdfk wr POVH lq xqfhuwdlq hqylurqphqwv1 LHHH Wudqv1 Frp0
pxq1/ 76+52627,=687697/ Ihe2Pdu2Dsu 4<<81
^8;` W1 V1 Udssdsruw1 Zluhohvv Frppxqlfdwlrqv/ Sulqflsohv dqg Sudfwlfh1
Suhqwlfh Kdoo/ Xsshu Vdggoh Ulyhu/ QM/ XVD/ 4<<91
^8<` S1 Urehuwvrq1 Forvhwrrswlpdo rqhvkrw iuhtxhqf| v|qfkurql}dwlrq iru
RIGP xvlqj slorw fduulhuhv1 Lq Frppxqlfdwlrq Wkhru| Plql Frqihuhqfh
lq frqmxqfwlrq zlwk LHHH Jorehfrp *<:/ sdjhv <:435/ Skrhql{/ D]/ XVD/
Qry1 4<<:1
^93` S1 Urehuwvrq dqg V1 Ndlvhu1 Dqdo|vlv ri wkh hhfwv ri skdvhqrlvh lq
ruwkrjrqdo iuhtxhqf| glylvlrq pxowlsoh{ +RIGP, v|vwhpv1 Lq Surf1 Lqwhuq1
Frqi1 Frppxq1/ sdjhv 4985498:/ Vhdwwoh/ XVD/ 4<<81
^94` P1 Uxvvho dqg J1 Vw ehu1 Whuuhvwuldo gljlwdo ylghr eurdgfdvwlqj iru preloh
uhfhswlrq xvlqj RIGP1 Zluhohvv Shuv1 Frppxq1/ 5+405,=7899/ 4<<81
^95` J1 Vdqwhood1 D iuhtxhqf| dqg v|pero v|qfkurql}dwlrq v|vwhp iru RIGP
vljqdov= Dufklwhfwxuh dqg vlpxodwlrq uhvxowv1 LHHH Wudqv1 Yhklf1 Whfkqro1/
7<+4,=5875:8/ Mdq1 53331
^96` W1 P1 Vfkplgo dqg G1 F1 Fr{1 Urexvw iuhtxhqf| dqg wlplqj v|qfkurql}d0
wlrq iru RIGP1 LHHH Wudqv1 Frppxq1/ 78+45,=49464954/ Ghf1 4<<:1
^97` P1 Vshwk/ V1 D1 Ihfkwho/ J1 Irfn/ dqg K1 Ph|u1 Rswlpxp uhfhlyhu gh0
vljq iru zluhohvv eurdg0edqg v|vwhpv xvlqj RIGP 0 Sduw L1 LHHH Wudqv1
Frppxq1/ 7:+44,=499;49::/ Qry1 4<<<1
^98` E1 Vwdqwfkhy dqg J1 Ihwwzhlv1 Exuvw v|qfkurql}dwlrq iru RIGPedvhg
fhooxodu v|vwhpv zlwk vhsdudwh vljqdoolqj fkdqqho1 Lq Surf1 LHHH Yhklf1
Whfkqro1 Frqi1/ sdjhv :8;:95/ Rwwdzd/ Fdqdgd/ Pd| 4<<;1
^99` K1 Vwhhqgdp dqg P1 Prhqhfodh|1 Dqdo|vlv dqg rswlpl}dwlrq ri wkh shu0
irupdqfh ri RIGP rq iuhtxhqf|0vhohfwlyh wlph vhohfwlyh idglqj fkdqqhov1
LHHH Wudqv1 Frppxq1/ 7:+45,=4;444;4</ Ghf1 4<<<1
^9:` D1 Vwhlqjdvv/ D1 Zlmqjddughq/ dqg Z1 Whlfk1 Iudph v|qfkurql}dwlrq xvlqj
vxshulpsrvhg vhtxhqfhv1 Lq Surf1 LHHH Lqwhuq1 V|ps1 Lqi1 Wkhru|/ sdjh
7;</ Xop/ Jhupdq|/ Mxqh 4<<:1
^9;` K1 Wdndqdvkl dqg P1 Qdndjdzd1 Dqwhqqd dqg pxowl0fduulhu frpelqhg
glyhuvlw| v|vwhp1 LHLFH Wudqv1 Frppxq1/ H:<E+<,=45544559/ Vhsw1
4<<91
^9<` I1 Wxiyhvvrq1 Vrph fkdqqho hvwlpdwlrq vwudwhjlhv iru d fhooxodu RIGP
v|vwhp1 Lq Surf1 UYN*<9 UdglrVflhqfh dqg Frppxq1 Frqi1/ sdjhv 494
497/ Oxoh/ Vzhghq/ Mxqh 4<<91
ELEOLRJUDSK\ 84
^:3` I1 Wxiyhvvrq1 Fkdqqho uhodwhg rswlpl}dwlrq ri zluhohvv frppxqlfdwlrq

v|vwhpv1 Olfhqwldwh wkhvlv qr1 6/ LVVQ 47350;995/ Ghsduwphqw ri Dssolhg
Hohfwurqlfv/ Oxqg Xqlyhuvlw|/ Oxqg/ Vzhghq/ Ihe1 4<<;1
^:4` I1 Wxiyhvvrq dqg R1 Hgiruv1 D frpsdudwlyh dqdo|vlv ehwzhhq glhuhqw
ghwhfwlrq vfkhphv lq dw Ud|ohljk0idglqj fkdqqhov1 Lq Surf1 LHHH Yhklf1
Whfkqro1 Frqi1/ Wrn|r/ Mdsdq/ Pd| 53331
^:5` I1 Wxiyhvvrq dqg R1 Hgiruv1 Suhdpeoh0edvhg wlph dqg iuhtxhqf| v|qfkur0
ql}dwlrq iru RIGP v|vwhpv1 Vxeplwwhg wr LHHH M1 Vhohfw1 Duhdv Frppxq1/
53331
^:6` I1 Wxiyhvvrq/ P1 Idxonqhu/ dqg R1 Hgiruv1 Wlph dqg iuhtxhqf| v|qfkur0
ql}dwlrq iru RIGP xvlqj SQ0vhtxhqfh suhdpeohv1 Lq Surf1 LHHH Yh0
klf1 Whfkqro1 Frqi1/ sdjhv 5536553:/ Dpvwhugdp/ Wkh Qhwkhuodqgv/ Vhsw1
4<<<1
^:7` I1 Wxiyhvvrq/ P1 Idxonqhu/ R1 Hgiruv/ dqg W1 Pdvhqj1 Wlph dqg iuhtxhqf|
v|qfkurql}dwlrq iru EUDQ xvlqj SQ0vhtxhqfh suhdpeohv1 Lq Surf1 UYN*<<
0 UdglrVflhqfh dqg Frppxq1 Frqi1/ sdjhv 98798;/ Nduovnurqd/ Vzhghq/
Mxqh 4<<<1
^:8` I1 Wxiyhvvrq/ P1 Idxonqhu/ S1 Krhkhu/ dqg R1 Hgiruv1 RIGP wlph
dqg iuhtxhqf| v|qfkurql}dwlrq e| vsuhdg vshfwuxp slorw whfkqltxh1 Lq
Hljkwk Frppxqlfdwlrq Wkhru| Plql Frqihuhqfh lq frqmxqfwlrq zlwk LHHH
LFF *<</ sdjhv 44844</ Ydqfrxyhu/ Fdqdgd/ Mxqh 4<<<1
^:9` I1 Wxiyhvvrq/ P1 Idxonqhu/ dqg W1 Pdvhqj1 Suh0frpshqvdwlrq iru Ud|ohljk
idglqj fkdqqhov lq wlph glylvlrq gxsoh{ RIGP v|vwhpv1 Lq Surf1 9wk LHHH
Lqw1 Zrunvkrs rq Lqwhooljhqw Vljqdo Surf1 dqg Frppxq1 V|vwhpv/ sdjhv
868:/ Phoerxuqh/ Dxvwudold/ Qry1 4<<;1
^::` I1 Wxiyhvvrq/ P1 Idxonqhu/ dqg W1 Pdvhqj1 Suh0frpshqvdwlrq iru Ud|ohljk
idglqj fkdqqhov lq wlph glylvlrq gxsoh{ RIGP v|vwhpv1 Dffhswhg iru
sxeolfdwlrq lq Zluhohvv Shuv1 Frppxq1/ 53331
^:;` I1 Wxiyhvvrq dqg S1 Krhkhu1 Fkdqqho hvwlpdwlrq xvlqj vxshulpsrvhg slorw
vhtxhqfhv1 Vxeplwwhg wr LHHH Wudqv1 Frppxq1/ 53331
^:<` I1 Wxiyhvvrq dqg S1 Nduovvrq1 Vrph ghlvjq lvvxhv iru RIGP KLSHU0
ODQ 5 v|vwhpv edvhg rq udglr fkdqqho phdvxuhphqwv1 HWVL HS EUDQ
grf1 qr1 6WUV3;4D/ Hxurshdq Whohfrppxqlfdwlrqv Vwdqgdugv Lqvwlwxwh/
Vrskld Dqwlsrolv/ Iudqfh/ Dsu1 4<<;1
^;3` I1 Wxiyhvvrq dqg S1 Nduovvrq1 Iuhtxhqf| vwdelolw| dqg ghvljq ri slorw
sdwwhuqv iru fkdqqho hvwlpdwlrq dqg frkhuhqw ghwhfwlrq1 HWVL HS EUDQ
grf1 qr1 KO4518WUV5d/ Hxurshdq Whohfrppxqlfdwlrqv Vwdqgdugv Lqvwl0
wxwh/ Vrskld Dqwlsrolv/ Iudqfh/ Ihe1 4<<<1
85 ELEOLRJUDSK\
^;4` I1 Wxiyhvvrq dqg W1 Pdvhqj1 Rq fkdqqho hvwlpdwlrq vwudwhjlhv iru d fho0

oxodu RIGP v|vwhp1 Lq Surf1 Qruglf Udglr Vhplqdu/ sdjhv 5;64/ Oxqg/
Vzhghq/ Dxj1 4<<91
^;5` I1 Wxiyhvvrq dqg W1 Pdvhqj1 Slorw dvvlvwhg fkdqqho hvwlpdwlrq iru RIGP
lq preloh fhooxodu v|vwhpv1 Lq Surf1 LHHH Yhklf1 Whfkqro1 Frqi1/ sdjhv
496<4976/ Skrhql{/ XVD/ Pd| 4<<:1
^;6` I1 Wxiyhvvrq dqg W1 Pdvhqj1 Rswlpl}dwlrq ri vxe0fkdqqho edqgzlgwk
iru preloh RIGP v|vwhpv1 Lq G1 Hyhulww dqg P1 Uxpvhzlf}/ hglwruv/
Pxowldffhvv/ prelolw| dqg whohwud!f 0 dgydqfhv lq zluhohvv qhwzrunv/ sdjhv
4364471 Noxzhu Dfdghplf Sxeolvkhuv/ Gruguhfkw/ Wkh Qhwkhuodqgv/ 4<<;1
^;7` K1 V1 Zdqj dqg S10F1 Fkdqj1 Rq yhuli|lqj wkh uvw rughu Pdunryldq
dvvxpswlrq iru d Ud|ohljk idglqj fkdqqho prgho1 LHHH Wudqv1 Yhklf1
Whfkqro1/ 78+5,=68668:/ Pd| 4<<91
^;8` V1 Zdudqdeh/ W1 Vdwr/ dqg W1 Deh1 Iruzdug vxefkdqqho frqwuro vfkhph
iru WGG pxowl0fduulhu preloh frppxqlfdwlrq v|vwhp1 LHLFH Wudqv1 Ixq0
gdphqwdov/ H;50D+:,=44:544:;/ Mxo| 4<<<1
^;9` V1 E1 Zhlqvwhlq dqg S1 P1 Hehuw1 Gdwd wudqvplvvlrq e| iuhtxhqf|0glylvlrq
pxowlsoh{lqj xvlqj wkh glvfuhwh Irxulhu wudqvirup1 LHHH Wudqv1 Frppxq1/
FRP04<+8,=95;967/ Rfw1 4<:41
^;:` P1 G1 \dfrxe1 Irxqgdwlrqv ri Preloh Udglr Hqjlqhhulqj1 FUF Suhvv/
Erfd Udwrq/ IO/ XVD/ 4<<61
^;;` S1 ]kl|rqj/ \1 [ldrkx/ dqg F1 Vkl{lq1 Wudqvplvvlrq dqg uhfhswlrq ri
WGG pxowlfduulhu FGPD vljqdov lq preloh frppxqlfdwlrqv v|vwhp1 Lq
Surf1 LHHH Yhklf1 Whfkqro1 Frqi1/ sdjhv 5467546;/ Kxvwrq/ W[/ XVD/
Pd| 4<<<1
^;<` Z1 ]kxdqj dqg Z1 Y1 Kxdqj1 Skdvh suhfrglqj iru iuhtxhqf|0vhohfwlyh
Ud|ohljk dqg Ulfldq vorzo| idglqj fkdqqhov1 LHHH Wudqv1 Yhklf1 Whfkqro1/
79+4,=45<475/ Ihe1 4<<:1
^<3` Z1 ]kxdqj/ Z1 D1 Nu}|phlq/ dqg S1 D1 Jrxg1 Dgdswlyh fkdqqho suhfrg0
lqj iru shuvrqdo frppxqlfdwlrqv1 Hohfwurq1 Ohww1/ 63+4<,=48:348:4/ Vhsw1
4<<71
Sduw LL
Lqfoxghg Sdshuv
86
Paper A
Paper A
This part has been published as:
Fredrik Tufvesson and Torleiv Maseng, "Pilot Assisted Channel Estimation for
OFDM in Mobile Cellular Systems", Proceedings of IEEE Vehicular Technology
Conference, Phoenix, USA, pp. 1639-1643, May 1997.
IEEE 1997, reprinted with permission.
Pilot Assisted Channel Estimation for OFDM
in Mobile Cellular Systems
Fredrik Tufvesson and Torleiv Maseng
Department of Applied Electronics, Lund University,
Box 118, S-221 00 Lund, Sweden
E-mail:Fredrik.Tufvesson@tde.lth.se
Abstract
The use of pilot symbols for channel estimation introduces overhead and it is
thus desirable to keep the number of pilot symbols to a minimum. The number
of needed pilot symbols for a desired bit error rate and Doppler frequency is
highly dependent on the pilot pattern used in orthogonal frequency division mul-
tiplexed, OFDM, systems. Five different pilot patterns are analysed by means of
resulting bit error rate, which is derived from channel statistics. Rearrangement
of the pilot pattern enables a reduction in the number of needed pilot symbols up
to a factor 10, still retaining the same performance. The analysis is general and
can be used for performance analysis and design of pilot patterns for any OFDM
system.
1 Introduction
The mobile channel introduces multipath distortion of the signalling waveforms.

Both the amplitude and phase are corrupted and the channel characteristics changes
because of movements of the mobile. In order to perform coherent detection, reliable
channel estimates are required. These can be obtained by occasionally transmitting
known data or so called pilot symbols. The receiver interpolates the channel infor-
mation derived from the pilots to obtain the channel estimate for the data signal, see
figure 1.
OFDM, orthogonal frequency division multiplexing, is used and proposed for

several broadcast systems and there is a growing interest in using the technique for
the next generation land mobile communication system. In OFDM systems the infor-
mation signal can be seen as divided and transmitted by several narrowband sub-car-
riers. Typically, for practical OFDM systems, the frequency spacing is less than the
coherence bandwidth and the symbol time is less than the coherence time. This
means that a receiver and pilot estimation pattern that take advantage of the relatively
large coherence bandwidth and coherence time can manage with less pilot symbols,
1
2 PILOT PATTERN DESIGN
data
transmitter decision
channel
pilot estimator
pattern
receiver
Figure 1: The pilot pattern is known both by the receiver and the transmitter. The pur-
pose is to minimize the number of transmitted pilot symbols without increasing the
bit error rate.
thereby minimizing the overhead introduced by the pilot symbols. The problem is to
decide where and how often to insert pilot symbols. The spacing between the pilot
symbols shall be chosen small enough to enable reliable channel estimates but large
enough not to increase the overhead too much. This paper includes among all an
algorithm of how to design a suitable pilot estimation pattern.
In a multicarrier system there exist a unique opportunity to determine various

parts of the channel impulse response, as opposed to a single-carrier system. It is no
use to make efforts to determine already known parts. Until now it seems like no one
has looked into the impact of the placement of the pilot symbols for multicarrier sys-
tems. Cavers [5] made an exhaustive theoretical analysis for single-carrier systems.
He pointed out that it was appropriate that 14% of the sent symbols were pilot sym-
bols to be able to handle large Doppler values (fdTs=0.05). Some pilot estimation pat-
terns for OFDM has been presented in the literature, see e.g. [6], [7]. Comparisons
between these and the one proposed here are shown later.
2 Estimation Strategies
Five different pilot patterns are analyzed, see figure 2.
1. Measure all channels at the same time, compare to a broadband single-carrier sys-
tem.
2. Measure the channels in increasing order, one at a time.
3. Measure the channels one at a time in a smart, but predetermined, way. The mea-
surement order is derived upon channel statistics and is optimal in the sense that
the total bit error rate is kept at a minimum each symbol time.
4. A pilot pattern presented by T. Mueller et al. [6], where the pilots are located with
equidistant spacings in time and frequency.
PILOT PATTERN DESIGN 3
64
1) 2)
......
...
Pilot
symbol
Channel
Data
symbol
2
1 ... ...
1 2 Time 65
3) 4) 5)
...
...
..
... ... ...
Figure 2: Examples of pilot patterns analyzed.
5. A pilot pattern by P. Hoeher [7], used in [4]. The pilot symbol locations are
shifted one step in frequency each pilot interval.
For comparison the same amount of pilots is used, one of 64 sent symbols. This
means that only 1.6% extra overhead is introduced by the pilots, but this is not suffi-
cient for large Doppler frequencies in some of the cases.
3 System
At different signal to noise ratios the resulting bit error rate from each pilot pattern is
evaluated using the algorithms given in Section 8. A matched filter receiver and
coherent BPSK or QPSK are used. Additive white Gaussian noise is assumed for
every sub-channel. The channel has delays and Rayleigh distributed amplitudes
according to the COST 207: Typical Urban profile. The complex parts of the trans-
fer function are assumed to change according to a first order auto-regressive process
as described in Section 6. The reason for using this model is to get a linear system
which rather easy can be hand-led algebraically. A Kalman filter is used to estimate
the frequency response of the channel. From the filter, a time dependent covariance
matrix is given as described in Section 7. This is used to calculate the expected bit
error rate for each channel. See figure 3 for a description of the system.
4 Results
The resulting bit error rate curves of the pilot patterns are presented for different
Doppler frequencies in figure 4.
I Typical urban X
PSK channel, H +
Used for design

noise: e
pilot
P, K, C channel
Estimated BER estimator

H

Real BER I decision H
matched filter
Figure 3: Overview of the system used to analyze the estimation patterns.
BER for QPSK when different pilot patterns are used, Eb/No=10
Pattern 1
Pattern 2
Pattern 3
Pattern 4
Pattern 5
Coherent
-1
10
BER
-2
10 -4 -3 -2 -1
10 10 10 10
fd*Ts
Figure 4: Resulting bit error rate as a function of Doppler value when the different
pilot patterns are used. The same number of pilot symbols are used for all curves.
For a given Doppler frequency the pilot pattern used sets the limit for the lowest pilot
density to be used, alternatively for a given pilot density it limits the maximum Dop-
pler frequency allowed. The calculations are made at 1800 MHz using 10 kHz chan-
nel separation between 64 OFDM channels carrying in total 640 ksymbols/s. No
intersymbol interference, perfect synchronization and known Doppler frequency, fd,
is assumed. 1.6% of the sent symbols are pilot symbols and the average bit error rate
of the 64 channels is presented.
The bit error rate is degraded both by imperfect channel estimates and noise dis-
turbances. The pilot pattern used determines the conversion between noise limited
and estimate limited region, see figure 5.
BER for QPSK when different pilot patterns are used, fdTs=0.001
0
10
Pattern 1
Pattern 2
Pattern 3
Pattern 4
Pattern 5
-1 Coherent
10
BER
-2
10
-3
10
-4
10
0 5 10 15 20 25 30
Eb/No
Figure 5: Bit error rate at different signal to noise ratios. Note the difference between
the error floors.
It is interesting to study the resulting pattern when estimating the channel that
gives the lowest bit error rate (strategy 3). A steady state pilot pattern is often
achieved where only few of the sub-channels are measured, see figure 6. Channel
estimates of the other sub-channels are achieved by filtering.
To see the effect of mismatch between the pilot pattern design parameters and the
actual parameters, the optimal pilot patterns (pattern 3) for three Doppler values were
Pilot pattern, fdTs=0.002 Pilot pattern, fdTs=0.02
60 60
50 50
40 40
Channel
Channel
30 30
20 20
10 10
20 40 60 80 100 20 40 60 80 100
Time Time
Figure 6: Resulting pilot pattern when sending a pilot at the channel which gives the
lowest total bit error rate. Note that only few channels will be used for pilot symbols.
used when moving at another speed. The actual Doppler frequency was as before
assumed to be known by the estimator, just to see the effect of the pilot pattern with-
out influences of the estimation algorithm. The designed pilot patterns for the typi-
cal urban environment were also used in hilly terrain and rural area specified in
[2] to see the influence of the propagation environment on the bit error rate, see
figure 7. As seen in the figure, the pilot patterns are robust to mismatches in the
design parameters.
5 Discussion
To minimize the bit error rate it is desirable to spread the pilot symbols both in time
and in frequency, as seen in figure 4 and figure 5. Normally a worst case design is
made for the channel estimation system and then we suggest to tailor the pilot pattern
to each base station site. A suitable pilot pattern can be calculated once the propaga-
tion environment and maximum expected speed in the particular cell is known. In
such a system, the pilot pattern used in the cell is among the information transferred
to the mobile when it logs into a new cell. When designing the pilot pattern one also
has to take the estimation algorithm into account. Here the estimation algorithm is
used only for evaluation and the complexity of the used algorithm is not a problem.
In some cases the pilot pattern has to aid the estimation algorithm to enable a less
complex one. The estimation algorithm used here, the Kalman filter based on an AR-
process, has no delay and the received signal can be detected immediately, i.e. no
future pilot symbols is taken into account when making the channel estimate. If the
BER when using pattern designed for another Doppler freq. or channel, Eb/No=10
-1
10
Rural
Hilly
Urban
fd*Ts=0.02
fd*Ts=0.002
fd*Ts=0.0002
-2
10 -4 -3 -2 -1
10 10 10 10
fd*Ts
Figure 7: Changes in the bit error rate due to Doppler mismatch and power delay pro-
file mismatch. The pilot patterns designed for fdTs=0.02, 0.002 and 0.0002 were used
at different Doppler values and the designed patterns for typical urban environment
were used in hilly terrain and rural area.
received signal is stored in a memory, the pilot symbols can be used in both time
directions and the time spacings between pilot symbols can be increased, retaining
the same performance.
The degradation due to mismatch in design parameters is mainly caused by the

estimation algorithm and therefore the curves for different Doppler values do not dif-
fer much. When the pilot pattern is designed for higher Doppler values than the
actual one, an increased error rate is achieved since the pilot symbols are not located
as close to each other in frequency as desired. In rural area the bit error rate becomes
lower due to the increased frequency correlation while the opposite happens in hilly
terrain. In the first case, an even better result is achieved with less pilots along the
frequency axes and more along the time axes.
The bit error rates within the sub-channels differ depending on where the pilots
are located. When minimizing the total bit error rate (pattern 3) the channels of the
sides get higher error rate, see example in figure 8.
6 Channel Model
The time dependent impulse response, h ( , t ) , is assumed to be a sum of reflections,

see (1) where ( t ) denotes the dirac-function.
0.12
0.11
0.1
BER
0.09
0.08
0.07
0.06
80
60
40
70
60
20 50
40
30
20
0 10
Time 0
Channel
Figure 8: The bit error rate becomes higher for the side channels when the total bit
error rate is kept at a minimum
N
h ( , t ) = an ( t ) ( t n ) (1)
n=1
The tap coefficients, an(t), and the tap delays, n , are chosen according to the COST
207 Typical Urban model in the GSM specification [2]. The transfer functions,
H(f,t), are obtained by the Fourier transform and these are the functions we want to
estimate for the different carriers. These channel transfer functions are regarded as
flat fading and constant during a symbol time.
A first order AR-process is used to model how the different taps may change from
one time instant to another. If we look at all the transfer functions at the same time, it
is possible to set up a state-space model of the form:
H( k + 1 ) = H(k) + v( k)
(2)
y(k) = C(k)H(k) + e(k)
The matrix is a diagonal N*N-matrix (here treated as a scalar) with elements
k AR 2 f d T s
e (3)
that define the AR-process. Ts is the symbol time including any cyclic prefix or guard
space. The white noise v(k) has covariance matrix R 1 = F R GSM F , where RGSM cor-
responds to the multipath intensity profile described in [2]. The vector y(k) is the
measured transfer functions. C(k) is an observation vector with ones only at the posi-
tions (channels) measured at time KTs and e(k) is measurement noise with a diagonal
covariance matrix R2.
The parameter kAR in the auto-regressive process for the channel changes is cho-
sen to adjust the memory so that it corresponds to the memory of Jakes model.
Channels corresponding to the U-shaped spectrum given in [8] were simulated and
then estimators based on an AR-model with different kAR were used, see Figure 9:
Simulations were performed with one sub-channel (Eb/No= 10, fdTs=0.002) with
every tenth symbol as a pilot symbol. In the figure the bit error rates of the nine data
symbols are shown. figure 9 shows that the best fit, in this case, is reached for
kAR0.15.
The adjustment of the AR-process can also be seen as an adjustment of the band-
width. If we set the 90% power bandwidth of the AR process equal to the bandwidth
of the Doppler spread, see figure 10, a value of kAR0.158 can be calculated. This is
the value used for all bit error rate calculations throughout the paper.
Simulated BER for QPSK, estimates based on AR-models with different fdTs
0.15
data symbols 1-9

mean
0.14
0.13
BER
0.12
0.11
0.1
0.09 -1 0 1
10 10 10
fdTs used in estimator (fdTs of estimator)/(fdTs of simulated channel)
Figure 9: Bit error rate for a simulated channel by Jakes when estimates are based on
an AR-process. The minimum value is reached for kAR=0.15.
Doppler power spectrum of AR-process and channel by Jakes

S(f)
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2

frequency (f-fc)/fd
Figure 10: Power density spectrum of the channel by Jakes and a first order AR-pro-
cess. The latter is adjusted to have the 90% power bandwidth equal to the Doppler
spread.
7 Channel Estimator
For the analysis and pilot pattern design a Kalman filter is used to estimate the trans-
fer functions. The Kalman filter is causal and uses measurements up to time k to esti-
mate the transfer functions, H(k). The Kalman filter is given [3] by (4)-(8), where X
denotes conjugate transpose of X:

H ( k k ) = H ( k k 1 ) + P ( k k 1 ) C ( k ) (4)
1
{ C ( k ) P ( k k 1 ) C ( k ) + R 2 } ( y ( k ) C ( k ) H ( k k 1 ) )

H( k + 1 k) = H( k k 1) + (5)

K(k )[y( k ) C( k )H(k k 1 )] = H(k k )
K ( k ) = P ( k k 1 ) C ( k ) (6)
1
[ C ( k ) P ( k k 1 ) C ( k ) + R 2 ]
P ( k + 1 k ) = P ( k k 1 ) + R 1 (7)
K ( k ) [ C ( k ) P ( k k 1 ) C ( k ) + R 2 ] K ( k )
P ( k k ) = P ( k k 1 ) P ( k k 1 ) C ( k ) (8)
1
[ C ( k ) P ( k k 1 ) C ( k ) + R 2 ] C ( k ) P ( k k 1 )

The reconstruction error H (k k ) = H (k ) H (k k ) is given by:

H(k k ) = [ K( k)C( k) ]H( k 1 k 1 ) + v(k 1 ) (9)
1
P ( k k 1 ) C ( k ) [ C ( k ) P ( k k 1 ) C ( k ) + R 2 ]
{ e( k ) + C( k) v( k 1 ) }
The Kalman filter is optimal in the sense that the variance of the reconstruction error
is minimized. The matrix P(k|k) is the variance matrix and this is used together with
K to make an estimate of the bit error rate, which in turn is used to decide the order in
which the channels are going to be measured. For pattern 3, the channel is chosen
that minimizes the total bit error rate of the channels after the measurement. The
matrices P(k|k) and K are independent of the measured values and therefore it is pos-
sible to precompute the order in which the channels are going to be measured.
8 Bit Error Rate Calculations
The bit error rate is calculated for BPSK and QPSK. A matched filter receiver is
assumed. The sampled output of this filter is given by
2
j ------ ( n 1 )
M
Xm = 2 Hm e + em (10)
where is the signal energy, Hm is the current transfer function at channel m, n is the
signal alternative among M sent and em is white gaussian noise. The bit error rate at
channel m is calculated as [1]
1
P bmBPSK = --- ( 1 m )
2
1 m (11)
P bmQPSK = --- 1 --------------------
2
2 m
2
where
E [ X m Hm ]
m = ------------------------------------------------ (12)
2
E [ X ] E [ H ]
2
m m

Hm are the outputs of the Kalman filter. These are not known in advance and there-
fore (13) -(15) are used. If matrix notation is used and signal alternative 1 is used for
the pilot symbol (does affect the analysis here, but in practise different pilot symbol
alternatives should be used), the expectations for all the channels can be calculated
as:

E[ X(k )H( k) ] = E[{ 2H( k) + N(k )} { H( k ) H( k) } ] (13)

= 2 E [ H ( k ) H ( k ) ] 2 E [ H ( k ) H ( k ) ]
2
E [ X ( k ) ] = E [ { 2 H ( k ) + N ( k ) } { 2 H ( k ) + N ( k ) } ] (14)
2
= 4 E [ H ( k ) H ( k ) ] + E [ N ( k ) N ( k ) ]
2
E[ H(k ) ] = E[{ H( k ) H( k )} { H( k ) H(k ) } ] (15)

= E [ H ( k ) H ( k ) ] 2E [ H ( k ) H ( k ) ] + E [ H ( k ) H ( k k ) ]
The expected values E [ X ( k ) H ( k ) ] , E [ X ( k ) 2 ] and E [ H ( k ) ] are independent of the
2
measured values but dependent upon which channels that are measured. Therefore
they are time dependent. The expectations can be calculated as:
E [ H ( k ) H ( k ) ] (16)
= E [ { H ( k 1 ) + v ( k 1 ) } { H ( k 1 ) + v ( k 1 ) } ]
2
= E [ H ( k 1 ) H ( k 1 ) ] + E [ v ( k 1 ) v ( k 1 ) ]
2
= R1 ( 1 )
E [ N ( k ) N ( k ) ] = R 2 (17)

E [ H ( k ) H ( k k ) ] = E [ H ( k ) H ( k ) H ( k ) H ( k k ) ] = (18)

E [ H ( k 1 ) H ( k 1 ) + R 1 R 1 C ( k ) a

H ( k 1 ) H ( k 1 ) C ( k ) a ]

= ( E [ H ( k 1 ) H ( k 1 ) ] + R 1 ) ( I C ( k ) a )
where

E[H(k k)H(k k ) ] = P(k k) (19)
1
a = P ( k k 1 ) C ( k ) { C ( k ) P ( k k 1 ) C ( k ) + R 2 } (20)
The bit error rate calculation is compared and verified by simulations. Proakis [1]
gives expressions for the bit error rate when estimating a constant Rayleigh channel
using different numbers of pilot symbols, see figure 11.
The bit error rate when using only one pilot symbol corresponds to estimating a
rapidly changing channel or the first estimate of an unknown channel. Then, only the
latest measurement is useful. In a similar way, the bit error rate when estimating a
constant channel with use of (infinitely) many pilot symbols corresponds to that of
coherent detection. Normally in a practical system the effect of the estimation is
somewhere between these two cases.
BER for BPSK, constant Rayleigh channel, Eb/No=0, 10, 20 and 30

0
10
Proakis
coherent
own calculations
own simulations
-1
10
BER
-2
10
-3
10
-4
10
0 2 4 6 8 10 12 14 16 18 20
Number of pilot symbols used for the estimate
Figure 11: Bit error rate for 2PSK when estimating a constant Rayleigh channel with
several pilot symbols at different signal to noise ratios.
9 Conclusions
The bit error rate for pilot assisted QPSK modulation is calculated when using differ-
ent pilot patterns. The ability to estimate the channel reliably when it changes due to
e.g. vehicle movements is highly dependent on the pilot pattern used. By rearranging
the pilot pattern it is, in some cases, possible to handle 10 times higher Doppler fre-
quency alternatively possible to reduce the number of needed pilot symbols the same
amount, still retaining the same bit error rate. Alternatively the new pilot pattern
could be used to reduce the bit error rate up to a factor 5, even more in a low noise
environment. The pilot symbols in the proposed pilot pattern are spread out both in
time and frequency. For a given propagation environment, e.g. a base station site, it is
possible to pre-calculate a suitable pilot pattern.
References
[1] J. G. Proakis, Digital Communications, third edition, McGraw-Hill, USA,

1995.
[2] European Telecommunications Standard Institute, GSM 05.05, version 4.6.0,
July 1993.
[3] J. M. Mendel, Lessons in Estimation Theory for Signal Processing, Communi-
cations and Control, Prentice-Hall, USA, 1995.
[4] M. Sandell, Design and Analysis of Estimators for Multicarrier Modulation and
Ultrasonic Imaging, Ph.D. thesis, Lule University of Technology, Sweden,
1996.
[5] J. K. Cavers, An Analysis of Pilot Symbol Assisted Modulation for Rayleigh
Fading Channels, IEEE Trans. Vehic. Tech. vol 40, no 4, pp 686-693, nov 1991.
[6] T. Mueller, K. Brueninghaus and H. Rohling, Performance of Coherent OFDM-
CDMA for Broadband Mobile Communications, Wireless Personal Communi-
cations 2, pp 295-305, Kluwer, Netherlands, 1996.
[7] P. Hoeher. TCM on Frequency-Selective Land-Mobile Fading Channels, Proc.
of 5th Tirrenia Int. Workshop in Dig. Communications, Tirrenia, Italy, Sept.
1991.
[8] W. Jakes, Microwave Mobile Communications, Wiley-Interscience, USA 1974
Paper B
Paper B
Fredrik Tufvesson and Torleiv Maseng, "Optimization of Sub-Channel Bandwidth

for Mobile OFDM Systems", in D. Everitt and M. Rumsewicz, editors, Multiaccess,
mobility and teletraffic - advances in wireless networks, pp. 103-114, Kluwer Aca-
demic Publishers, Dordrecht, The Netherlands, 1998.
Kluwer Academic Publishers 1998, reprinted with permission.
Optimization of Sub-Channel Bandwidth for
Mobile OFDM Systems
Fredrik Tufvesson and Torleiv Maseng
Department of Applied Electronics, Lund University,
Box 118, SE-221 00 Lund, Sweden.
E-mail: Fredrik.Tufvesson@tde.lth.se
Abstract
The bit error rate in orthogonal frequency division multiplex (OFDM) sys-
tems is affected by the number of sub-channels used, changes in the channel
characteristics and, to some extent, the excess delay of the channel. This paper
presents an analytical expression for the bit error rate on Rayleigh fading chan-
nels when interchannel interference (ICI) caused by channel changes during a
symbol and energy loss due to the cyclic prefix are regarded. This expression is
used to optimize the number of sub-channels, and thereby the sub-channel band-
width in the system. It is argued that the system can be optimized neglecting the
effect of imperfect channel estimation and on a worst case assumption for the
Doppler frequency and signal to noise ratio. The analysis is general and can be
used for performance analysis and optimization of any mobile OFDM system.
1 Introduction
OFDM, orthogonal frequency division multiplexing, is used and proposed for several
broadcast systems [1] [2] and there is a growing interest in using the technique for the
next generation land mobile communication system. In OFDM systems the informa-
tion signal can be seen as divided among and transmitted by several narrowband sub-
carriers. The bandwidths of the sub-carriers depend on the bit rate sent on each of
them and consequently, for a given total bitrate the sub-carrier bandwidth is depen-
dent on the number of sub-channels used. The number of sub-channels used is set in
the system design and the problem is to find a good trade off between bandwidth,
limitations by the hardware and the physical channel. As the sub-carrier bandwidth is
reduced, the symbol time on each carrier gets longer, the channel changes during a
symbol gets larger and channel compensation gets more difficult. When the changes
in the channel characteristics during a symbol time become evident, the orthogonality
between the subchannels is lost and interchannel interference (ICI) arises. On the
other hand as the carrier bandwidth and the sub-channel symbol rate increases,
intersymbol interference (ISI) becomes a problem. This problem can be avoided by
1
2 SUB-CHANNEL BANDWIDTH
introducing a cyclic prefix [3], but this results in an energy loss. In between these two
extremes there exists an optimal bandwidth where the bit error rate is minimized, this
bandwidth is found in this paper.
There is some previous work regarding the choice of number of sub-channels in

OFDM systems, see e.g. [4] [5], but often the optimization is performed empirically
or by simulation. In [4] a noncoherent OFDM DPSK system is analysed with respect
to random FM noise and a frequency selective channel. It is concluded that a sub-
channel bandwidth of 1.5 kHz is suitable for a bitrate of 200 kbit/s and 100 Hz Dop-
pler frequency, but without consideration to the interchannel interference caused by
channel changes during a symbol. The latter can cause severe degradation of the bit
error rate if the sub-channel bandwidth is chosen too small.
2 System description
When designing an OFDM system one often start with a requirement on the total bit
rate. Assume that a total bitrate of Rtot bits per second is required. If these are equally
divided between the channels then the bitrate
R tot
R 1 = --------- (1)
N
is transferred on each of the N sub-channels. The symbol time for M-ary modulation
on each sub-channel becomes
log 2M N log 2M
T sub = --------------- = -------------------- . (2)
R1 R tot
In the following the influence of some of the most important design parameters are
discussed.
2.1 Cyclic prefix
To avoid intersymbol interference and to maintain orthogonality between the sub-

channels in a time dispersive channel, a cyclic prefix can be used [3]. ISI is avoided if
the length of the cyclic prefix, Tcp, is chosen large enough to exceed the maximum
excess delay of the channel. The total symbol time, Tsub, is extended when a cyclic
prefix is used but the sub-channel bandwidth, f , is equal to the inverse of the sym-
bol time excluding the cyclic prefix [4], Ts, see figure 1.
SUB-CHANNEL BANDWIDTH 3
copy
f
N*f
f
t
Tcp Ts
Tsub
Figure 1: Eight symbols sent on four sub-channels. The striped area indicates inser-
tion of the cyclic prefix.
1 1 R tot
f = ----- = ------------------------- = ------------------------------------------ (3)
Ts T sub T cp N log 2M T cp R tot
Insertion of a cyclic prefix means that the sub-channel bandwidth has to be increased
in order to keep the bit rate R1 constant. The bit rate determines the symbol time Tsub
and if a part of this is used for the cyclic prefix, then the time Ts has to be shortened
which in turn leads to increased sub-channel bandwidth.
The cyclic prefix also leads to a power loss, cp . The receiver uses the energy
received during the time Ts and discards the energy corresponding to the duration of
the cyclic prefix. The remaining signal energy can be calculated as cp E s where
Ts T cp R tot
cp = ---------- = 1 -------------------- . (4)
T sub N log 2M
2.2 Interchannel interference
Even though the channel is perfectly estimated there are some variations in the trans-
fer function during each symbol interval. These variations becomes evident when
many sub-channels are used due to the long symbol time, they are hard to track and
result in interchannel interference, ICI. When the number of sub-channels is suffi-
ciently large the resulting ICI can be modeled as additive white Gaussian noise,
which is added to the channel noise [6] with spectral density N0. The variance of the
ICI-noise, 2ICI = E [ n ICI n ICI ] , is calculated as[6]:
N1
log 2M T cp
= 2E s 1 ---- ------ ( N n ) J 0 2 f D n --------------- --------

2 1 2
ICI (5)
N N 2 R tot N
n=1
2.3 Bit error rate for coherent detection
If we neglect the effect of channel estimation errors, the equivalent signal to noise
ratio after the cyclic prefix is removed and the ICI-noise is added becomes
E E s cp
-----s- = -----------------------
-. (6)
N 0 eq 2
N 0 + ICI
The BER for QPSK/4QAM in a Rayleigh fading channel with coherent detection can
then be calculated as

Es
p ( y ) Q y ------ dy = ---
1
pb = N 0 eq 2
(7)
y=0
1 NE s ( 2N T cp R tot )
--- ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
-
2 N1
2 log 2M T cp
N ( 4N 0 + 10E s ) NE s ( 8 + T cp R tot ) 16E s ( N n ) J 0 2 f D n --------------- --------
R tot N
n=1
where y is the power attenuation of the channel with an exponential distribution of

mean 1, i.e. p(y)=e-y. In figure 2 the bit error rate for coherent detection is presented
for four different Es/N0 at 50 Hz Doppler frequency and a total bit rate of 320, 640
respectively 1280 kb/s. The reference curve represents coherent detection in a Ray-
leigh channel without losses due to cyclic prefix or ICI.
A cyclic prefix of duration 10 s is chosen to combat time dispersion, a duration

long enough to exceed the expected maximum excess delay of the channel. In this
paper a COST 207 Typical Urban channel [8] is considered with an maximum
excess delay of 5 s. In figure 2 the bit error rate is shown insensitive to bitrate
changes but the effect of the ICI-noise and the energy loss due to the cyclic prefix is
apparent. On the right hand side few sub-channels are used, the symbol time is short
and the length of the cyclic prefix needed to suppress ISI causes a big loss of symbol
energy. On the left hand side the symbol time is long, the changes in the channel
characteristics during a symbol becomes significant and ICI arises. The optimal sub-
channel bandwidth is sensitive to the Doppler frequency since a large Doppler fre-
quency results in large ICI-noise and the optimum is pushed toward broader chan-
nels, see figure 3.
3 dB
-1
10
13 dB
-2
10
BER
23 dB
-3
10
33 dB
loss incl.
no loss
-4
10
3 4 5
10 10 10
Figure 2: The BER curves for Rtot=320, 640 and 1280 kbit/s overlap each other
totally in this case. Coherent QPSK is used in a Rayleigh fading channel with Tcp=10
s and fD=50 Hz when the number of sub-channels and signal to noise ratio are var-
ied. The dotted lines represent coherent detection without losses.
3 dB
-1
10
13 dB
-2
10
23 dB
BER
-3
10 33 dB
-4
10 10 Hz
50 Hz
200 Hz
-5
10
3 4 5
10 10 10
Sub-channel bandwidth/Hz
Figure 3: Bit error rate for coherent detection when the Doppler frequency is varied,
fD={10, 50, 200} Hz, Rtot=320 kbit/s and Es/N0= {3, 13, 23, 33} dB. The dotted lines
represent coherent detection without losses.
2.4 Channel estimation
The radio channel corrupts the transmitted signal and in order to make coherent
detection possible we have to know the impact of the channel. Both the amplitude
and phase are corrupted by the fading channel, whose characteristics vary because of
movements of the mobile terminal. In order to keep track of the channel characteris-
tics pilot symbol assisted modulation, PSAM, can be used. This means that known
training symbols are multiplexed into the data stream at certain sub-channels and cer-
tain times. The receiver interpolates the channel information derived from the pilot
symbols and makes channel estimates for the data symbols. Since the pilot symbols
carry no data the pilot density is to be kept at a minimum not to increase the overhead
too much. In order to make it possible for the receiver to achieve nearly maximal
channel information from each of the pilots the pilot pattern can be made balanced
[7]. This means that the spacings between the pilots are approximately the same in
both frequency and time when normalized by the minimum expected coherence
bandwidth and the minimum expected coherence time respectively. Few sub-chan-
nels means that we have to insert several data symbols between the pilot symbols in
the time domain. Many sub-channels gives us the opportunity to use several of the
channels for pilots simultaneously without increasing the pilot density, see figure 4.
Let NT and NF denote the pilot time distance respectively pilot frequency distance
in number of symbols, fDmax denote the maximum expected Doppler frequency (Hz)
and max denote the maximum expected excess delay (s) of the channel, then the con-
dition for balanced design becomes [7]
f Dmax T sub N T max f N F . (8)

1
The overall pilot density is -------------- , and as a rule of thumb the pilot spacings can be
NT NF
chosen as [7]
f f f f
t t t t
N=1 N=2 N=4 N=8
Pilot symbols Data Symbols
Figure 4: Time - frequency distribution of a pilot symbol differs depending on the

number of sub-channels, N, used. Note that the same pilot density, 1/4, is used for the
three alternatives.
NT
f
Data
Pilot
NF
Figure 5: An example of a balanced pilot pattern with pilot time distance NT=5 and
pilot frequency distance NF=4.
1
f Dmax T sub N T = max f N F = --- (9)
4
An example of a pilot pattern with NF=4 and NT=5 is given in figure 5.
With a fixed time-frequency block, edge problems can arise when designing the
pilot pattern. In order to get rid of this problem in the following and in order to sup-
press the bit error rate at the edges, pilots are put on the outer sub-channels respec-
tively time instants and then sufficiently many pilot symbols are evenly distributed
between these edge pilots in frequency respectively time in order not to exceed the
pilot distances given by (9).
To see the effect of the channel estimation only, the bit error rate was calculated
without ICI-noise and energy losses due to the cyclic prefix. For the estimation a
two-dimensional Wiener filter [7] was used, which is optimal in the sense that the
variance of the estimation error is minimized. Furthermore, known Doppler fre-
quency, perfect synchronization, a COST 207 Typical Urban channel [8] and a
matched filter receiver was assumed, see Appendix A for more details. The resulting
bit error rate under these assumptions with a bit rate of 320 kbit/s, 200 Hz Doppler
frequency is presented infigure 6.
The reason why the different pilot patterns virtually do not affect the bit error rate
is explained by the fact that the pilot symbols are placed at approximately the same
frequencies and time instances independent of the sub-channel bandwidth and sym-
bol time used. There are some changes in the bit error rates due to the fact that the
pilot distances can not be below one, due to numerical reasons in the pilot pattern
design and due to the slight increase in the total bandwidth to estimate when few
channels are used, but these changes are rather small.
3 dB
-1
10
13 dB
-2
10
BER
23 dB
-3
10 33 dB
-4
10
3 4 5
10 10 10
Sub-channel bandwidth/HZ
Figure 6: Changes in the bit error rate caused by the pilot pattern, Es/N0={3, 13, 23,
33} dB. As seen, the pilot pattern has virtually no effect on the bit error rate when
balanced design is used.
3 Complete System
In a real system the channel estimation is of course not perfect. The bit error rate
when both channel estimation and the losses due to the cyclic prefix and ICI-noise
are included is lower bounded by the bit error rate for coherent detection in (7) and
the deviation gets smaller as the pilot density increases. The bit error rate for the
same parameters as in last section (Rtot=320 kbit/s fD=200 Hz, Typical Urban chan-
nel, Tcp=10 s) but with the losses included is presented in figure 7.
4 Discussion
As indicated in figure 7 it is possible to optimize the number of sub-channels and

thereby the sub-channel bandwidth by assuming perfect channel estimates. The
resulting optimum when channel estimation effects are included is principally unaf-
fected by the number of sub-channels used. The reason for this is that the variance of
the estimation error, see Appendix A, is approximately the same, independent of the
sub-channel values chosen. This, in turn, means that it is possible to optimize the
number of channels neglecting the effect of channel estimation.
More narrowband sub-channels can be used when terminal movements are slow
3 dB
-1
10 13 dB
-2
10 23 dB
BER
-3
33 dB
10
est. incl.
coherent
-4
10
3 4 5
10 10 10
Sub-channel bandwidth/Hz
Figure 7: Bit error rate for coherent detection and for the total system when channel
estimation errors are included, Es/N0={3, 13, 23, 33} dB.
since the channel is not expected to change during the longer symbol time. This
means that the relative duration of the cyclic prefix becomes shorter and that the
bandwidth expansion is kept small. However, the sub-channel bandwidth has to be
optimized on a worst case assumption for the Doppler frequency, otherwise the
increase in bit error rate can become severe for high Doppler values.
The optimal sub-channel bandwidth is unfortunately dependent on the signal to

noise ratio. Higher noise levels call for more narrowband sub-channel since the noise
makes the energy loss due to the cyclic prefix more evident. In the same way as for
the Doppler frequency a worst case design has to be made with respect to the noise
level. It is important to keep the maximal bit error rate as low as possible, even
though this results in a larger BER increase for lower noise levels. Actually, often it is
the performance of a coded system that is interesting and since the coding gives more
distinct optima it is important not to increase the worst case BER.
The effect of frequency offsets is not included in the analysis and this effect
becomes more significant when many narrowband sub-channels are used. The carrier
to interference ratio due to frequency offsets is proportional to f 2 [9], thus moving
the optimal sub-channel bandwidth towards broader channels when frequency offsets
are present. Finally, the calculations are based on the assumption that the sub-chan-
nels are flat fading, which may not be the case when broadband sub-channels are
used in some channels, e.g. hilly terrain.
5 Conclusions
The impact of the number of sub-channels, and thereby the sub-channel bandwidth,
used in a mobile OFDM-system was analysed. An analytical expression for the bit
error rate in Rayleigh fading channels was presented where the impact of the cyclic
prefix and the channel changes during a symbol time are encountered. This BER
showed an optimum for the sub-channel bandwidth to be used depending on Doppler
frequency, length of cyclic prefix and signal to noise ratio. The effect of channel esti-
mation on the optimum was also examined and this showed that the optimization can
be performed on the coherent system, without taking the channel estimation into
account. The excess delay of the channel, high noise levels and efficient usage of the
spectrum call for many narrowband sub-channels while a rapidly moving terminal
call for the use of fewer more broadband sub-channels. Finally it was argued that the
sub-channel bandwidth has to be designed on a worst case assumption for the Dop-
pler frequency and the signal to noise ratio.
Appendix A. System Models
In the following section the channel model, estimation algorithm and bit error rate
calculations are presented in detail.
The time dependent channel impulse response, h ( , t ) , is assumed to be a sum of

reflections, see (A.1) where ( t ) denotes the dirac-function.
N
h ( , t ) = an ( t ) ( t n ) (A.1)
n=1
The tap coefficients, an(t), and the tap delays, n , are chosen according to the COST
207 Typical Urban model in the GSM specification [8]. The transfer functions,
H(f,t), are obtained by the Fourier transform of (A.1) and these are the functions we
want to estimate for the different carriers. These channel transfer functions are
regarded as flat fading.
For the estimation and analysis a two-dimensional Wiener filter [7] is used, which
is optimal in the sense that the variance of the reconstruction error is minimized. The
sampled signal from the matched filter receiver of a pilot symbol at channel m in time
k is given by
2
j ------ ( p 1 )
k k M k
Xm = Es Hm e + nm (A.2)
where Es is the signal energy, p is the signal alternative sent among the M possible
ones and n is white Gaussian noise. The noise terms have a variance given by (7), but
the covariance matrix R ke = E [ n k ( n k ) ] is not diagonal due to the correlation between
the channels introduced by the ICI. The filter coefficients are given by
k k T 1 T
m = ( ( m ) ) (A.3)
where = E [ XX ] is the covariance matrix for the received pilot symbols and
k k k
m = E [ H m ( X m ) ] is a cross-covariance vector between the transfer function to be esti-
mated and the received pilot symbols. The estimates of the channel transfer functions
can then be calculated as
k k T
Hm = ( m ) X (A.4)
where X is a vector of the sampled values from all of the pilot symbols.
Finally, the mean square error of the estimates is calculated as
k 2 k T 1 k
J m = H ( m ) ( m ) (A.5)
and this matrix is in turn used to calculate the QPSK bit error rate for each of the
symbols. The BER is given by [10] [11]
1
P b = --- ---------------------- (A.6)
2 2
2 2
where
2
H
= E s -------------------------------------------------- . (A.7)
2 2 2
( H + e ) ( H + J )
References
[1] Digital broadcasting systems for television, sound and data services. Euro-
pean telecommunications standards institute, prETS 300 744, Valbonne,
France, 1996
[2] Radio broadcasting systems; Digital Audio Broadcasting (DAB) to mobile,
portable and fixed receivers. European telecommunications standards institute,
ETS 300 401, Valbonne, France, 1995
[3] A. Peled and A. Ruiz, Frequency domain data transmission using reduced
computational complexity algorithms, Proc. IEEE Int. Conference Acoustics,
speech, signal processing, Denver, USA, pp 964-967, 1980
[4] S. Hara, M. Mouri, M. Okada and N. Morinaga, Transmission performance
analysis of multi-carrier modulation in frequency selective fast Rayleigh fading
channel, Wireless Personal Communications, nr 2, Kluwer Academic Publish-
ers, Dordrecht, Netherlands, pp 335-356, 1996
[5] M. Wahlqvist, C. stberg, J.-J. van de Beek, O. Edfors and P. O. Brjesson, A
conceptual study of OFDM-based multiple access schemes: Part 1 - Air inter-
face requirements. Technical report Tdoc 116/96 ETSI STC SMG2 meeting no
18, Helsinki, Finland, May 1996
[6] M. Russell and G. L. Stber, Terrestrial digital video broadcasting for mobile
reception using OFDM, Wireless Personal Communications, vol 2, nr 1-2,
Kluwer Academic Publishers, Dordrecht, The Netherlands, pp 45-66, 1995
[7] P. Hoeher, S. Kaiser and P. Robertson, Pilot-symbol-aided channel estimation
in time and frequency, In K. Fazel and G. P. Fettweis, editors, Multi-Carrier
Spread-Spectrum, chapter IV, pp 169-178. Kluwer Academic Publishers, Dor-
drecht, The Netherlands, 1997.
[8] European Telecommunications Standard Institute, GSM 05.05, version 4.6.0,
July 1993.
[9] Y. Zhao and S.-G. Hggman, Sensitivity to Doppler shift and carrier frequency
errors in OFDM systems -- The consequences and solutions, IEEE Vehicular
Technology Conference, Atlanta, USA, April 29 - May 2, pp. 1564-1568, 1996.
[10] S. K. Wilson, Digital audio broadcasting in a fading and dispersive channel,
Ph.D. thesis, Stanford University, USA, 1994.
[11] J. G. Proakis, Digital communications, third edition, McGraw-Hill, New
York, USA, 1995.
Paper C
This part will be published as:
Fredrik Tufvesson, Mike Faulkner and Torleiv Maseng, "Pre-Compensation for Ray-
leigh Fading Channels in Time Division Duplex OFDM systems", Wireless Personal
Communication, 2000.
Kluwer Academic Publishers 2000, reprinted with permission.
Pre-Compensation for Rayleigh Fading
Channels in Time Division Duplex OFDM
systems
Fredrik Tufvesson1, Mike Faulkner2 and Torleiv Maseng1
Abstract
Coherent demodulation results in good detection performance but requires
channel estimation. Fading pre-compensation (precoding) at the transmitter can
lead to low-complex receiver structures with good performance capabilities,
without the need for channel estimation. Time division duplex systems based on
orthogonal frequency division multiplex (OFDM) are particularly suited to this
because intersymbol interference effects can be neglected, simplifying transmit-
ter adjustments. Methods that involve amplitude and/or phase pre-compensation
are compared in terms of resulting bit error rate and increase in peak-to-average
power ratio. Dynamic channels degrade the performance as the block lengths get
longer. For a certain block length the performance degrades below that of tradi-
tional differential decoding. A block length of up to 40 times that used in the
Digital European Cordless Telephone system, DECT, is possible using channel
estimation and ideal Wiener prediction.
1 Introduction
In time division duplex (TDD) systems the same frequency is used both by the base
station and the mobile terminal, but at different time instants. In these systems the
possibility of making predictions of the channel impact is rather straightforward. The
used radio channel is often assumed to be reciprocal, see e.g. [1] [2], since we are
interested in compensating the multipath channel and not the interference. If the
block length is short compared to the speed of change of the channel, then predictions
of the channel influence are likely to be accurate for adjacent blocks and we can
adjust the signal to be transmitted to achieve better performance.
Pre-compensation can be used for different purposes, for example to control the
1 Department of Applied Electronics, Lund University, Box 118,

SE - 221 00 Lund, Sweden, E-mail: Fredrik.Tufvesson@tde.lth.se
2 School of Communications and Informatics, Victoria University of Technology,
PO Box 14428, MCMC, Melbourne, Vic. 8001, Australia
1
2 PRE-COMPENSATION
output power [3], to reduce interference from other users [4], or to modify the trans-
mitted signal in order to reduce intersymbol interference (ISI) [1]. In this paper we
examine the possibilities and limits when using pre-compensation of the channel
impact on the downlink symbols in orthogonal frequency division multiplex (OFDM)
systems. Figure 1 shows the principle of the technique. In OFDM systems it is
information
to be sent mobile
pilot terminal
pilot symbols transmitter
multiplexer
received antenna
information receiver
data received
base symbols information
receiver equalizer
station
pilot
symbols channel estimate
antenna channel estimator
channel
estimate information
to be sent
transmitter pre-compen-
sation
Figure 1: Pre-compensation where the channel influence is compensated in advance

on the downlink.
straightforward to compensate for a time dispersive channel since its impact after the
FFT only is an amplitude attenuation and a phase shift of the data symbols. Channel
estimation is normally a complex operation to perform and every mobile terminal has
to make channel estimates in order to use coherent detection. By pre-compensation it
is possible to move the task of channel estimation to the base station, but still achieve
the performance benefits of coherent detection at the mobile terminal [5]. This means
that cheap, less complex, mobile receivers can be used. We address here the problem
of channel pre-compensation in a simple uncoded OFDM system, but the results can
be applied for other systems where the channel can be modeled as a single complex
tap or together with coding and diversity arrangements.
Compensating channel attenuation means that we transmit more power where the
channel attenuation is high and less power where the channel attenuation is low. This
contradicts the water pouring theorem, which states that more power and higher order
modulation should be used where the channel attenuation is low in order to maximize
PRE-COMPENSATION 3
the transmission capacity. However, here the aim is to use fixed modulation and
enable a low-complex receiver.
A drawback when compensating channel attenuation is also that the amplitude

variations of the transmitted signal become larger. The resulting peak-to-average
power ratio has to be limited. For PSK systems it is not necessary to compensate
amplitude variations and in this paper we show that phase compensation only may be
adequate in these cases. In the paper we are interested in evaluating the performance
limits for the system. We therefore assume perfect time synchronization and ideal
oscillators without phase or frequency jitter. In a real system such impairments natu-
rally degrade the performance and their influence needs to be taken into account.
The paper is organized as follows: In section 2, the system and the model for the
OFDM system are described. In section 3 we then present the three analyzed com-
pensation methods, and in section 4 we derive their performance in case where the
channel is known by the base station. In section 5 the resulting increase in peak-to-
average power ratio is derived for the three methods, respectively. We then proceed
in section 6 with the method where only the phase shift of the channel is compen-
sated and we derive the maximum block length before the performance gets worse
than that of differential detection. Finally, the conclusions are presented in section 7.
2 System
In the paper we use an OFDM TDD-FDMA system with M sub-channels, see Figure
2 for a description of the transmitter and receiver. The OFDM signal is transmitted
through a time dispersive Rayleigh fading channel. It is assumed that the fading is
flat for each of the sub-channels and constant during a symbol interval. We assume
further a cyclic prefix exceeding the maximum excess delay, and that the demodu-
lated signals do not suffer from interchannel interference nor ISI.
Under these assumptions we can use a model for the OFDM system where the data
s1[k] r1[k]
data
S/P IFFT P/S D/A h(t,) A/D S/P FFT P/S

sM[k] rM[k]
n(t)
Figure 2: Block description of the transmitter-receiver chain.

4 PRE-COMPENSATION
signal is transmitted on separate, but correlated, Rayleigh fading channels with addi-
tive Gaussian noise. The transmitted OFDM signal is convolved with the channel
impulse response, h ( t, ) , but after demodulation the convolution only results in a
phase shift, m [ k ] , and a power gain, ym[k], of the data symbols. The sampled com-
plex representation of the received signal on channel m at time k can therefore be
written as [6]
j m [ k ]
rm [ k ] = ym[k ]e sm [ k ] + nm [ k ] , (1)
where sm[k] is the complex representation of the transmitted data signal and nm[k] is
independent samples of white Gaussian noise.
3 Compensation Methods
We evaluate three compensation methods:
1. Phase compensation only, the transmitted phase is altered by m [ k ] radians.

This method gives no increase in the transmitted signal dynamics.
2. Phase and amplitude compensation when y>a2, otherwise no transmission. The
transmitter output is inhibited when the channel gain does not exceed the thresh-
old a2. This saves power when the channel is poor.
3. Phase and amplitude compensation when y>a3, otherwise power limiting. The
transmitter output saturates when the channel gain does not exceed the threshold
a3. This method limits the output peak power to a reasonable level.
In the following the sub-script m is omitted to simplify notation, but every sub-chan-
nel is compensated individually by its own amplitude and/or phase.
The base station makes estimates of the channel amplitude and channel phase e.g.
by using known pilot symbols from the uplink. Then the signal to be transmitted is
compensated. If we let u q[ k ] denote the complex representation of the uncompen-
sated signal alternative q, the complex representation of the compensated signals,
u q [ k ] , becomes
j
( uq [ k ] ) 1 = e u q[ k ] (2)
0 y < a2

( uq [ k ] )2 = 2 j y a2 (3)
------ e u q[ k ]
y
PRE-COMPENSATION 5
3 j
------ e u q[ k ]
a3 y < a3
( uq [ k ] ) 3 = , (4)
3 j y a3
------ e u q[ k ]
y
for the three methods respectively. Here 2 and 3 are power normalization coeffi-
cients. The probability density function of the power gain, y, in a Rayleigh channel
y
with unity mean is p Y ( y ) = e . Hence, if the channel is perfectly estimated and
E q denotes the symbol energy before compensation, then the respective mean trans-
mitted symbol energies are
( E q ) 1 = E q (5)
y
e
( E q ) 2 = 2 E q ------- dy (6)
y
a2
a3
3 y 3 y
( Eq )3 = ------ e E qdy +
a3 -----
y
-e E qdy (7)
0 a3
y
a 3
= 3 E q ----- ( 1 e ) + ------- dy .
1 e
a3 y
a3
In order to maintain the same transmitted mean energy for the three methods, the
power normalization coefficients are consequently chosen as
y
2 = 1 ------- dy
e
(8)
y
a2
y
a3
3 = 1 ----- ( 1 e ) + ------- dy .
1 e
(9)
a3 y
a3
6 PRE-COMPENSATION
4 Performance when the Channel is Known by the

Base Station
The performance limit for each of the given methods is achieved when all downlink
transfer functions are known by the base station and the channel is compensated per-
fectly. This is of course not the situation in a practical case, but it is anyhow interest-
ing to get a lower limit for the resulting bit error rate (BER).
The BER, Pb, when the Rayleigh fading channel is perfectly known is for the first
method the same as for ordinary coherent detection of QPSK signals, i.e. [6]

2E b y
( Pb )1 = Q y ---------- e dy
N0
(10)
0
1 E b N 0
= --- 1 --------------------------
- .
2 1 + E b N 0
Here N0 denotes the density of the noise and E b denotes the bit energy before com-
pensation, E b = E q 2 for QPSK. The BER, when the base station knows the chan-
nel impact, can for the two other methods be calculated as
a2
1 y 2E b y
( Pb )2 = --- e dy +
2 Q 2 ---------- e dy
N0
(11)
0 a2
1 a2 a2 2E b
= --- ( 1 e ) + e Q 2 ----------
2 N0
a3
3 2E b y 2E b y
( Pb )3 = Q y -----
- ---------- e dy +
a3 N0 Q 3 ---------- e dy
N0
(12)
0 a3
1 3 ( E b N 0 ) 2E b 1
- Q 2a 3 + 3 ---------- --- .
= --- + ----------------------------------------
2 a 3 + 3 ( E b N 0 ) N0 2
The resulting BER is shown in Figure 3 together with the curves for differential
QPSK [6].
At low signal to noise ratios phase-only compensation appears to be best for the
majority of power compensating limits. At medium and high signal to noise ratios
PRE-COMPENSATION 7
0
10
1
10
2
10
BER
3
10
4
10 diff.
ph. only
5
10 Tx off
Tx sat.
6
10 5 4 3 2 1 0 1 5 4 3 2 1 0 1
10 10 10 10 10 10 10 10 10 10 10 10 10 10
compensation limit compensation limit
Figure 3: BER for the compensation methods when the channel is known by the base
station, Eb/N0= 20 dB (left) and 10 dB (right). The compensation limit (i.e. a1, a2 or
a3) is the maximum power amplification (attenuation) of the channel to be entirely
compensated.
there is some value in adding amplitude compensation with peak power saturation,
but the question is whether one can afford the increased peak-to-average power ratio.
For the amplitude compensation methods a low compensation limit means that too
much energy is spent on compensating very bad channels, while a high limit means
that the transmitter is saturated or off even when communication normally is advis-
able.
5 Peak-to-Average Power Ratio
The compensation limits, a2 and a3, can be seen as measures of the dynamic range
for the power compensation. In practical measurements the peak-to-average ratio of
the transmitted power is often used, which for the pre-compensation methods is the
product of the peak-to-average ratio of the modulation method, e.g. OFDM, and the
increase due to the compensation. For a single sub-channel the increase in peak-to-
average power ratio can be calculated as 1, 2 a 2 and 3 a 3 for the three methods
respectively, but for an OFDM signal the increase is dependent on the correlation
between the sub-channels and which particular channels that are compensated. If we
compensate the phase only, then the transmitted signal is a sum of sinusoids with dif-
ferent phases and there is no increase in the peak-to-average power ratio. If we com-
pensate the amplitude as well, then the transmitted signal is a sum of weighted
sinusoids and we get a larger peak-to-average power ratio. In Figure 4 we present
8 PRE-COMPENSATION
10
ph. only
8
Tx off
peaktoaverage ratio (dB)
Tx sat.
6
high corr. full corr.
4
low corr.
2
3 2 1 0 1
10 10 10 10 10
compensation limit
Figure 4: Simulated values of the increase in peak-to-average power ratio when

applying the three methods for an OFDM-based wireless-LAN. The dashed lines rep-
resent a channel with no delay spread and fully correlated sub-channels. The dash-
dotted lines represent an indoor channel with short delay spread and high correlation
( rms =50ns), and the solid lines represent an indoor channel with long delay spread
and low correlation ( rms =250ns).
simulated values of the increase in peak-to-average power ratio when applying the
different methods in channels with various delay spreads. We use here the OFDM
based wireless local area network (LAN) specified in ETSI BRAN [7]1 as an exam-
ple. The system uses 52 sub-channels and has a sub-channel bandwidth of 312.5 kHz.
For the simulations we use three different channel models:
A one tap channel, giving full correlation between the sub-channels.
An indoor channel with short delay spread (channel A, rms =50ns), resulting
in large sub-channel correlation.
A channel with long delay spread modeling large open buildings (channel E,
rms =250ns). This gives low correlation between the sub-channels.
The simulated values of the increase in peak-to-average power ratio are achieved as
the difference between the 99.5% peak-to-average power ratio levels for the uncom-
pensated and pre-compensated OFDM signal.
1 This system is similar to IEEE 802.11a standardized in the U.S.

PRE-COMPENSATION 9
In Figure 4 we see that we get a larger increase in peak-to-average power ratio if the
sub-channels are correlated, i.e. the delay spread is short. A large peak in the trans-
mitted signal is achieved when the sub-channels contributing to the peak are ampli-
fied due to the amplitude compensation. When the correlation is small, the increase
becomes smaller. This is because the probability that all sub-channels contributing to
the peak are amplified becomes lower. The simulations are performed for a specific
system, but similar results can be expected for other OFDM systems if we take into
account the relation between the coherence bandwidth of the channel and the OFDM
bandwidth.
Phase-only compensation produces no increase in peak-to-average power ratio

and for known channels we get the same performance as for coherent detection. If it
is possible to increase the peak-to-average power ratio by 2 dB, we see in Figure 4
that we can use a3=0.15 for the method where the transmitter is saturated. As seen
from Figure 3, it is then possible to decrease the BER 2-3 times compared to phase-
only compensation. However, it is doubtful whether one can afford to increase the
peak-to-average power ratio for wireless OFDM applications, since power variations
already are a problem in these systems.
6 Maximum Block Length
In practice the transfer functions are not known, but they have to be estimated e.g. by
transmitting known so-called pilot symbols. The impact of the channel on the pilot
symbols is observed, and the channel coefficients for the data symbols are achieved
by filtering and prediction. Different alternatives can be used to form these channel
estimates. Here we use a two-dimensional Wiener filter, which uses both the correla-
tion in time and in frequency to derive the channel estimates. The Wiener filter is
optimal in the sense that the mean squared error is minimized. For further details
about channel estimation in OFDM systems see e.g. [8].
The channel predictions for the downlink block become obsolete after a certain
time. This means that the length of the downlink block is limited before the BER
reaches an unacceptable level. This particular level is dependent on the application.
Here we compare the BER to that of differential detection and use the latter as a
benchmark since the complexity of the mobile terminals is almost the same. The
maximum block length is a function of the Doppler frequency, fD, in the system since
it determines the speed of change in the channel. In the following we assume a classi-
cal U-shaped Doppler spectra such that the time correlation can be described by a
zeroth-order Bessel function. We also concentrate on the performance of the phase-
only compensated system, since there is no increase in peak-to-average power ratio.
In order to investigate the influence of the observation time we study the performance
of a single sub-channel in the OFDM system. We use almost noise free pilot symbols
10 PRE-COMPENSATION
uplink downlink
observation time, t1 prediction time, t2
pilot symbol data symbol
Figure 5: Block structure to investigate the influence of the observation time.
for channel estimation where the observation time t1, the time between the first and
last pilot symbol, is varied, see Figure 5. The low-noise pilot symbols are used in the
analysis since we want to separate the impact of the time varying channel. In the
analysis we have not compensated the symbol energy of the data symbols due to the
boosted pilots. For long and medium block lengths the performance is limited by the
time varying channel and not by the noise. Besides, in a well designed OFDM system
the noise affecting the pilot symbols are averaged over several symbols when calcu-
lating the channel estimates. Therefore, the variance of the channel estimation error is
much lower than the variance of the noise affecting data symbols.
In Figure 6 the BER as a function of prediction time t2, the time since the last
pilot symbol, is presented for different observation times. All the pilot symbols, are
0
10
fDt1=0
fDt1=0.01
fDt1=0.05
1
10 fDt1=0.30
BER
2
10
3
10
0 0.05 0.1 0.15 0.2
prediction time (fD*t2)
Figure 6: BER as a function of prediction time, fDt2, for different observation times,
fDt1. Eb/N0=20 dB for the dashed lines and Eb/N0=10 dB for the solid ones. The dash-
dotted lines represent differential detection.
PRE-COMPENSATION 11
equally distributed within the observation time t1. In this example pilot symbols at
P=4 different time instants are used for prediction. The signal to noise ratio, Eb/N0, is
set to 20 dB and 10 dB respectively, except for the pilot symbols where we use a 30
dB higher signal to noise ratio. We assume that the Doppler frequency is known and
use a Wiener filter for estimation and prediction of the channel. The BER is derived
2
from the mean squared error, e , of the estimate. For QPSK it is given by the corre-
lation coefficient, , between the backrotated received samples and the channel esti-
mate as [6, appendix C]
1
P b = --- ---------------------- . (13)
2 2
2 2
The correlation coefficient when using pilot symbols can be calculated as
Es y
est = --------------------------------------------------------- , (14)
2
( Es y + N0 ) ( Es y + e )
where y = E [ y ] is the mean power of the channel. The BER for DQPSK is also
given by (13), but with
Es
diff = ------------------ . (15)
Es + N0
The BER of the downlink block is determined by the ability of making correct
channel predictions. This ability is dependent on the location of the pilot symbols in
the uplink block, especially their spreading in time, the observation time. Longer
observation time means in general better predictions. An observation time of zero
corresponds to keeping the last channel estimate in the uplink block as a channel pre-
diction for the whole downlink block. By just adding some information about the
trend of the channel values, a considerable decrease in the BER could be achieved.
But the enhanced predictions are naturally dependent on the validity of the channel
model and the ability to suppress the noise affecting the pilot symbols.
At a certain prediction time, the differential system and the pre-compensated sys-
tem have the same bit error performance, see Figure 6. This point, the break even
point, is evaluated in Figure 7 as a function of the observation time and the number of
used pilot symbols using the same assumptions as before. For block lengths smaller
than the ones in Figure 7 the (instantaneous) BER is lower for a pre-compensated
system than for a differential system. Beyond the break even point, differential detec-
tion without pre-compensation gives better results for similar receiver complexity. As
an example, look at Figure 6, Eb/N0=10 dB and the curve for observation time
fDt1=0.05. After a certain time fDt2=0.17 the BER is the same as for a differential sys-
12 PRE-COMPENSATION
0.35 P=1
P=2
0.3
P=3
P=4
0.25
P=5
break even (fD*t2)
0.2
0.15
0.1
0.05
0
0 0.1 0.2 0.3 0.4
observation time (f *t )
D 1
Figure 7: Break even time where the BER for a pre-compensated system starts to
exceed the one for a differential system, Eb/N0= 10 dB. P is the number of time
instants in the uplink block used for pilot symbols.
tem. This point can be found in Figure 7. For P=4 and an observation time of 0.05 the
break even point is fDt2=0.17. In Figure 7 we clearly see the effect of the observation
time. By using all the pilot symbols for prediction and not only the last estimate there
is an increase in the break even time of at least a factor 3. We also see the importance
of using sufficiently many pilot symbols to ensure that there always are pilot symbols
with high correlation to the point to be predicted. Pilot symbols at 3 or 4 time instants
in the uplink block seems enough for this.
To compare the above block lengths to an existing TDD system we can use the
Digital European Cordless Telephone (DECT) system, which has a block length of
368 s [9]. For a terminal at pedestrian speed (10 km/h, 1900 MHz carrier frequency)
this results in a normalized block length of fDt=0.0065. With pre-compensation,
Eb/N0=10 dB, P=4 and a known Doppler frequency the break even time becomes
fDt2=0.26 (when the observation time is equal to the break even time). This means
that it is possible to use approximately 40 times longer block lengths than the one in
the DECT system before the BER becomes worse than that of differential detection.
In a real system the statistics of the time variations may not be known and one
often avoids the requirement of measuring Doppler frequency. The estimation and
prediction filters are therefore often adjusted to the worst case scenario for the chan-
nel changes. This of course decreases the feasible block lengths in the system. In Fig-
ure 8 the break even points are given when the Wiener filter is designed for k times
PRE-COMPENSATION 13
0.15
k=1
k=2
k=4
k=7
break even (fD*t2)
0.1 k=10
0.05
0
0 0.02 0.04 0.06 0.08 0.1
observation time (f *t )
D 1
Figure 8: Break even time when the Wiener filter is adjusted to k times the actual
Doppler frequency and pilot symbols at P=4 time instants are used for prediction,
Eb/N0= 10.
higher Doppler frequency than the actual one. If a pedestrian for example reduces the
speed to 2.5 km/h when the Wiener filter is designed for 10 km/h, the mismatch fac-
tor k is 4 and the break even time is reduced from fDt2=0.18 to 0.07. But this means
on the other hand that absolute block length measured in number of symbols or the
time t2 is longer in the slow but mismatched case since the Doppler frequency is
smaller there, ( t 2 ) 2.5 km/h 0.016 s instead of ( t 2 ) 10 km/h 0.010 s. The feasible
block length in number of symbols is in the general case still determined by the worst
case Doppler frequency and therefore mismatch is not regarded as a problem.
7 Conclusions
The BER when applying different methods of pre-compensating QPSK signals in a

Rayleigh fading channel is presented. Pre-compensation can be used to achieve a per-
formance better than that of traditional differential detection without using channel
estimation and equalization in the mobile receivers. Phase-only compensation is pro-
posed since there is no increase in the peak-to-average power ratio. It is shown that
block lengths 40 times longer than the one in the DECT system could be used, and
anyhow get a performance better than in differential systems. The block length analy-
sis was made for a single sub-channel assuming low-noise pilot symbols. This
assumption is not as restrictive as one may think, since many pilot symbols normally
are used in OFDM systems to form the channel estimates. The correlation in time and
in frequency results in noise averaging and therefore the channel estimates become
14 PRE-COMPENSATION
accurate. For unknown Doppler frequencies where the filters are adjusted to the
worst case scenario, the feasible block lengths are determined by the maximum Dop-
pler frequency, though the filters are mismatched for lower Doppler frequencies.
References
[1] A. Masoomzadeh-Fard and S. Pasupathy, Combined equalization and differen-

tial detection using precoding, IEEE Transactions on Communications, vol.
45, no. 3, pp. 274-278, 1997.
[2] W. Zhuang and V. Huang, Phase pre-coding for frequency selective Rayleigh
and Ricean slowly fading channels, IEEE Transactions on Vehicular Technol-
ogy, vol. 46, no. 1, pp. 129-142, 1997.
[3] H. Matsuki and H. Takanashi, Theoretical analysis of BER improvement by
adaptive transmitter power control for Rayleigh fading compensation, Elec-
tronics Letters, 29, (17), pp. 1520-1521, 1993.
[4] I. Ghauri and R. Knopp, Power control for diversity reception in time-division
duplex CDMA, Proc. of IEEE Vehicular Technology Conference, Ottawa,
Canada, pp. 2343-2347, May 1998.
[5] P. S. Henry and B. S. Glance, A new approach to high-capacity digital mobile
radio, The Bell system technical journal, Vol. 60, No. 8, pp. 1891-1904, 1981.
[6] J. G. Proakis, Digital communications, third edition, McGraw-Hill: New York,
USA, 1995.
[7] European Telecommunications Standards Institute, Broadband Radio Access
Networks (BRAN); HIPERLAN type 2; Physical (PHY) Layer, ETSI TS 101
475 V1.1.1, ETSI, Sophia Antipolis, France, 2000.
[8] P. Hoeher, TCM on frequency-selective land-mobile fading channels, in Proc.
5th Tirrenia International Workshop in Digital Communication, Tirennia, Italy,
pp. 317-328, Sept. 1991; E. Biglieri and M. Luise (eds.), Coded modulation and
bandwidth-efficient transmission, Elsevier Science Publishers: Amsterdam, the
Netherlands, 1992.
[9] J. D. Gibson, editor, The mobile communications handbook, CRC-press: Boca
Raton, Fl, USA, 1996.
Fredrik Tufvesson was born in Lund, Sweden in 1970. He received the M.S. degree
in Electrical Engineering from Lund University in 1994, the Licentiate Degree in
1998 and he is currently a Doctoral student at the Department of Applied Electronics,
Lund University. His research interests include channel estimation and synchroniza-
PRE-COMPENSATION 15
tion problems for wireless communication, especially in wireless OFDM systems.
Mike Faulkner received the B.S.E.E. from Queen Mary College, London Univer-
sity, U.K., the M.E. degree from university of New South Wales, Australia and the
Ph.D. degree from the University of Technology, Sydney in 1970, 1978 and 1993
respectively. Between 1988 and 1996 he spent three periods at Lund University, Swe-
den, the last time as a Visiting Professor working in radio hardware design. Since
1977 he has been a Lecturer and then Professor at Victoria University of Technology,
Melbourne, Australia. His current interests are signal processing, radio circuit design
and spread spectrum.
Torleiv Maseng is Director of Research at the Norwegian Defence Research Estab-

lishment where he is responsible for communications and information systems. He
has worked as a scientist at SINTEF in Trondheim where he was involved in design
and standardization of GSM for ten years. He has also been scientist at NC3A, former
STC, which is a NATO research centre in The Hague, for seven years. During 1992-
94 he was the technical manager for the new private mobile operator in Norway.
Since 1994 he holds a chair in radio communications at Lund University, Sweden.
This work is down to 20% since 1996, when he took up his employment at the Nor-
wegian Defence Research Establishment outside Oslo. He is the author of more than
100 papers, holds patents and is a Technical Editor of the IEEE Communications
Magazine. He has received an award for outstanding research, has arranged large
international conferences and is quoted in Who is Who in the World.
16 PRE-COMPENSATION
Paper D
This part has been submitted as:
Fredrik Tufvesson and Ove Edfors, "Preamble-based Time and Frequency Synchro-
nization for OFDM Systems", submitted to IEEE Journal on Selected Areas in Com-
munications, January 2000.
Preamble-based Time and Frequency
Synchronization for OFDM Systems 1
Fredrik Tufvesson, student member, IEEE, and
Ove Edfors, member, IEEE
Department of Applied Electronics, Lund University, Sweden
Abstract
Fast and reliable time and frequency synchronization is crucial for packet-
based orthogonal frequency division multiplex (OFDM) systems. Various syn-
chronization methods are analyzed to determine their performance when using
preambles based on repeated short pseudo noise (PN) sequences or conventional
preambles based on repeated OFDM data symbols. We make analytical evalua-
tions for AWGN channels and simulate the performance for one- and two-tap
Rayleigh fading channels. Conventionally, the synchronization signal is calcu-
lated as a sum of products between received samples. However, it is shown that
switching the order of summation and multiplication improves the detection per-
formance in terms of lower probability of false detection and of missing the syn-
chronization signal. The false detection probability is decreased by several
orders of magnitude for reasonable parameter settings. The performance in terms
of frequency-offset estimation is similar for both preambles. The PN-based pre-
ambles have low peak-to-average power ratio and they make it possible to use
analog-to-digital converters with only one-bit quantization in stand-by mode.
Therefore, the PN-based preambles allow for a great reduction in stand-by mode
power consumption.
1 Introduction
There is great interest in using orthogonal frequency division multiplex (OFDM) for
high-speed wireless local area network applications. Standardization of systems
where OFDM is used is occurring both in the United States [1], Japan [2] and Europe
[3]. The systems are based on packet data transmission and therefore fast and reliable
time and frequency synchronization with low overhead is crucial to the systems.
Often, a repeated OFDM data symbol is proposed for synchronization issues in

OFDM systems, see e.g. [4] [5]. The synchronization signal is achieved by correlat-
1 Parts of this work have been presented at the IEEE Vehicular Technology Confer-
ence, Amsterdam, The Netherlands, September 1999.
1
2 PREAMBLE-BASED SYNCHRONIZATION
ing samples one symbol or half a symbol apart. The timing signal is then the ampli-
tude of the correlator output. Time synchronization is achieved when the amplitude
exceeds a certain threshold. The frequency-offset estimate is then obtained from the
phase of the same signal. When using a cyclic prefix the timing signal has a flat
region, which causes an uncertainty about where the actual start of the packet is [6].
On the other hand, if no cyclic prefix is used, the frequency-offset estimate is cor-
rupted by intersymbol interference. This has to be suppressed by weighting the
received samples [7].
Synchronization techniques like this, where the synchronization signal is calcu-

lated by correlation, require heavy computations while less complex solutions are
favorable. Alternative methods resulting in lower complexity include e.g. the use of a
preamble, which uses only half of the OFDM spectrum for transmission [8]. The syn-
chronization signal is then achieved as the ratio between the powers in each half of
the frequency band. This method works well for frame synchronization at medium
and high signal-to-noise (SNR) ratios, but gives only a coarse time synchronization
signal and no frequency-offset estimate. There is still a timing uncertainty and, for
low SNR values, the probability of missing the synchronization signal is large.
In CDMA systems synchronization is generally achieved by correlation to a

known code sequence. The correlation means that the influence of noise and interfer-
ence is reduced so that synchronization can be achieved at very low SNR values. Fur-
ther, since the transmitted signal is a single carrier signal, the power variations can be
made low. However, the calculation of the frequency-offset estimate is not as
straightforward as in OFDM systems [9].
In this paper we propose a synchronization technique for OFDM systems that

combines the advantages of time synchronization in CDMA systems with those of
the simple frequency estimation scheme in OFDM systems. We propose a preamble
that consists of a known short repeated pseudo noise sequence (PN-sequence) to be
used together with a synchronizer that correlates received samples before processing
them further. The advantages of this technique compared to conventional OFDM
synchronization are:
The synchronization signal has a sharp synchronization peak and no ambiguity

due to the cyclic prefix.
There are low power variations within the preamble, which means relaxed
requirements on the automatic gain control (AGC).
It is possible to use analog-to-digital converters with only one-bit quantization for

coarse time synchronization. This means low power consumption in stand-by
mode.
PREAMBLE-BASED SYNCHRONIZATION 3
The amplitude of the synchronization signal is small when the timing is wrong
compared to the amplitude when the timing is correct. This means low probability
of false detection and of missing the correct synchronization signal.
To our knowledge, this is a novel approach for OFDM system synchronization and
no analysis of the statistics of the synchronization signal has been made before. The
idea of using one-bit quantization has been presented before for conventional OFDM
synchronization [10] but there the quantization loss was large. By using the technique
presented here this loss becomes low enough to make a one-bit solution attractive.
The paper is organized as follows: In section 2 the synchronization system is

described. In section 3 we analytically analyze the performance in AWGN channels.
In section 4 the performance for two-tap Rayleigh fading channels is investigated by
means of simulations and, finally, in section 5 we present the conclusions.
2 System Description
The preambles are designed for an OFDM system with M sub-channels. The aim is to
estimate the start of the packet and achieve a reliable frequency-offset estimate using
a single-symbol preamble, but other preamble lengths are of course also possible. We
compare two preambles, a PN-based and a conventional one. The comparison is per-
formed using two alternative synchronizer structures, the conventional OFDM syn-
chronizer and a pre-correlating synchronizer.
2.1 Preamble Structures
The PN-based preamble was derived from a R/2 chip long m-sequence, where R
denotes the repetition length of the preamble. The sequence was oversampled twice
and repeated until the total length became 2R+G samples, as shown in Figure 1,
where G denotes the length of the guard interval. In order to decrease the power vari-
R R G
AGC m-seq m-seq guard
G R R
AGC guard OFDM OFDM
Figure 1: Evaluated preambles and their length in samples.

ations within the preamble we used an offset QAM constellation for transmission
where the quadrature sequence was delayed one sample compared to the inphase
sequence. The preamble can be generated using the FFT of the PN-sequence as input
to the OFDM modulator so no extra device is required for preamble generation.
The conventional preamble consists of a repeated OFDM data symbol as shown

in Figure 1. The total preamble length was 2R+G samples, where R normally is half
the OFDM symbol length, i.e. R=M/2. The data symbols were chosen to give good
autocorrelation properties and the repetition of the R samples was achieved by using
only the even sub-channels for transmission.
2.2 Synchronizer Structures
The main parts of the pre-correlating synchronizer are presented in Figure 2. First,
pre-correlation of length K with the known sequence is performed. When the
received sequence coincides with the known sequence, a large peak occurs with a
phase angle determined by the channel. When there is no synchronization signal, the
output is noise only. The correlator output is multiplied by the conjugate of the output
delayed PR times. The positive integer P denotes an extra delay, which can be used to
decrease the variance of the frequency-offset estimate. Most often no extra delay is
used and P=1. Correlation peaks spaced PR samples apart produce a large synchroni-
zation signal with a (small) phase shift proportional to the frequency-offset. The
maximum frequency offset is inversely proportional to the pre-correlator length. If
the pre-correlator length is limited, e.g. due to a large frequency offset, and the varia-
tions of the synchronization signal is too large it is possible to sum consecutive peaks
to decrease the variance. However, normally a long correlator is preferred and no
summation is necessary.
samples
K
| |2
cK-1 cK-2 c3 c2 c1 c0
x
timing
( )*
PR
freq. est.
arg
[ k, a ]
Figure 2: Structure of the pre-correlating synchronizer when using only one pre-corr-
elator product for synchronization (L=1).
The conventional synchronizer multiplies samples spaced R samples apart and

uses a sliding window to sum R consecutive products. This synchronizer structure is
presented in Figure 3. When the repeated signal is present, the phases of the products
within the window are nearly the same and they essentially add in phase. Signals that
are not repeated generally result in small products with random phases. Then the
products do not add in phase and the synchronization signal becomes small.
For both methods the absolute value of the synchronization signal is compared to
the short time mean of the received power. If a threshold is exceeded, time synchroni-
zation is achieved and the frequency estimate is available from the phase of the syn-
chronization signal.
2.3 Synchronization Signal
For the analytical performance evaluation we consider an AWGN channel. The base
band representation of the received signal is
2 c t
r 0 ( t ) = s 0 ( t ) exp j -------------- + + n 0 ( t ) , (1)
T
s
where n0(t) is white Gaussian noise with one-sided spectral density N0, and s0(t) is
the transmitted preamble. Further, is the carrier phase and c = f T s is the fre-
quency-offset to be estimated when normalized by the sub-channel spacing 1/Ts.
The synchronization signals from the two synchronizer structures are represented
in a common framework. Let r[k] denote the sampled received signal and c[k] the
sequence used for pre-correlation. If the pre-correlating synchronizer is used then the
pre-correlator length is equal to the repetition length, i.e. K=R. If no pre-correlator is
samples
PR
| |2 ( )*
x
timing
freq. est. arg [ k, a ] x
Figure 3: Structure of the conventional OFDM synchronizer.

used, as for conventional OFDM synchronization, then K=1 and all c[k]=1. Further,
let a denote the slip between the received sequence and the locally generated pre-cor-
relation sequence. In the following analysis this slip is assumed to be an integer. The
received sampled signal is multiplied by the conjugate of the known pre-correlation
sequence to get a remodulated signal, which is summed over length K to get the
pre-correlator output. The remodulated signal can be calculated as
c [ k a ] r [ k ] (2)
2 c k
= s c [ k a ] s [ k ] exp j --------------- + c + n [ k ] ,
M
2
where * denotes complex conjugation, s is the power of the transmitted signal, M is
the number of sub-channels in the OFDM system, n[k] is white Gaussian noise with
2
variance n and s[k] is the transmitted preamble.
The synchronization signal for both the conventional and the pre-correlating syn-
chronizer is given by
K 1
L1

[ k, a ] = c [ k n lR a ] r [ k n lR ]

(3)
l = 0 n = 0

K 1
c [ k n ( l + P ) R a ] r [ k n ( l + P ) R ] ,

n = 0
where, again, L is the number of products used for the synchronization signal, P is the
extra delay when calculating the products and K is the pre-correlator length. When
using the pre-correlator we often have only one product for synchronization and no
extra delay so L=1 and P=1, without the pre-correlator these parameters are set to
L=R and P=R.
A PN-based preamble permits a very low arithmetic resolution in the synchroniz-

ers due to its low peak-to-average power ratio and good autocorrelation properties.
There is of course a loss in performance but, if for example the length of the pream-
ble is extended, it is possible to have a sleep mode operation where the synchronizer
uses only one-bit resolution for time synchronization. The synchronization unit then
only consists of a threshold detector, a K-input adder, a multiplier and a comparator.
The synchronization signal is still given by (3), but the received signal now only take
the values r [ k ] = 1 j . In this way only the sign of the received signal is used, thus
the receiver does not have to compare the timing signal to the received power and can
instead use a fixed threshold. This decrease in complexity renders the low-resolution
operation especially suitable in acquisition/sleep mode, where low power consump-

tion is important.
3 Performance Evaluation
In this section we analytically analyze the performance of time and frequency syn-
chronization for the two preambles and the two synchronizer structures in AWGN
channels. To compare different parameter choices, we normalize the synchronization
2 2
signal as [ k, a ] = [ k, a ] ( LK s ) . As we will see below, when the timing is
correct the mean of the synchronization signal is almost unity, and when the timing is
wrong the mean is close to zero.
In subsection 3.1, we analyze the distribution of the timing signal when the timing
is correct, i.e. when a=0 and there is no slip between the received and known
sequence. In subsection 3.2 we derive the distribution of the timing signal when the
timing is wrong. By wrong timing we mean that there is no synchronization signal
present and the synchronizer is fed by white Gaussian noise only. We concentrate on
the cases where the synchronization signal consists of one, two or a large number of
products. Normally, only one product is used together with the pre-correlating syn-
chronizer since this results in the best performance. Further, in subsection 3.3, we
analyze the performance of time synchronization when using the low-resolution
mode. In subsection 3.4 we summarize our results on time synchronization and
finally, in subsection 3.5, we analyze the performance of the frequency-offset esti-
mate. For the figures and numerical examples in this section we use an AWGN chan-
nel with 0 dB SNR and a preamble length of 60 samples (R=30), except for the
evaluation of the frequency-offset estimate where we have other SNR values and the
preamble length is 64 for the OFDM-based preamble. The extension is used in that
case in order to get a full OFDM symbol as a preamble.
3.1 Time Synchronization, Correct Timing
When the pre-correlator is used, the synchronization signal is calculated as a sum of

L products between delayed pre-correlator outputs. Let [ k, a ] denote the pre-corre-
lator output. If the frequency offset is within the permitted range, c < M ( 2PR ) ,
then [ k, 0 ] can be calculated for both the conventional and the PN-based preamble
as [12]
c K 2 c
[ k, 0 ] = ------- sinc --------- exp j ------------ k + --- + n [ k, 0 ] ,
1 1
(4)
L M M 2
2 2 2
where n [ k, 0 ] is white Gaussian noise with variance = n ( KL s ) and the
2 2
output has been normalized by LK s . Define, in the same way as in [5], inphase
as the phase of the sum,
2 c PR
arg [ k, 0 ] [ k PR, 0 ] -------------------- . (5)
M
L
For reasonable SNR values the noise-times-noise products are negligible and the
inphase part of the synchronization signal [ k, 0 ] is much larger than its quadrature
part. The amplitude, i.e. the timing signal, can therefore be approximated by the
inphase components as
2 c K
--- sinc --------
1
[ k, 0 ] [ k, 0 ] = L M
- + (6)
L
c K 2 c ( k PR + 1 2 )
InPhase n [ k, 0 ] ------- sinc --------- exp j -------------------------------------------------
1
L M M

c K 2 c ( k + 1 2 )
+ ------- sinc --------- exp j ------------------------------------ n [ k PR, 0 ] .
1
L M M

The approximation, [ k, 0 ] , is Gaussian with mean
2 c K
E [ [ k, 0 ] ] = sinc --------- , (7)
M
and variance
2 c K
var ( [ k, 0 ] ) = n sinc --------- ( LK s ) .
2 2
(8)
M
The timing signal is always positive and, since the expected value is large compared
to the variance, the approximation as a Gaussian gives a negligible probability of
negative values. When there is a frequency offset, the mean decreases slightly since
the samples do not add totally in phase. However, this loss is small in the interesting
frequency-offset range. The cumulative density function (CDF) at the correct timing
point can be expressed as
x E [ [ k, 0 ] ]
F [ k, 0 ] ( x ) F [ k, 0 ] ( x ) = 1 Q ------------------------------------ , (9)
var ( [ k, 0 ] )
where Q is the tail probability of a normalized Gaussian variable.
When the timing is correct, the distribution of the timing signal with the conven-
tional synchronizer is similar to that of the pre-correlating synchronizer. However,

since there is no loss due to the phase rotation, the normalized synchronization signal
can in this case be regarded as Gaussian with unity mean,
E [ conv [ k, 0 ] ] = 1 , (10)
and variance
2
n
var ( conv [ k, 0 ] ) = -------------- . (11)
2
LK s
Monte Carlo simulations, not presented here, show a good agreement between the
true values and the approximations.
3.2 Time Synchronization, Wrong Timing
When using the pre-correlating synchronizer, the normalized complex outputs of the
2 2 2
sub-correlators have zero mean and variance = n ( KL s ) . If the synchroniza-
tion signal consists of one product only, i.e. L=1, we can rewrite the timing signal as
[ k, a ] = [ k lR, a ] [ k ( l + P ) R, a ] . (12)
Since the terms are the amplitude of complex Gaussians, they are Rayleigh distrib-
uted with probability density function (PDF)
2x x2
f [ k, a ] ( x ) = ------ exp ------ . (13)
2
2
The two terms in (12) are independent. By using the transformation theorem on their
joint distribution, the PDF of the timing signal can be calculated as

2 2 x 1 2 x 2
f [ k, a ] ( x ) = ------ --- exp ------ b + ----- db (14)
2 b 2 2
b
0
2 2 x
= ------ xk 0 2 ------ ,
2 2
where k0(x) denotes the zeroth order modified Bessel function. The CDF of the tim-
ing signal is, consequently, given by

F [ k, a ] ( ) = f [ k, a ] ( x ) dx = 1 2 ------ k 1 2 ------
2 2
(15)
0

2 2
K s K s
= 1 2 -------------- k 1 2 -------------- .
2
n 2n
For L=2 we are not able to derive exact expressions for the distributions of the
timing metric, but we determine bounds for the CDFs instead. By an upper bound we
mean that if the CDF of A is upper bounded by the CDF of B then F A ( x ) F B ( x ) .
The opposite is true for a lower bound. In appendix A we show that the CDF of
[ k, a ] is lower bounded by the CDF of an Erlang-2 variable X ,
2 2
F X ( x ) = 1 1 + ------- x exp ------- x . (16)
2

2
We also show that the CDF of the timing signal [ k, a ] when L=2 is upper
bounded by the CDF of the maximum, Y , of two independent exponential variables,
2 2
F Y ( y ) = 1 exp ------- y . (17)

2
When the number of terms, L, in the sum is large, the distribution of the timing
signal can be calculated in the same way for the pre-correlating synchronizer and the
conventional synchronizer. In such cases, the distribution of the real and the imagi-
nary part of the timing signal can be approximated as Gaussian due to the central
limit theorem. The real and imaginary part of the normalized timing signal,
Re { conv } and Im { conv } , both have zero mean and a variance
2 2 2
Re { = L( ) 2 . (18)
conv }
The normalized conventional timing signal can therefore be regarded as Rayleigh

distributed with CDF
2 2
x
2 2 2 s
F [ k, a ] ( x ) = 1 exp ---------------------------- = 1 exp x K L ------ . (19)
2
conv
2 Re { } 2n
conv
The timing signal from the pre-correlating synchronizer is generally smaller com-
pared to the same signal from the conventional synchronizer when the timing is
wrong. This in turn means that the false detection probability of the pre-correlating
synchronizer is much lower than that of the conventional synchronizer. In Figure 4

the CDFs of the timing signals from the pre-correlating synchronizer and the and
conventional synchronizer are presented for the cases of correct and wrong timing
and no frequency-offset. By correlating samples before multiplication the mean and
variance of the timing signal are decreased when the timing is wrong. This makes it
possible to use a low detection threshold without a resulting large false detection
probability. Although in the figure there is no difference between the detection per-
formance of the PN-based preamble and the conventional preamble, there is an
advantage for the PN-based preamble when analyzing the case when the timing is
almost, but not entirely, correct. The correlation properties are better for the PN-
based preamble and it results a sharp synchronization peak. This results in more
accurate time synchronization and this property is especially important in multipath
channels.
3.3 Low-Resolution Operation
There is a high correlation between the sign of the received signal and the sign of the
pre-correlating sequence when the timing is correct. The correlation is zero when the
timing is wrong and therefore it is possible to have a low-resolution operation mode
where only the sign of the received signal is used to find a coarse timing signal. The
probability that the sign of the received sample is equal to the sign of the pre-correlat-
ing sequence can, for the I-channel and the Q-channel respectively, be calculated as
2 c k
p I, equal [ k ] = 1 Q 2 ( s n ) cos --------------- + c
2 2
M
(20)
2 c k
p Q, equal [ k ] = 1 Q 2 ( s n ) sin --------------- + c ,
2 2
M
when the timing is correct. Note that if the phase of the received signal is between
and 2 then the probability of equal signs in e.g. the quadrature channel is lower than
0.5. However, the same is true for the second part of the preamble so that the product
will in general be positive anyway. The correlator adds all the equal and unequal
samples and therefore the two complex parts of the correlator output, low res ,
have a binomial distribution. The probability that the real part of pre-correlator has nI
samples of equal signs is given by
K nI K nI
P Re ( equal ( n I ) = ( p I, equal ) ( 1 p I, equal ) (21)
low res ), n I
with a corresponding expression for the imaginary part. For small frequency offsets
and correct timing the real part of the synchronization signal is much larger than the
imaginary part, so the absolute value can be approximated by the real part only. The
0.8
wrong
timing
0.6 correct
CDF
timing
0.4 p.c. L=1
p.c. L=2 u.b.
p.c. L=2 l.b.
0.2
conventional
correct
0
0 0.5 1 1.5
normalized timing signal,
Figure 4: CDFs of the timing signals when using the conventional synchronizer or the
pre-correlating synchronizer (p.c.) with L=1 or L=2. In case of L=2 and wrong tim-
ing we show an upper (u.b.) and lower bound (l.b.) for the distribution.
real part of the synchronization signal is the sum of the products between the two real
parts of the correlator outputs and the two imaginary parts of the correlator outputs.
The correlator output is given by low res = 2n I K + i ( 2n Q K ) . The probabil-
ity function of the products is calculated by adding all the probabilities for values
resulting in a certain product and the probability function, P Re ( ( n ) , of the
low res )
sum is then achieved by convolution. For larger frequency offsets we have a similar
distribution, but the signal is not dominated by the real part.
When the timing is wrong, the input can be seen as random and the probability
that the sign of the received sequence is equal to the sign of the pre-correlating
sequence is PI,equal=PQ,equal=50% for the I and Q-channel, respectively. In this case
none of the parts dominate and we derive bounds for the distribution. The absolute
value of the synchronization signal is larger than the maximum absolute value of the
real and imaginary part of the synchronization signal and smaller than the sum of the
amplitudes of the real and imaginary part respectively. The CDF of the synchroniza-
tion signal is therefore upper bounded by the CDF of the maximum absolute value of
the real and imaginary part and lower bounded by the CDF of the sum of the absolute
values of the real and imaginary parts, i.e.
F Re ( (n) F (n) (22)

low res ) + Im ( low res ) low res
F max ( Re ( ) )(n).
low res ) , Im ( low res
Due to the favorable correlation properties, the low-resolution timing signal

works well also for low SNRs in AWGN channels. In Figure 5 we plot the CDF of
0.8 wrong
timing
0.6
CDF
correct
0.4 timing
0.2
wrong u.b.
wrong l.b.
correct
0
0 200 400 600 800 1000 1200
lowresolution timing signal
Figure 5: CDFs of the low-resolution timing signal. For wrong timing we have a
lower and an upper bound for the distribution.
the low-resolution timing signal using the PN-based preamble and the same parame-
ters as in Figure 4. For wrong timing we show an upper and a lower bound for the
distribution. In the figure it can be seen that the probability is very small that the tim-
ing signal is larger when the timing is wrong than when it is correct.
3.4 Receiver Operation Characteristics
For the timing signals there is a relationship between the probability of missing the
synchronization signal and the probability of false detection, which is dependent on
the chosen detection threshold. We plot this relationship, the so-called receiver oper-
ation characteristics, in Figure 6 for the conventional and pre-correlated timing signal
together with an upper and lower bound for the low-resolution timing signal. Note
that we plot the probability of missed synchronization signal and not, as most often,
the probability of detection vs. the probability of false detection. We see in the figure
that the timing signal from the pre-correlating synchronizer results in much better
performance, compared to the conventional synchronizer. For a probability of missed
synchronization signal of 10-4 we get a false detection probability of 3*10-2 for the
conventional synchronizer while we get 2*10-8 for the pre-correlating synchronizer.
The lower variance for the pre-correlated timing signal when the timing is wrong also
means that it is possible to decrease the threshold for detection without having too
many false alarms. The low-resolution synchronization signal also results in good
detection performance, in AWGN channels it competes well with the conventional
full-resolution synchronizer. When the probability of missed synchronization signal
is 10-4 the false detection probability from the low-resolution synchronization signal
0
10
2
10
missed sync probability
4
10
6
precorr
10 conventional
lowres l.b.
lowres u.b.
8
10 10 8 6 4 2 0
10 10 10 10 10 10
false detection probability
Figure 6: Receiver operation characteristic for the conventional synchronizer, the

pre-correlating synchronizer and the low-resolution synchronizer. For the latter one
we show an upper and lower bound.
is in the order of 10-3. Compare this to the false detection probability of the conven-
tional, but full resolution, synchronization signal that for these parameters is 3*10-2.
In [10] a synchronizer based on the cyclic prefix using an A/D-converter with one-bit
quantization is analyzed. This system is similar to conventional preamble synchroni-
zation but the repetition due to the cyclic prefix is used instead of the repetition in the
preamble. There the low-resolution synchronizer gave a significant loss compared to
conventional full-resolution synchronization. Here the performance of the low-reso-
lution synchronizer is actually better than that of the conventional full resolution syn-
chronizer.
3.5 Frequency-Offset Estimation
The frequency-offset estimate is given by the phase of the synchronization signal

when the correct timing is found. The estimate is calculated as
M
c = ------------------- arg ( [ k, 0 ] ) . (23)
( 2 PR )
This phase increment method gives a (nearly) unbiased frequency estimate of a com-
plex exponential in white noise [11] and can cope with frequency-offsets which are
smaller than c < M ( 2PR ) .
The variance of the normalized frequency estimate can be approximated as [11]

var ( c ) (24)
2 2 2
M
2 1 n 1 n
= -------------------------------- --- ------------------------------------------ + ----------- - .
-----------------------------------------
2 2 2 2
( 4 KLPR ) L s sinc ( c K M ) 2KP 2s sinc 2 ( c K M )
The performance of the frequency-offset estimate is similar for the conventional and
the pre-correlating synchronizer. For low SNRs the variance from the pre-correlating
synchronizer is slightly lower. In this case the effective SNR after the pre-correlator
is often high enough that the second term in (24) can be neglected. In Figure 7 we
plot the theoretical standard deviation of the frequency-offset estimation error
together with simulated results for the conventional synchronizer using the OFDM-
based preamble and the pre-correlating synchronizer using the PN-based preamble.
The preamble length was 60 (K=30, L=1) for the PN-based preamble and 64 (K=1,
L=32) for the OFDM-based one in an AWGN channel with SNR 0 10 and 20 dB,
respectively. The simulated values agree well with the theoretical values. For the pre-
correlating synchronizer at large frequency offsets it is possible to observe the loss
due to the sinc-term in (24). However, as seen in the figure this increase in the stan-
dard deviation is small.
It is possible to use the low-resolution synchronization signal to derive an initial

frequency-offset estimate. Simulations show, however, that the loss compared to full-
resolution estimation is large, especially for high SNRs. For low SNRs the increase in
standard deviation is smaller due to the averaging effect from the noise.
0.4
precorr K=30
0.2 conv L=32
sim conv
0.1
sim precorr
0 dB
standard deviation
0.04
10 dB
0.02
0.01
20 dB
0.004 3 2 1 0
10 10 10 10
frequency offset,
Figure 7: Theoretical and simulated values of the standard deviation of the fre-
quency-offset estimation error. AWGN channel with SNR 0, 10 and 20 dB. Note that
the OFDM-based preamble is slightly longer than the PN-based one.
4 Performance in Fading Channels
In multipath channels the timing signal from the pre-correlating synchronizer has a
peak for each of the resolvable channel taps. The correlation peaks have the same
phase, except for a small distortion due to the noise, and their amplitudes correspond
to the tap-power for each delay. For time synchronization the synchronizer can either
use the largest peak, a sum of the largest peaks or a sum of all peaks within a time
window corresponding to the excess delay of the channel to find the correct timing.
In OFDM systems we want a timing signal that has a peak when the largest part
of the channel impulse response is within the window set by the guard interval. This
is straightforward to achieve with a rectangular window with length as the guard
interval. To get sharp synchronization peaks and to minimize the influence between
peaks it is important to use a sequence with good autocorrelation properties. Mini-
mizing the influence is especially important for the frequency-offset estimate, other-
wise there will be a small increase in the standard deviation of the estimation error.
The PN-based preamble used together with the pre-correlating synchronizer makes it
possible to resolve taps spaced one chip (i.e. two samples) apart.
In order to investigate the performance in multipath and fading channels we sim-

ulated a one-tap and a two-tap Rayleigh fading channel. In case of the two-tap chan-
nel the taps had equal mean power and the delay between them was varied. In Figure
8 we present simulated receiver operation characteristic in a one-tap rayleigh fading
channel for the low-resolution synchronizer, the conventional synchronizer using the
OFDM-based preamble and the pre-correlating synchronizer using the PN-based and
OFDM-based preamble. The preamble length was 60 for the PN-based preamble
(K=30, L=1) and 64 for the OFDM-based preamble (K=1, L=32). The OFDM-based
preamble was also designed to give good correlation properties by choosing the data
symbols properly, the mean SNR was 10 dB and a constant channel during the whole
preamble was assumed.
As before, we compare the timing signals when the timing is correct and when the
timing is wrong so that the input to the synchronizer in the latter case is noise only.
The pre-correlating synchronizer in this and the next figure used the largest peak and
not a sum to find the timing. In Figure 9 we present simulated receiver operation
characteristic for the two-tap Rayleigh fading channel. The delay between the taps
was 4 samples such that we got two fully resolvable Rayleigh fading taps when using
the PN-based preamble with the pre-correlating synchronizer. Due to the delay the
preambles were extended by 4 samples. For a probability of missed synchronization
signal of 10-3 the probability of false detection is here 2*10-1 for the conventional
synchronizer, 10-2 for the low-resolution synchronizer and 7*10-4 for the pre-corre-
lating synchronizer, respectively.
1
10
2
10
3
10
4
10 conventional
lowresolution
precorr, PN
5
10 precorr, OFDM
5 4 3 2 1 0
10 10 10 10 10 10
Figure 8: Simulated receiver operation characteristics for the different synchronizers

and preambles. One-tap Rayleigh fading channel, mean SNR 10 dB, preamble length
60 (PN-based preamble) or 64 (OFDM-based preamble).
1
10
2
10
3
10
4
10 conventional
lowresolution
precorr, PN
5
10 precorr, OFDM
5 4 3 2 1 0
10 10 10 10 10 10
Figure 9: Simulated receiver operation characteristics for the different synchronizers

and preambles. Two-tap Rayleigh fading channel with equal power and 4 samples
delay, mean SNR 10 dB, preamble length 64 (PN-based preamble) or 68 (OFDM-
based preamble) including 4 samples of cyclic extension.
The pre-correlating synchronizer gives better detection performance both for the
one-tap and the tow-tap channel. The reason is again that the mean amplitude is
lower for that timing signal when the timing is wrong. The low-resolution synchro-
nizer also performs well compared to the conventional one. It also gives lower proba-
bility of false detection and lower probability of missing the synchronization signal.
When using the pre-correlating synchronizer the difference between the OFDM-
based and the PN-based preambles is small. Both preambles have good autocorrela-
tion properties and therefore their performance is similar. The PN-based preamble
has a slightly better performance when the threshold is low and the false detection
probability thereby is high. This is because the received samples are weighted
equally for the PN-based preamble while some samples have higher weight for the
OFDM-based preamble. For short preambles this implies that the variance of the tim-
ing signal is slightly smaller when the timing is wrong.
In a real application a two-step procedure for synchronization in Rayleigh fading

channels can be used. The PN-sequence is repeated four times such that the preamble
length becomes 4R+G samples. During the first 2R samples the low-resolution mode
is used to find a coarse timing signal where the received power is unknown. In low-
resolution mode the synchronizer has low complexity and therefore the stand by
mode power consumption becomes low. At the same time as the coarse synchroniza-
tion signal is calculated the AGC is set. In the second step, during the following sam-
ples, fine time synchronization is performed and a frequency-offset estimate is
derived. In this part the synchronization signal is normalized by the received power.
We evaluate the performance of the frequency-offset estimate by comparing the

standard deviation of the estimation error. In Figure 10 the standard deviation is pre-
sented for different frequency offsets using the conventional synchronizer with the
OFDM-based preamble or the pre-correlating synchronizer with the PN-based pre-
amble. It should be noted that also in this example only the maximum peak is used
for estimation. A slightly lower standard deviation for the pre-correlating synchro-
nizer is achieved by using more peaks. The simulations showed that the performance
is similar for both preambles and that the standard deviation is in principle indepen-
dent of the actual frequency offset as long as it is within the permitted range. The
simulations also showed that outliers caused by fading dips dominate the standard
deviation of the frequency-offset estimate.
5 Conclusions
We have analyzed the performance of different preamble-based synchronization

schemes in OFDM systems. We analyzed two synchronizer structures, the conven-
tional OFDM synchronizer and the pre-correlating synchronizer, used together with a
conventional preamble or a PN-based preamble. It was shown that the pre-correlating
0.04 10 dB
standard deviation
0.02
20 dB
0.01
conventional
precorr, PN
0.004 precorr, OFDM
3 2 1 0
10 10 10 10
frequency offset
Figure 10: Simulated standard deviation of the frequency-offset estimation error.

Rayleigh fading channel with two Rayleigh fading taps. Mean SNR 10 (solid) and 20
dB (dashed), preamble length 64 (PN-based preamble) or 68 (OFDM-based pream-
ble) including 4 samples of cyclic extension.
synchronizer results in better detection performance than the conventional synchro-

nizer. By summing received samples coherently before multiplication, as in the pre-
correlating synchronizer, it is possible to decrease the influence from noise or inter-
ference when the timing is wrong. Therefore a lower detection threshold can be used,
which results in better detection performance in terms of lower false detection proba-
bility and lower probability of missing the synchronization signal. By using the PN-
based preamble the peak-to-average power ratio of the transmitted signal becomes
low and the timing signal from the pre-correlating synchronizer shows a sharp syn-
chronization peak. This is in contrast to conventional synchronization, which is based
on repeated OFDM symbols. For frequency-offset estimation the different methods
result in similar performance. Synchronization based on PN-sequence preambles and
the pre-correlating synchronizer offers great power reductions in stand-by mode.
Using this method it is possible to use A/D-converters with only one-bit resolution in
stand-by mode. The low-resolution synchronization signal is independent of the
received power and therefore a coarse time synchronization signal can be calculated
at the same time as the AGC level is set.
Appendix A. Distribution of the Timing Signal, Wrong

Timing, L=2
If L is even, the synchronization signal can be rewritten as
L1
[ k, a ] = [ k lR, a ] [ k ( l + P ) R, a ] (25)
l=0
L21
= d 2k + 1 c 2k + 1 + b 2k + 1 e 2k + 1 .
k=0
All the terms (b, c, d and e) in (25) are independent zero mean Gaussian variables; d
2 2 2
and b are real valued and have variance b = d = E [ bb ] = 2 , whereas c and
e are complex valued with variance of the both complex parts
2 2 2
Re { c } = Im { c } = . Defining new variables as g k = d 2k + 1 Re { c 2k + 1 } and
h k = d 2k + 1 , the joint probability function for g and h becomes
g 2 1 2 1
f G, H ( g, h ) = ------------------------------ exp ----------------- --- ------ h --- .
1 1
(26)
2 Re { c } d 2 h 2
h
Re { c } d
Integrating [13] (3.478.4) over h gives the PDF for the products, i.e.

= --------------------------- k 0 -----------------------
1 g
fG ( g ) = f G, H ( g , h ) d h Re { c } d Re { c } d
(27)

2 2 g
= ---------- k 0 ------------- .
2 2

Naturally this PDF also applies for b 2k + 1 e 2k + 1 and for the corresponding imagi-
nary parts. The characteristic function of the product, g, can be calculated as [13]
(6.671.14)

ig 2 2 g ig
g ( ) = fG ( g )e dg = ------------ k 0 ------------- e dg
2 2
(28)

a
2 1
= ------- ------------------------------------------ .
2
( 2 2 )2 + 2

The PDF of the sum of products, i.e. the real part of , is given by the inverse trans-
form of the product between the L characteristic functions,

L 2
1 2 2 2 2
L
i x
f Re { } ( x ) = ------ -------
2 2 ------2- +

e d . (29)

For L=2, we get a Laplacian random variable with PDF [13] (3.389.5)
1 2x
f Re { } ( x ) = -------------- exp ------------- . (30)
2
2 2

Re { [ k, a ] } and Im { [ k, a ] } are therefore independent exponential variables

and their sum is an Erlang-2 distributed variable. This means that the CDF of
[ k, a ] when L=2 is lower bounded by the CDF of an Erlang-2 variable X ,
2 2
F X ( x ) = 1 1 + ------- x exp ------- x . (31)
2

2
The CDF of [ k, a ] when L=2 is upper bounded by the CDF of the maximum, Y ,
of two independent exponential variables,
2 2
F Y ( y ) = 1 exp ------- y . (32)

2
6 Acknowledgment
The authors would like to thank Prof. Mike Faulkner, Victoria University of Technol-
ogy, Australia, for helpful discussions and suggestions.
References
[1] Institute of Electrical and Electronics Engineers, IEEE P802.11a/D7.0, IEEE,

New York, USA, July 1999
[2] MMAC system architecture, Ver. 1: MMAC Promotion Council, Tokyo, Japan,
March 1998.
[3] European Telecommunications Standards Institute, Broadband Radio Access
Networks (BRAN); HIPERLAN Type 2 Functional Specification Part 1 - Phys-

ical (PHY) layer. <DTS/BRAN030003-1> V0.j (1999-09), ETSI, Sophia Antip-
olis, France, Sept. 1999.
[4] P. Moose, A technique for orthogonal frequency division multiplexing fre-
quency offset correction, IEEE Trans. on Communications, Vol. 42, no. 10,
Oct. 1994, pp. 2908-14.
[5] T.M. Schmidl and D.C. Cox, Robust frequency and timing synchronization for
OFDM, IEEE Trans. on Communications, vol. 45, no. 12, Dec. 1997, pp.
1616-1621.
[6] S.H. Mller-Weinfurtner, J.F. Rler and J.B. Huber, Analysis of a frame- and
frequency synchronizer for (bursty) OFDM, Proc. of IEEE Communications
Theory Mini Conference in conjunction to Globecom98, Sydney, Australia,
Nov. 1998, pp. 201-206.
[7] F. Platbrood, S. Rievers and J. Farserotu, Analysis of coarse frequency syn-
chronization for Hiperlan Type-2, Proc. of IEEE Vehicular Technology Con-
ference, Amsterdam, The Netherlands, September 1999, pp. 688-692.
[8] K. Bruninghaus; M. Radimirsch, Coarse frame synchronisation for OFDM
based wireless communication systems, IEEE International Symposium on
Personal, Indoor and Mobile Radio Communications, 1998, pp. 806-810.
[9] N. Nishinaga; M. Nakagami; Y. Iwadare, A simple synchronization acquisi-
tion method for DS/SS system under carrier frequency offset, IEICE Transac-
tions on Fundamentals of Electronics, Communications and Computer
Sciences, vol.E80-A, no.11; Nov. 1997, pp. 2162-2171.
[10] J.J van de Beek, M. Sandell, M. Isaksson and P.O. Brjesson, Low-complex
frame synchronization in OFDM systems, Proc. of 4th International Confer-
ence on Universal Personal Communications, Tokyo, Japan, Nov. 1995, pp.
982-986.
[11] H. Meyr and M. Moeneclaey, Digital Communication Receivers, John Wiley
& Sons, New York, U.S.A, 1998.
[12] U. Fawer, A coherent spread-spectrum diversity receiver with AFC for multi-
path fading channels, IEEE Trans. on Communications, vol. 42 no. 2/3/4, Feb-
April 1994, pp. 1300-1311.
[13] I.S. Gradshteyn And I.M. Ryzhik, Table Of Integrals, Series, And Products,
Alan Jeffrey, Editor, Academic Press, London, UK, 1994.
Paper E
Fredrik Tufvesson, Mike Faulkner, Peter Hoeher and Ove Edfors, "OFDM Time and
Frequency Synchronization by Spread Spectrum Pilot Technique", 8th IEEE Com-
munication Theory Mini Conference in conjunction to ICC99, Vancouver, Canada,
June 1999, pp. 115-119.
OFDM Time and Frequency Synchronization
by Spread Spectrum Pilot Technique
Fredrik Tufvesson1, Mike Faulkner2, Peter Hoeher3 and Ove Edfors1
1
Department of Applied Electronics, Lund University, Box 118, SE-
221 00 Lund, Sweden, e-mail: Fredrik Tufvesson@tde.lth.se
2School of Communications and Informatics, Victoria University of
Technology, Melbourne, Australia
3Information and Coding Theory Lab, University of Kiel, Kiel,
Germany
Abstract
Correct time and frequency synchronization is important for the perfor-
mance of orthogonal frequency division multiplex (OFDM) systems. We evalu-
ate a system where a pseudo noise sequence (PN-sequence) is used for
estimation of these synchronization parameters. The PN-sequence is superim-
posed on the OFDM signal. The system is evaluated by means of the variance of
the frequency estimation error and the probability for correct timing synchroni-
zation. Both theoretical and simulated results are presented. The proposed tech-
nique is very flexible and works also at low signal to noise ratios. The frequency
offset range has been significantly increased compared to conventional OFDM
synchronizer schemes.
1 Introduction
Time and frequency synchronization are two important issues for the performance of
orthogonal frequency division multiplex (OFDM) systems. Correct frequency syn-
chronization is crucial in order to preserve orthogonality of the sub-channels. Timing
synchronization is necessary in order to locate the cyclic prefix and identify the start
of new packets or frames. Often the synchronization process is divided into two parts,
acquisition and tracking. Burst transmission acquisition includes a continuous search
for new packets, finding the beginning of the symbols as well as the packets and
making a coarse frequency estimate so that demodulation can be performed. For con-
tinuous transmission, acquisition means finding the start of symbols and frames and
making a coarse frequency estimate. In tracking mode the synchronization parame-
ters are continuously updated to keep the performance of the demodulator as good as
possible under changing conditions.
1
2 SYNCHRONIZATION USING SUPERIMPOSED PILOTS
Existing algorithms are often based on correlation of either a repeated [1][2][3] or

a known [4][5] symbol. A correlation peak is achieved when the timing is correct and
then the frequency offset estimate is achieved by the phase shift between certain sam-
ples. The distance between these samples determines the frequency offset range.
When, e.g., the cyclic prefix is used as a repeated symbol the frequency offset range
is limited to half the sub-channel spacing.
This paper presents a new method for acquisition and tracking of OFDM synchro-
nization parameters, which is based on a continuous transmission of a pseudo noise
(PN) sequence. Similar techniques has previously been proposed for synchronization
of single carrier systems [6] and for frame synchronization in OFDM systems [7]. In
this paper we propose and investigate a system for frequency and timing synchroni-
zation, including both symbol and frame synchronization. The system can cope with
very large offsets, up to half the OFDM bandwidth, and gives the frequency estimate
in one step. The synchronization signal has a sharp peak and is independent of the
OFDM parameters used. It can be used for any length of the cyclic prefix and does
not increase the transmitted bandwidth.
2 Spread Spectrum Pilot Technique
The estimation of synchronization parameters is based on transmission of a PN-

sequence which is superimposed on the OFDM information signal. The chip time of
the PN-sequence is equal to or double the sample time in the OFDM system, depend-
ing on spectrum and sampling requirements. The sequences are synchronous and
transmitted in the same band. Figure 1 shows a block description of how the transmit-
ted signal is generated. During acquisition the PN-sequence is sent without transmis-
sion of the information signal. After one or more PN-symbols, when the correct
timing and carrier frequency have been found, the transmission of the information
signal starts and the receiver enters the tracking mode, see Figure 2. In tracking mode
the amplitude of the PN-sequence can be chosen so that the distortion of the informa-
tion symbol is small. The amplitude of the PN-sequence can be adjusted to a specific
application and the synchronization scheme works for low signal to noise ratios. It is
desirable to use a code length equal to or a multiple of the OFDM symbol length
(including the cyclic prefix) so that correlation peaks occur at the symbol boundaries.
data data data PN

OFDM +
scaling PN-sequence
1 c [k ]
Figure 1: Generation of the transmitted signal

SYNCHRONIZATION USING SUPERIMPOSED PILOTS 3
power
PN
t
data
acq. tracking
Figure 2: Structure of the transmitted signal.
For frame synchronization it is sometimes advantageous to use a long code.
3 System Description
In the analysis we assume an OFDM system with M sub-channels, having a sym-

bol time Ts in an AWGN channel. The discrete time representation of the received
base band signal is
r [ k ] = s [ k ] + n[ k ] (1)
2 c k
j --------------- + c
M
= ( s c [ k ] + 1 s d [ k ] ) e + n[ k ] .
where c[k] is the transmitted PN-sequence, d[k] is the OFDM data sequence and n[k]
is white Gaussian noise with spectral density N0. Further, c is the carrier phase and
c = f T s denotes the frequency offset to be estimated when normalized by the
sub-channel spacing. Both the PN-sequence and the data sequence has unity power
and the amount of power used for the PN-sequence is controlled by the code power
2 2
ratio . The received power is given by s E s T s = E [ s [ k ] ] , whereas the
2 2
noise power is given by n ( N 0 T s ) = E [ n [ k ] ] . The received signal is
despread by the local PN-sequence to get
j ( 2 c k M + c )
c [ k a ] r [ k ] = s c [ k a ] c [ k ] e (2)
j ( 2 c k M + c )
+ 1 s d [ k ] c [ k a ] e + n[k ].
For time synchronization we want to find the integer slip, a, between the two codes.
For the purpose of time and frequency estimation, the two latter terms in (2) are act-
ing as disturbances. For data demodulation, no code multiplication is performed in
the receiver and the first and the last terms cause disturbances. The resulting SNR
becomes 2s ( ( 1 ) 2s + 2n ) for the code sequence and ( 1 )2s ( 2s + 2n ) for the
data sequence. In acquisition mode, only the code sequence is sent and the code
power ratio equals one. In tracking mode a low code power ratio is then used in
order to minimize the impact on data transmission.
The despread signals are summed in groups of K samples to get a so called sub-
correlation. The synchronization signal, [ k, a ] , for a given slip, a, is then achieved
as the sum of L products between sub-correlations spaced KP samples apart (P is the
delay in number of correlator lengths),
L1 K 1
[ k, a ] = c [ k n lK a ] r [ k n lK ]

(3)

l = 0 n = 0

K 1
c [ k n ( l + P ) K a ] r [ k n ( l + P ) K ] .

n = 0
Time synchronization is given by the absolute value of this signal whereas the fre-
quency offset estimate is given by the phase. The synchronizer structure is given in
Figure 3. This scheme can be seen as a modification of the schemes in [4] and [5].
Here, the length of the sub-correlators is K instead of 1 as in [5], and the differential
detectors work on the sub-correlator outputs instead of the correlator outputs as in
[4].
4 Timing Estimator Performance
In this section we analyze the distribution of the timing metric in an AWGN chan-
nel. Firstly we analyze the case when the timing is correct and later when it is wrong.
The calculations are quite tedious and therefore we just summarize the results and
verify them by means of simulations. Often it is desirable to have a signal which is
2 2
independent of received power. By normalizing as [ k, a ] = [ k, a ] ( LK s )
we get a timing metric with a mean close to one when the timing is correct and just
above zero when it is wrong.
K1
c [ k n a ] r [ k n ]
r[k]
n=0
freq. est.
( )* arg
L1
timing
z-KP l=0 abs
Figure 3: Block diagram of the estimator.

4.1 Correct Timing
When the timing is correct, a=0, the distribution of the timing metric can be ana-
lyzed in a similar way as in [3]. The frequency offset entails that the chips are not
added totally in phase in the sub-correlator, which leads to an energy reduction. We
assume that no ISI is present and c M . Normalizing the sub-correlator signal by
1
LK s gives the normalized output, [ k, 0 ] , where [4]
2 2
2 c
[ k, 0 ] = ------- sinc ( c K M ) exp j ------------ k + --- + n [ k, 0 ] .
1 1
(4)
L M 2
All differential products then add coherently since they have nearly the same phase.
For medium and high SNR:s, the contribution from the noise vector perpendicular to
the signal vector can be neglected. If we consider the data signal as Gaussian, or if K
is large, [ k, 0 ] can be approximated as a Gaussian variable with mean
2
E [ [ k, 0 ] ] = sinc ( c K M ) (5)
and variance
2 2 2
( ( 1 ) s + n ) sinc ( c K M )
var [ [ k, 0 ] ] = ----------------------------------------------------------------------------- . (6)
2
LK s
As seen in (6) the variance is inversely proportional to LK. The particular choice of
correlator length has small impact on this product if the observation interval is kept
constant. The probability of missing the sync signal is therefore mainly determined
by the observation interval and the equivalent SNR for the PN-sequence.
4.2 Incorrect Timing
When the timing is wrong, a 0 , the sub-correlators do not peak. The sub-corre-
lator outputs, [ k, a ] , can be regarded as Gaussian variables with zero mean. The
timing metric is therefore the absolute value of a sum of products between zero mean
Gaussian variables,
L1
[ k, a ] a0
= a [ k lK, a ] a [ k ( l + P ) K, a ] . (7)
l=0
1 A typo in the original publication has been corrected, the square root was missing.
0.8
0.6
cdf
0.4
K=1
K=16
0.2
K=128
K=256
0
0 0.5 1 1.5
normalized timing metric
Figure 4: CDF for the normalized timing metric when the timing is wrong (solid) and
correct (dashed), SNR=10 dB, =0.1, observation interval 512 samples and
( K, L ) = { ( 1, 256 ), ( 16, 16 ), ( 128, 2 ), ( 256, 1 ) } . Marker values are obtained by
simulations, whereas the lines represent theoretical values.
In appendix A the distribution of the timing metric is analyzed for different choices
of the number of differential signals. In Figure 4 the cumulative distribution function
(CDF) of the timing metric is presented for various correlator lengths, K, and number
of differential signals, L, when the observation interval is 512 samples, SNR=10dB
and =0.1.
4.3 Receiver Operation Characteristics
Depending on how the detection threshold is set we will get a specific probability
of missing the sync signal and a probability of false alarm. In Figure 5 we plot the
receiver operation characteristics (ROC) for different parameter choices. It is of
course desirable to be in the lower left corner where both probabilities are low. Here,
again, we see the importance of using long sub-correlator lengths. By correlating
samples before differential detection we are not so sensitive to single noise samples.
In conventional OFDM synchronization no correlation is performed before differen-
tial detection The proposed technique will therefore give better performance, also
when only used as a preamble in acquisition mode.
5 Frequency Estimator Performance
The frequency estimator structure is the same both in tracking and acquisition
0
10
missed synch probability
2
10
4
10
K=1
K=16
6
10 K=128
K=256
8 6 4 2 0
10 10 10 10 10
Figure 5: Receiver operation characteristics, probability of missing the synch signal

vs. false detection probability. SNR=10 dB, =0.1, ( K, L ) = { ( 1, 256 ), ( 16, 16 ) ,
( 128, 2 ), ( 256, 1 ) } and observation interval 512 samples.
mode, but a two step procedure can be used in order to enhance the performance. In
acquisition mode the sub-correlator length, K, and the number of delay elements, P,
are chosen according to the maximum expected frequency offset. In tracking mode,
initial compensation has already been made and therefore larger K and P can be used
to decrease the variance of the frequency estimation error.
The phase increment method used gives a (nearly) unbiased frequency estimate of a
complex exponential in white noise [8] and can cope with frequency offsets which
are upper bounded by1 c < M ( 2PK ) . The normalized frequency estimate when
the timing is known is given by2
M arg ( [ k, 0 ] )
c = -------- ------------------------------ . (8)
KP 2
The frequency offset estimators in [1], [2], [4] and [5] can, from a performance point
of view, be seen as a special case of the proposed estimator in acquisition mode. See
Table 1 for a comparison of parameters, where G denotes the length of the cyclic pre-
fix.
1 A typo in the original paper has been corrected, the factor 2 was missing.
2 A typo in the original paper has been corrected, 2 was missing.
Table 1: Comparison of parameters for some frequency

offset estimators based on one symbol measurements.
reference K P L
van de Beek [1] 1 M G
Moose [2] 1 M/2 M/2
Nishinaga [5] 1 1 M -1
Fawer [4] M/2 1 1
PN-based M
K < ---------------
M
P < ---------------
M
----- P
2P c 2K c K
The variance of the normalized frequency estimates can be calculated as [8]
2
M
var ( c ) = ----------------------- (9)
2 3
4 K LP
2 2 2 2 2
1 ( 1 ) s + n 1 ( 1 ) s + n
2 2
- + ----------- ---------------------------------------------
--- --------------------------------------------- - .
L
s sinc ( c K M ) 2KP 2s sinc2 ( c K M )
Strictly speaking this is only valid for zero frequency offset, but simulations indicate
[8] that it is a good approximation for other frequency offsets as well. An interesting
property to note is that the same minimum variance can be achieved for different sub-
correlator lengths as long as the number of delay elements are chosen properly.
Figure 6 shows theoretical and simulated values of the variance as a function of

the frequency offset when the estimator is designed for a worst case normalized off-
set of 1, 4 and 10 respectively. Due to the flat region we can conclude that the perfor-
mance is determined by the frequency offset for which the estimator is designed and
that a worst case design has to be made. For large frequency offsets a two mode
approach can be used to decrease the variance. First the parameters are adjusted for
the worst case scenario and after a first coarse estimate has been made and corrected
longer sub-correlators and larger delay can be used. As a rule of thumb the normal-
ized frequency error must not exceed a few percent, see e.g. [2][9]. In order to fulfil
this in tracking mode we need to extend the observation interval or increase the code
power ratio compared to the parameters used in the figure. A possible solution could
be to have correlation length 1 and an observation interval of 3 symbols. Then the
4
frequency error variance becomes 1.3 10 for small offsets in tracking mode.
des=1
1
10 des=4
=10
estimation error variance
des
2
10
3
10
4
10
5
10 3 2 1 0 1
10 10 10 10 10
frequency offset
Figure 6: Variance of the normalized frequency estimation error as a function of the

offset for SNR=10 dB and code power ratio =0.1 (dashed) and =1 (solid). The
sub-correlators are designed for frequency offsets 1, 4 and 10. Marker values are
obtained by simulations.
6 Conclusions
We have presented and analyzed a new approach for synchronization in OFDM

systems. The technique is very flexible and has a large frequency offset range. Off-
sets up to half the OFDM bandwidth can be detected in one step, instead of the com-
mon offset range equal to one or half the sub-channel spacing. The synchronization
signal is achieved by correlating received samples before they are differentially
detected. This makes the synchronizer less sensitive to single noise values and results
in better detection properties compared to conventional synchronizers. The amount
of synchronization signal is controlled by a single parameter and can be tailored to a
specific application. Therefore the presented synchronizer scheme is robust and
works for low signal to noise ratios. The complexity of the synchronizer is low and it
is therefore suitable for implementation.
Appendix A. Distribution of the Timing Metric
In this section we analyze the distribution of the timing metric when the timing is
not correct. We analyze the case with one or two differential signals, L, or when the
number of differential signals is large. The normalized complex outputs of the sub-
2 2 2 2
correlators have zero mean and variance = ( ( 1 ) s + n ) ( KL s ) if an
a
ideal code with perfect autocorrelation properties is used, and variance

2 2 2 2
= ( s + n ) ( KL s ) for a random code.
a
For L=1 we can rewrite the timing metric as
[ k, a ] = a [ k lK, a ] a [ k ( l + P ) K, a ] , (10)
where the two terms are Rayleigh distributed. By the transformation theorem we can
derive the joint PDF for [ k, a ] and a [ k lK, a ] and then integrate over
a [ k lK, a ] to get the probability density function (PDF) for the timing metric.
The PDF for incorrect timing becomes
2 2 d
f [ k, a ] ( d ) = -------- dk 0 2 -------- , (11)
2 2
a a
where k0(x) denotes the zeroth-order modified Bessel function. The CDF of the tim-
ing metric is consequently given by

P { [ k, a ] } = F D ( ) = 1 2 -------- k 1 2 -------- , (12)
2 2

a a
where k1 is the first order modified Bessel function.
For L=2 we derive bounds for the timing metric. If L is even the metric can be
rewritten as
L1
[ k, a ] = a [ k lK, a ] a [ k ( l + P ) K, a ] (13)
l=0
L21
= d 2k + 1 c 2k + 1 + b 2k + 1 e 2k + 1 .
k=0
All the terms (b, c, d and e) in (13) are independent zero mean Gaussian variables, d
2 2 2
and b are real valued and have variance b = d = E [ bb ] = 2 , whereas c
a
and e are complex valued with variance of the both complex parts
2 2 2
Re { c } = Im { c } = . Defining new variables as g k = d 2k + 1 Re { c 2k + 1 } and
a
h k = d 2k + 1 , the joint probability function for g and h becomes
g 2 1 2 1
f G, H ( g, h ) = ------------------------------ exp ----------------- --- ------ h --- .
1 1
(14)
2 Re { c } d 2 h 2
Re { c } h
d
Integrating [10, integral 3.478.4] over h gives the PDF for the products, i.e.

= --------------------------- k 0 -----------------------
1 g
fG ( g ) = fG, H ( g, h ) dh Re { c } d Re { c } d
(15)

2 2 g
= ------------ k 0 ------------- .
2 2

a a
Naturally this PDF also applies for b 2k + 1 e 2k + 1 and for the corresponding imagi-
nary parts. The characteristic function of the product, g, can be calculated as [10,
integral 6.671.14]

ig 2 2 g ig
g ( ) = fG ( g ) e dg = ------------ k 0 ------------- e dg
2 2
(16)

a a
2 1
= -------- -------------------------------------------- .
2
( 2 2 )2 + 2
a a
The PDF of the sum of products, i.e. the real part of , is given by the inverse trans-
form of the product between the L characteristic functions,
L 2
1 2 2 2 2
L
i x
f Re { } ( x ) = ------ --------
2 2 2
-------
- +

e d . (17)
a
a
For L=2, we get [10, integral 3.389.5] a Laplacian random variable with PDF
1 2 x
f Re { } ( x ) = ---------------- exp ------------- . (18)
2
2 2
a a
Re { [ k, a ] } and Re { [ k, a ] } are therefore independent exponential variables

and the sum of them is therefore an Erlang-2 distributed variable. This means that the
CDF of [ k, a ] is upper bounded by the CDF of an Erlang-2 variable X ,
2 2
F X ( x ) = 1 1 + -------- x exp -------- x . (19)
2

2
a a
The CDF of [ k, a ] is lower bounded by the CDF of the maximum, Y , of two

independent exponential variables,
2 2
F Y ( y ) = 1 exp -------- y . (20)

2
a
Finally, if L is large Re { } and Im { } can be approximated as independent

Gaussian variables due to the central limit theorem. In this case Re { } and Im { }
have zero mean and variance
2 2
Re { } = L ( ) 2 .
2
(21)
a
The normalized conventional timing metric is therefore Rayleigh distributed with

CDF

2
P { [ k, a ] } = F ( ) 1 exp -------------------- . (22)
2
2 Re { }
References
[1] J.J. van de Beek, M. Sandell and P. O. Brjesson, ML estimation of time and
frequency offset in OFDM systems, IEEE Trans. on Signal Processing, vol.
45, no. 7, July 1997, pp. 1800-1805.
[2] P. Moose, A technique for orthogonal frequency division multiplexing fre-
quency offset correction, IEEE Trans. on Communications, vol.42, no. 10,
Oct. 1994, pp. 2908-14.
[3] T.M. Schmidl and D. C. Cox, Robust frequency and timing synchronization
for OFDM, IEEE Trans. on Communications, vol. 45, no. 12, Dec. 1997, pp.
1616-1621.
[4] U. Fawer, A coherent spread-spectrum diversity receiver with AFC for multi-
path fading channels, IEEE Trans. on Communications, vol. 42, no. 2/3/4,
Feb.-April 1994, pp. 1300-1311.
[5] N. Nishinaga, M. Nakagami and Y. Iwadare, A simple synchronization acqui-
sition method for DS/SS system under carrier frequency offset, IEICE Trans.
on Fundamentals, vol. E80-A, no. 11, Nov. 1997, pp. 2162-2171.
[6] T.P. Holden and K. Feher, A spread spectrum based system technique for syn-
chronization of digital mobile communication systems, IEEE Trans. on Broad-
casting, vol. 36, no. 3, Sept. 1990, pp. 185-194.
[7] A. Steingass, A.J. van Wijngaarden and W.G. Teich, Frame synchronization
using superimposed sequences, in Proc. IEEE International Symposium on
Information Theory, Ulm, Germany, June 29-July 4, 1997, p. 489.

[8] H. Meyr and M. Moeneclaey, Digital Communication Receivers. New York,
U.S.A.: John Wiley & Sons, 1998.
[9] T. Pollet, M. Van Bladel and M. Monelcaey, BER sensitivity of OFDM sys-
tems to carrier frequency offset and Wiener phase noise, IEEE Trans on Com-
munications, vol. 43, no. 2/3/4, Feb.-April 1995, pp. 191-193.
[10] I.S. Gradshteyn And I.M. Ryzhik, Table Of Integrals, Series, And Products.
Alan Jeffrey, Editor, London, UK: Academic Press, 1994.
Paper F
This part has been submitted as:
Fredrik Tufvesson and Peter Hoeher, "Channel Estimation using Superimposed Pilot
Sequences", submitted to IEEE Transactions on Communications, March 2000.
Fkdqqho Hvwlpdwlrq xvlqj Vxshulpsrvhg Slorw
Vhtxhqfhv
Iuhguln Wxiyhvvrqh dqg Shwhu Krhkhun
h
Ghsduwphqw ri Dssolhg Hohfwurqlfv/ Oxqg Xqlyhuvlw|
Er{ 44;/ VH0554 33 Oxqg/ Vzhghq
Id{= .79079 45 << 7;/ H0pdlo= Iuhguln1WxiyhvvrqCwgh1owk1vh
n Lqirupdwlrq dqg Frglqj Wkhru| Ode/ Xqlyhuvlw| ri Nlho
Ndlvhuvwu1 5/ G057476 Nlho/ Jhupdq|/
Id{= .7<07640::8:50736/ H0pdlo= skCwhfkidn1xql0nlho1gh1
Lqgh{ whupv= Fkdqqho hvwlpdwlrq/ skdvh v|qfkurql}dwlrq/ htxdol}huv/

Ylwhuel ghwhfwlrq/ vhtxhqfhv/ Ud|ohljk fkdqqhov/
Skdvh Vkliw Nh|lqj/ Txdgudwxuh Dpsolwxgh Prgxodwlrq
Devwudfw
Iru wkh sxusrvh ri ydulrxv v|qfkurql}dwlrq wdvnv +lqfoxglqj fduulhu
skdvh/ wlph/ iuhtxhqf|/ dqg iudph v|qfkurql}dwlrq,/ rqh pd| dgg d nqrzq
slorw vhtxhqfh/ w|slfdoo| d svhxgr0qrlvh vhtxhqfh/ wr wkh xqnqrzq gdwd
vhtxhqfh1 Wklv dssurdfk lv nqrzq dv d vsuhdg0vshfwuxp slorw whfkqltxh
ru dv d vxshulpsrvhg slorw vhtxhqfh whfkqltxh1 Lq wklv sdshu/ zh dsso|
wkh vxshulpsrvhg slorw vhtxhqfh whfkqltxh iru wkh sxusrvh ri fkdqqho hv0
wlpdwlrq +FH,1 Zh sursrvh d wuxo| frkhuhqw uhfhlyhu edvhg rq wkh Ylwhuel
dojrulwkp/ dqg ghulyh lwv sdluzlvh huuru suredelolw|1 Wklv dqdo|wlfdo uh0
vxow jlyhv ixuwkhu lqvljkw dqg ohdgv wr dq dssur{lpdwlrq ri wkh elw huuru
udwh/ zklfk lv frquphg e| Prqwh Fduor vlpxodwlrqv1 Ixuwkhupruh/ zh
suhvhqw d jhqhulf orz0frvw uhfhlyhu vwuxfwxuh edvhg rq uhgxfhg0vwdwh vh0
txhqfh hvwlpdwlrq1 Dprqj wkh glvwlqfw dgydqwdjhv frpsduhg wr frqyhq0
wlrqdo slorw0v|pero0dvvlvwhg FH duh +l, qr edqgzlgwk h{sdqvlrq dqg +ll,
d vljqlfdqwo| lpsuryhg shuirupdqfh lq idvw idglqj hqylurqphqwv1 Wkh
sursrvhg Ylwhuel uhfhlyhu pd| dovr eh xvhg dv dq dowhuqdwlyh uhfhlyhu iru
slorw0v|pero0dvvlvwhg FH/ pdnlqj wklv vfkhph pruh urexvw lq idvw0idglqj
fkdqqhov1
W Sduwv ri wklv zrun zhuh suhvhqwhg dw wkh Lqw1 Zrunvkrs rq Pxowl0Fduulhu Vsuhdg0
Vshfwuxp/ Rehusidhqkrihq/ Jhupdq|/ Vhsw1 4<<< dqg LHHH JOREHFRP *<</ Ulr gh
Mdqhlur/ Eud}lo/ Ghf1 4<<<1
4
5 FKDQQHO HVWLPDWLRQ XVLQJ VXSHULPSRVHG SLORWV
4 Lqwurgxfwlrq
Frqvlghu gljlwdo gdwd wudqvplvvlrq ryhu wlph0vhohfwlyh dw0idglqj fkdqqhov1 Dv0
vxplqj frkhuhqw ghprgxodwlrq/ rqh ri wkh pdlq sureohpv lv fduulhu skdvh v|q0
fkurql}dwlrq/ erwk lq whupv ri dftxlvlwlrq dqg wudfnlqj/ sduwlfxoduo| zkhq wkh
fkdqqho lv idvw dqg zkhq d olqh0ri0vljkw frpsrqhqw lv devhqw ^4`^5`1
D srsxodu whfkqltxh wr pdlqwdlq frkhuhqw ghprgxodwlrq iru d zlgh fodvv
ri gljlwdo prgxodwlrq vfkhphv kdv ehhq sursrvhg e| Prkhu dqg Orgjh ^6` ^7`/
dqg lv nqrzq dv slorw0v|pero0dvvlvwhg FH 4 1 Wkh pdlq lghd ri slorw0v|pero0
dvvlvwhg FH lv wr pxowlsoh{ nqrzq slorw v|perov +dovr fdoohg wudlqlqj v|perov,
lqwr dq xqnqrzq gdwd vwuhdp1 Frqyhqwlrqdoo|/ wkh uhfhlyhu uvwo| rewdlqv whq0
wdwlyh fkdqqho hvwlpdwhv dw wkh srvlwlrqv ri wkh slorw v|perov e| phdqv ri
uh0prgxodwlrq/ dqg wkhq frpsxwhv qdo hvwlpdwhv e| phdqv ri lqwhusrodwlrq1
Djkdprkdppdgl hw do1 dqg Fdyhuv zhuh dprqj wkh uvw dqdo|}lqj dqg rswl0
pl}lqj slorw0v|pero0dvvlvwhg FH jlyhq glhuhqw lqwhusrodwlrq owhuv ^8` ^9`1 Gxh
wr wkh slorw v|perov wkh edqgzlgwk voljkwo| lqfuhdvhv1
Lq wklv sdshu/ zh h{soruh d uhodwhg whfkqltxh sursrvhg d gr}hq |hduv djr e|
Pdnudnlv/ Ihkhu dqg Kroghq ^:`^;`1 Wkh whfkqltxh zdv ruljlqdoo| fdoohg vsuhdg0
vshfwuxp slorw whfkqltxh dqg wkh lghd lv wr olqhduo| dgg d nqrzq slorw vhtxhqfh
wr wkh xqnqrzq gdwd vhtxhqfh/ vhh Vhfwlrq 51 Vlplodu phwkrgv duh dssolhg lq
wkh hog ri dxwrpdwlf frqwuro/ zkhuh lqwhqwlrqdo shuwxedwlrq vljqdov duh xvhg/
h1j1/ iru gxdo frqwuro/ forvhg orrs lghqwlfdwlrq dqg h{wuhpxp frqwuro ri g|0
qdplfdo v|vwhpv ^<`1 Iru frppxqlfdwlrq v|vwhpv Pdnudnlv/ Ihkhu dqg Kroghq
dssolhg wkh whfkqltxh wr skdvh v|qfkurql}dwlrq ^:`^;`1 Odwhu/ wkh whfkqltxh kdv
ehhq dssolhg iru wkh sxusrvhv ri iudph v|qfkurql}dwlrq ^43` dqg mrlqw wlph dqg
iuhtxhqf| v|qfkurql}dwlrq ri RIGP vljqdov ^44`/ uhvshfwlyho|1 D nh| ihdwxuh ri
wkh vfkhph lv wkdw wkh slorw lqirupdwlrq dqg wkh gdwd duh wudqvplwwhg lq wkh
vdph glphqvlrq rq wkh vdph fkdqqho1 Dowhuqdwlyh whfkqltxhv pd| eh edvhg rq
wudqvplwwlqj erwk vhtxhqfhv rq vhsdudwh fkdqqhov1 Rqh fdq/ h1j1/ xvh wzr ru0
wkrjrqdo owhuv dv vxjjhvwhg e| Ehmmdql dqg Ehoruh ^45`/ ru xvh dq ruwkrjrqdo
vljqdo glphqvlrq +h1j1 wkh txdgudwxuh frpsrqhqw,/ dv grqh lq wkh XPWV2Z0
FGPD xsolqn ^46`1 Lq erwk fdvhv/ krzhyhu/ wkh edqgzlgwk h!flhqf| zrxog eh
uhgxfhg1
Zlwklq wklv sdshu/ zh dvvxph v|qfkurqrxv v|pero0e|0v|pero vxshusrvlwlrq
ri wkh slorw v|perov/ zkhuh wkh srzhu ri wkh slorw v|perov lv w|slfdoo| pxfk
ohvv wkdq wkh srzhu ri wkh gdwd v|perov1 E| frqvwuxfwlrq/ wkhuh lv qr lqfuhdvh
lq edqgzlgwk zkhq xvlqj d svhxgr0qrlvh slorw vhtxhqfh1 Wkhuhiruh/ zh xvh
wkh qrwlrq ri d vxshulpsrvhg slorw vhtxhqfh whfkqltxh dv vxjjhvwhg lq ^43`/
ehfdxvh vsuhdg0vshfwuxp whfkqltxhv pd| vxjjhvw d zlghqlqj ri wkh vshfwuxp1
Dv rssrvhg wr slorw0v|pero0dvvlvwhg FH/ qr lqwhusrodwlrq lv qhfhvvdu|1 Wkh
vxshulpsrvhg slorw vhtxhqfh whfkqltxh lv wkhuhiruh pruh edqgzlgwk dqg srzhu
4 Fkdqqho hvwlpdwlrq pd| eh vhhq dv d jhqhudol}dwlrq ri fduulhu skdvh v|qfkurql}dwlrq/
vlqfh hvwlpdwhv ri wkh txdgudwxuh frpsrqhqwv fryhu wkh fduulhu skdvh dv zhoo dv uholdelolw|
lqirupdwlrq/ riwhq fdoohg fkdqqho vwdwh lqirupdwlrq1
FKDQQHO HVWLPDWLRQ XVLQJ VXSHULPSRVHG SLORWV 6
uk yk ^
u
Viterbi k-D
detector
fk nk
transmitter channel receiver
Iljxuh 4= Eorfn gldjudp ri wkh wudqvplvvlrq vfkhph1
h!flhqw wkdq slorw0v|pero0dvvlvwhg FH/ sduwlfxoduo| lq idvw idglqj frqglwlrqv/

dv vkrzq lq wklv sdshu1
Lq wklv sdshu zh irfxv rq vlqjoh0fduulhu v|vwhpv dqg iuhtxhqf|0 dw idglqj
fkdqqhov1 Dvshfwv ri pxowl0fduulhu v|vwhpv zhuh suhvhqwhg lq ^47`1 Rxu pdlq
frqwulexwlrq lv wkh ghulydwlrq dqg dqdo|vlv ri d uhfhlyhu vwuxfwxuh iru wkh sxusrvh
ri FH +skdvh v|qfkurql}dwlrq,1 Zh ghulyh wkh rswlpdo phwulf dqg dq dgglwlyh
phwulf zklfk lv vxlwdeoh iru uhfxuvlyh dojrulwkpv1 Zh wkhq dqdo|}h wkh shuiru0
pdqfh erwk dqdo|wlfdoo| dqg lq whupv ri vlpxodwlrqv1 Lq Vhfwlrq 6 wkh uhfhlyhu/
dqg vwuxfwxuhg vlpsolfdwlrqv wkhuhri/ lv suhvhqwhg1 Wkh sursrvhg wudqvplvvlrq
vfkhph lv wkhq hydoxdwhg dqdo|wlfdoo| lq Vhfwlrq 7 dqg e| vlpxodwlrqv lq Vhfwlrq
81 Ilqdoo|/ wkh frqfoxvlrqv duh gudzq lq Vhfwlrq 91
5 Wudqvplvvlrq Vfkhph zlwk Vxshulpsrvhg Sl0

orw Vhtxhqfh
D eorfn gldjudp ri wkh wudqvplvvlrq vfkhph ihdwxulqj wkh wudqvplwwhu/ wkh
fkdqqho dqg wkh uhfhlyhu lv vkrzq lq Ilj1 41 Lq wkh iroorzlqj/ zh xvh frpsoh{
edvhedqg qrwdwlrq dqg dvvxph shuihfw wlph/ iuhtxhqf| dqg iudph v|qfkurql}d0
wlrq1 Lq wkh wudqvplwwhu d nqrzq slorw vhtxhqfh/ w|slfdoo| d svhxgr0qrlvh +SQ,
vhtxhqfh/ lv v|qfkurqrxvo| dgghg wr wkh xqnqrzq gdwd vhtxhqfh1 Dowhuqdwlyho|/
d slorw vhtxhqfh pd| eh dgghg wr wkh skdvh ri wkh gdwd vhtxhqfh/ zklfk pd|
eh wkh ehwwhu fkrlfh iru FSP1 Wkh fklsv ri wkh nqrzq slorw vhtxhqfh dqg wkh
l1l1g1 gdwd v|perov duh ghqrwhg dv sn 5 F dqg xn 5 F/ uhvshfwlyho|/ zkhuh
n lv wkh wlph lqgh{1 Zh dovr dvvxph wkdw wkh fkls gxudwlrq lv htxdo wr wkh
v|pero gxudwlrq1 Wkh uhfhlyhu uholhv rq d orz fruuhodwlrq ehwzhhq wkh slorw
vhtxhqfh dqg wkh gdwd vhtxhqfh lq rughu wr uhvroyh skdvh dpeljxlwlhv1 Wkh
gdwd vhtxhqfh lv dvvxphg wr eh P0du| dqg erwk vhtxhqfhv kdyh }hur phdq=
H^xn ` @ H^sn ` @ 31 Ohw wkh hqhuj| ri wkh slorw v|perov eh H^msn m5 ` @ Hv
dqg wkh hqhuj| ri wkh gdwd v|perov eh H^mxn m5 ` @ +4 , Hv / uhvshfwlyho|1
Khqfh/ wkh dyhudjh hqhuj| ri wkh frpsrvlwh v|perov lv Hv / li wkh vhtxhqfhv duh
pxwxdoo| lqghshqghqw= H^xn sn ` @ 31 Wkh dyhudjh hqhuj| shu lqirupdwlrq elw
lv He @ Hv @ orj5 P 1
Wkh qrupdol}hg srzhu ri wkh slorw v|perov/ / fruuhvsrqgv wr wkh dprxqw
ri nqrzq lqirupdwlrq wudqvplwwhg/ dqg vkrxog eh rswlpl}hg1 Lq slorw0v|pero0
dvvlvwhg FH v|vwhpv wkh uhodwhg sdudphwhu lv wkh vsdflqj ri wkh slorw v|perov1
Lq wkh uhpdlqlqj/ lv dvvxphg wr eh frqvwdqw iru doo n/ vlqfh rxu irfxv lv rq
wudfnlqj1 Lq rughu wr lpsuryh wkh dftxlvlwlrq ehkdylru/ pd| eh odujhu iru wkh
uvw v|perov zlwklq d gdwd iudph ^44`1
Qrwh wkdw lq slorw0v|pero0dvvlvwhg FH v|vwhpv wkh gdwd dqg slorw v|perov
duh ruwkrjrqdo e| phdqv ri wlph0glylvlrq pxowlsoh{lqj/ zkhuhdv iru vxshulp0
srvhg slorw vhtxhqfhv wkh furvv0fruuhodwlrq ehwzhhq wkh gdwd dqg slorw v|perov
w|slfdoo| lv qrq0}hur1 Krzhyhu/ wkh vxshulpsrvhg slorw vhtxhqfh lv nqrzq dqg
wkhuhiruh wkh glvwxuedqfh fdq eh pdgh vpdoo1 Wkh orvv gxh wr wkh slorw vh0
txhqfh lv ryhu0frpshqvdwhg e| wkh odfn ri lqwhusrodwlrq/ dv zh zloo vkrz lq
Vhfwlrq 81 Wkh wudqvplvvlrq v|vwhp pd| eh lqwhusuhwhg dv d wzr0xvhu v|v0
whp/ zkhuh rqh xvhu wudqvplwv lqirupdwlrq xqnqrzq wr wkh uhfhlyhu/ dqg wkh
rwkhu xvhu wudqvplwv nqrzq lqirupdwlrq1 Fruuhvsrqglqjo|/ wkh wudqvplvvlrq
whfkqltxh lv uhodwhg wr pxowl0xvhu v|vwhpv/ dqg wkh v|qfkurql}dwlrq2ghwhfwlrq
sureohp lv uhodwhg wr pxowl0xvhu ghwhfwlrq1
Jlyhq rxu dvvxpswlrqv/ wkh pdwfkhg owhu rxwsxw vdpsohv/ |n 5 F/ fdq eh
zulwwhq dv
|n @ +xn . sn , in . qn > +4,
zkhuh in 5 F lv d pxowlsolfdwlyh +wlph0vhohfwlyh, idglqj surfhvv zlwk H^min m5 ` @

4/ dqg qn 5 F duh }hur0phdq zklwh Jdxvvldq qrlvh vdpsohv zlwk rqh0vlghg
srzhu vshfwudo ghqvlw| Q3 1 Wkh fkdqqho xqghu frqvlghudwlrq kdv d Ud|ohljk
idglqj dpsolwxgh dqg d fodvvlfdo Grssohu vshfwuxp zlwk dxwrfruuhodwlrq fr0
h!flhqwv uo @ M3 +5iGpd{ Wv o,/ zkhuh iGpd{ ghqrwhv wkh pd{lpxp Grssohu
iuhtxhqf|/ dqg Wv wkh v|pero gxudwlrq1 Ryhuvdpsolqj lv qrw wuhdwhg zlwklq
wklv sdshu/ exw frxog eh xvhg wr ixuwkhu lpsuryh wkh shuirupdqfh lq idvw idg0
lqj1
Lq wkh wudqvplwwhu/ wkh pdmru glhuhqfh ehwzhhq wkh vsuhdg0vshfwuxp sl0
orw whfkqltxh sursrvhg e| Ihkhu hw do1 dqg wkh vfkhph ghvfulehg deryh lv
wkdw lq wkh ruljlqdo sursrvdo wkh olqhdu dgglwlrq zdv grqh diwhu G2D frqyhu0
vlrq/ zkhuhdv zh gr gljlwdo edvhedqg surfhvvlqj1 Wkhuhiruh/ zh fdq dvvxuh
v|pero0v|qfkurqrxv wudqvplvvlrq/ zklfk lv dq lpsruwdqw surshuw| iru rxu qryho
uhfhlyhu1 Dv d vlgh hhfw/ wkh srzhu ghqvlw| vshfwuxp lv xqdhfwhg e| wkh vx0
shulpsrvhg slorw vhtxhqfh zkhq xvlqj d SQ slorw vhtxhqfh1
Dq lpsruwdqw dvshfw lv wkh ghvljq ri wkh slorw vhtxhqfh1 Li sn lv fkrvhq wr
eh frqvwdqw iru doo n/ wkh slorw vhtxhqfh ehfrphv d slorw wrqh1 Jlyhq d SQ gdwd
vhtxhqfh/ wkh shuirupdqfh zrxog rq wkh dyhudjh eh wkh vdph frpsduhg wr d
SQ slorw vhtxhqfh1 Krzhyhu/ vlpxowdqhrxv wlph dqg iuhtxhqf| v|qfkurql}dwlrq
zrxog qr orqjhu eh srvvleoh zlwk d frqvwdqw slorw vhtxhqfh1 Ixuwkhupruh/
jlyhq d SQ slorw vhtxhqfh/ zh pd| h{whqg rxu uhfhlyhu wr frsh zlwk iuhtxhqf|0
vhohfwlyh idglqj dv zhoo1 Ilqdoo|/ lw lv lqwhuhvwlqj wr qrwh wkdw wkh frqfhsw ri
vxshulpsrvhg slorw vhtxhqfhv lv dssolfdeoh wr prvw h{lvwlqj dqdorj dqg gljlwdo
wudqvplvvlrq vfkhphv dv dq dgg0rq ihdwxuh
6 Uhfhlyhu Vwuxfwxuh
Wkh frqyhqwlrqdo uhfhlyhu iru wkh vxshulpsrvhg slorw vhtxhqfh whfkqltxh vsolwv
wkh uhfhlyhg vljqdo/ dqg ihhgv lw lqwr d fduulhu uhfryhu| xqlw dqg d ghwhfwlrq
xqlw ^:`^;`/ uhvshfwlyho|1 Wkh rxwsxw vljqdo ri wkh fduulhu uhfryhu| xqlw lv ihg
lqwr wkh ghwhfwlrq xqlw1 Lq rughu wr sduwo| frpshqvdwh wkh lqwhuihuhqfh gxh wr
wkh slorw vhtxhqfh/ d vlpsoh fdqfhoodwlrq vfkhph kdv ehhq sursrvhg1
Wkh frqyhqwlrqdo uhfhlyhu iru wkh slorw0v|pero0dvvlvwhg FH whfkqltxh rs0
hudwhv vlploduo|= wkh uhfhlyhg vljqdo lv ghpxowlsoh{hg/ dqg ihg lqwr d fkdqqho
hvwlpdwlrq xqlw +l1h1/ dq lqwhusrodwru, dqg d ghwhfwlrq xqlw ^6`^7`^8`^9`/ uhvshf0
wlyho|1 Lqwhuihuhqfh fdqfhoodwlrq lv qrw qhfhvvdu| e| ghvljq1
Dv rssrvhg wr wkhvh uhfhlyhuv/ zh zloo qrz lqwurgxfh d uhfxuvlyh uhfhlyhu
edvhg rq wkh Ylwhuel dojrulwkp/ zklfk gluhfwo| rxwsxwv hvwlpdwhv ri wkh gdwd
v|perov1 Wklv uhfhlyhu lv sulpdulo| ghvljqhg iru wkh vxshulpsrvhg slorw vh0
txhqfh whfkqltxh/ exw lv dovr dssolfdeoh iru wkh frqyhqwlrqdo slorw0v|pero0
dvvlvwhg FH whfkqltxh1 Glhuhqw fkdqqho hvwlpdwhv duh frpsxwhg iru glhu0
hqw k|srwkhvlv e| phdqv ri wkh sulqflsohv ri shu0vxuylyru surfhvvlqj ^48`^49`1
Lqwhuihuhqfh fdqfhoodwlrq lv grqh lqkhuhqwo|1 Wkh sulqflsoh ri wkh uhfhlyhu lv
wkh iroorzlqj= Wkh uhfhlyhu vwduwv zlwk d k|srwkhvlv ri wkh wudqvplwwhg gdwd
vhtxhqfh1 Dvvxplqj wkh k|srwkhvlv lv fruuhfw/ wkh uhfhlyhg v|perov duh wkhq
xvhg wrjhwkhu zlwk wkh k|srwkhvlv wr fdofxodwh fkdqqho hvwlpdwhv iru hdfk ri
wkh v|perov lq wkh eorfn1 Wkhq wkh glvwdqfh +phwulf, ehwzhhq wkh uhfhlyhg
vhtxhqfh dqg wkh k|srwkhvlv dhfwhg e| wkh fruuhvsrqglqj fkdqqho hvwlpdwhv lv
fdofxodwhg1 Wkh vdph surfhgxuh lv wkhq xvhg iru wkh qh{w k|srwkhvlv1 Ilqdoo|/
wkh uhfhlyhu pdnhv d ghflvlrq lq idyru ri wkh vhtxhqfh zlwk wkh orzhvw glvwdqfh1
Iru wkh fdofxodwlrqv lw lv frqyhqlhqw wr xvh pdwul{ qrwdwlrq1 Zh ghqrwh wkh
uhfhlyhg vdpsoh yhfwru e| | @ +|4 > |5 > ===> |Q ,W @ Fi . q/ zkhuh i dqg q duh
yhfwruv ri wkh idglqj surfhvv dqg wkh qrlvh/ uhvshfwlyho|/ dqg F lv d gldjrqdo
pdwul{ zlwk wkh wudqvplwwhg v|perov rq lwv gldjrqdo= fn>n @ +xn . sn ,1
Wkh fkdqqho hvwlpdwhv duh fdofxodwhg xvlqj wkh O odwhvw uhfhlyhg v|perov1
Dffruglqj wr +4,/ d whqwdwlyh fkdqqho hvwlpdwh/ iano / fdq eh rewdlqhg e| uh0
prgxodwlrq=
|n
ian @ > +5,
x
n . sn
zkhuh o lv dq duelwudu| lqwhjhu qxpehu1 Qrwh wkdw wkh uvw sduw ri wkh ghqrplqd0
wru lv edvhg rq wkh k|srwkhvlv/ zkhuhdv wkh vhfrqg sduw lv nqrzq= wkh slorw v|p0
erov khosv wr uhvroyh wkh skdvh dpeljxlw|1 Jlyhq k|srwkhvhv x nO.4 /
n > = = = > x
wkh fkdqqho hvwlpdwh in fdq eh frpsxwhg e| olqhdu owhulqj ^48`=

4
O[
in @ 4 |>
dn>o iano @ dn F +6,
o@3
zkhuh O lv wkh owhu rughu dqg dn lv d yhfwru ri wkh fkdqqho hvwlpdwlrq frh!0
flhqwv iru v|pero qxpehu n1
Li wkh fkdqqho lv zlgh0vhqvh vwdwlrqdu|/ wkh rswlpdo owhu frh!flhqwv/ do /
3 o O 4/ duh wkh vroxwlrqv ri wkh Zlhqhu0Krsi htxdwlrqv=
+d3 > = = = > dO4 ,W @ dWrsw @ 4 > +7,
zkhuh o @ Hîn iano ` @ uo duh wkh hohphqwv ri wkh furvv0fruuhodwlrq yhfwru dv0
vxplqj d fruuhfw k|srwkhvlv1 Ixuwkhu/ o>m @ Hîano ianm ` @ 5q @ m
xno . sno m5
o>m . umo duh wkh hohphqwv ri wkh dxwrfruuhodwlrq pdwul{ / zkhuh o> m @
4> = = = > O/ uo @ uo lv wkh owk dxwrfruuhodwlrq frh!flhqw ri wkh fkdqqho/ dqg
5q @ Q3 @Hv lv wkh yduldqfh ri wkh qrlvh1 Wkh fruuhvsrqglqj yhfwru ri fkdqqho
hvwlpdwhv iru d fhuwdlq k|srwkhvlv fdq frqvhtxhqwo| eh zulwwhq dv
4 |=
i @ DF +8,
Iru h{dpsoh/ iru d eorfn ohqjwk ri 8 uhfhlyhg v|perov dqg O @ 6> wkh fkdqqho
hvwlpdwlrq pdwul{ kdv wkh irup
3 4
d4>3 3 3 3 3
E d5>4 d5>3 3 3 3 F
E F
D@E E d6>5 d6>4 d6>3 3 3 F>
F +9,
C 3 d7>5 d7>4 d7>3 3 D
3 3 d8>5 d8>4 d8>3
zkhuh dl>m duh wkh owhu frh!flhqwv1 Qrwh wkdw wkh rswlpdo owhu frh!flhqwv duh
ghshqghqw rq wkh k|srwkhvlv1 Iru orz ydoxhv wkh glhuhqfh ehfrphv vpdoo/
krzhyhu/ dqg d vwdwlf owhu jlyhv d jrrg dssur{lpdwlrq1
Wkh jrdo lv wr uhfxuvlyho| frpsxwh wkh pd{lpxp0olnholkrrg hvwlpdwh ri wkh
wudqvplwwhg vhtxhqfh1 Wkh uhfhlyhu wkhq whvwv doo srvvleoh k|srwkhvhv dqg gh0
flghv iru wkh vhtxhqfh uhvxowlqj lq wkh orzhvw phwulf/ l1h1 wkh kljkhvw suredelolw|1
Lw lv/ krzhyhu/ kdug wr frpsxwh wkh wuxh pd{lpxp0olnholkrrg hvwlpdwh uhfxu0
vlyho|/ dqg wkhuhiruh zh kdyh wr pdnh dssur{lpdwlrqv wr dfklhyh dq dgglwlyh
phwulf vxlwdeoh iru uhfxuvlyh uhfhlyhuv1 Zh ghulyh wkh phwulf lq Dsshqgl{ D1
Wkh rswlpdo phwulf lv jlyhq e|
K
. |K +FU
rsw @ oq+ghw+!,,
K 5 4
i F . q L, |> +:,
@H^+| |
zkhuh ! ghqrwhv wkh fryduldqfh pdwul{ ri wkh uhfhlyhg vdpsohv/ ! ,
K
, `1 Dvvxplqj wkh k|srwkhvlv lv fruuhfw wklv fryduldqfh pdwul{ fdq eh
+| |
FKDQQHO HVWLPDWLRQ XVLQJ VXSHULPSRVHG SLORWV :
Iljxuh 5= Wkh fkdqqho hvwlpdwhv iru wkh glhuhqw k|srwkhvhv ri wkh wzr uvw
v|perov1
fdofxodwhg dv
4 4
F
! @ H^+L FD ,||K +L FD
F ,K `
F
@ +L FD 4 ,+FU
K K 5 4 K
i F . q L,+L FDF , +;,
zkhuh Ui lv wkh fryduldqfh pdwul{ ri wkh fkdqqho frqwdlqlqj wkh dxwrfruuhodwlrq
frh!flhqwv1
Zlwk wkh dgglwlyh phwulf lqfuhphqwv lw lv srvvleoh wr xvh wkh Ylwhuel dojr0
ulwkp iru ghwhfwlrq1 Lq Dsshqgl{ D lw lv vkrzq wkdw d vxlwdeoh dssur{lpdwlrq
wkdw |lhogv dgglwlyh phwulf lv jlyhq e|
@ +| FD
F 4 |,K +| FD
F 4 |,= +<,
Wkh fruuhvsrqglqj dgglwlyh phwulf lqfuhphqw xvhg lq wkh Ylwhuel uhfhlyhu lv
wkhq jlyhq e|
fn > fn4 > = = = > fnO , @ m|n fn in m5 >
n + +43,
zkhuh in lv wkh fkdqqho hvwlpdwh edvhg rq wkh fxuuhqw k|srwkhvlv1 Lq Ilj1 5 wkh
fkdqqho hvwlpdwhv iru wkh uvw wzr v|perov lq d irxu0vwdwh wuhoolv lv suhvhqwhg1
Wkh fkdqqho hvwlpdwhv duh edvhg rq wkh k|srwkhvhv dqg rq wkh fruuhodwlrq
ehwzhhq d vshflf fkdqqho hvwlpdwh dqg wkh dfwxdo fkdqqho ydoxh lv kljk rqo|
zkhq wkh k|srwkhvlv lv fruuhfw1 Wkh fruuhodwlrq fdq eh fdofxodwhg dv
% O4 & O4
[ [ xno . sno
H + do iano ,in @ do Hîno in `= +44,
no . sno
x
o@3 o@3
; FKDQQHO HVWLPDWLRQ XVLQJ VXSHULPSRVHG SLORWV
Wkh iudfwlrq rq wkh uljkw kdqg vlgh htxdov rqh rqo| iru wkh fruuhfw k|srwkhvlv
wkhuhe| surgxflqj d fruuhfw fkdqqho hvwlpdwh1 Wkh uhfhlyhu fdq eh lqwhusuhwhg
dv dq hvwlpdwru0fruuhodwlqj ghwhfwru/ zkhuh sduwv ri wkh uhfhlyhg vljqdo duh
nqrzq dqg sduwv ri wkh vljqdo duh xqnqrzq1 Lq d uvw vwhs/ wkh xqnqrzq
fkdqqho uhvsrqvh +dqg wkhuhe| wkh uhfhlyhg vljqdo, lv hvwlpdwhg e| wkh uhfhlyhg
v|perov dqg wkh k|srwkhvlv1 Lq wkh qh{w vwhs wklv hvwlpdwh lv fruuhodwhg zlwk
wkh uhfhlyhg v|perov1 Li wkh fruuhodwlrq ehwzhhq wkh fkdqqho hvwlpdwh dqg wkh
dfwxdo fkdqqho ydoxh lv kljk/ wkhq wkh phwulf zloo eh orz dqg wkh k|srwkhvlv
wkdw fruuhvsrqgv wr wkh wudqvplwwhg gdwd zloo eh fkrvhq1
Wkh k|srwkhvhv xn4 > = = = > xnO.4 frqvwlwxwh wkh P O4 vwdwhv ri d wuhoolv/
dqg xn ehorqjv wr wkh dfwxdo eudqfk1 Wkh wuhoolv kdv P eudqfkhv2vwdwh1 Iru wkh
dgglwlyh phwulf wkh idploldu dgg0frpsduh0vhohfw rshudwlrq fdq eh dssolhg dqg
wkh prvw suredeoh vhtxhqfh fdq eh irxqg e| edfn0wudflqj ru uhodwhg rshudwlrqv1
D uxoh ri wkxpe iru wkh ghflvlrq ghod| lv G O = = = 7O1 Gxh wr wkh vhtxhqfh
hvwlpdwlrq/ wkh uhfhlyhu fdq frsh zlwk d orz0srzhu slorw vhtxhqfh1
Iru odujh P dqg O/ wkh frpsoh{lw| fdq eh vljqlfdqwo| uhgxfhg e| dsso|lqj
wkh sulqflsohv ri uhgxfhg0vwdwh vhtxhqfh hvwlpdwlrq ^4:` ^4;`=
N
[ 4 O
[4
in @ do |no @+
xno . sno , . do |no @+a
xno . sno ,> +45,
o@3 o@N
zkhuh wkh wudgh0r ehwzhhq frpsoh{lw| dqg shuirupdqfh fdq eh dgmxvwhg e|

wkh sdudphwhu N/ 3 N O1 x n4 > = = = > x
nN .4 ehorqj wr d wuhoolv zlwk
P N 4 vwdwhv dqg P eudqfkhv2vwdwh/ zkhuhdv x anN > = = = > x
anO.4 ehorqj wr wkh
sdwk klvwru| +vxuylyru vhtxhqfh, ri wkh fruuhvsrqglqj vwdwh1 Wkhuhiruh/ N @ O
fruuhvsrqgv wr pd{lpxp0olnholkrrg vhtxhqfh hvwlpdwlrq dqg N @ 3 lv uhodwhg
wr sdudooho ghflvlrq ihhgedfn hvwlpdwlrq1 Iru P A 5/ ixuwkhu vlpsolfdwlrqv
fdq eh dfklhyhg e| wkh sulqflsohv ri vhw0sduwlwlrqlqj ^4:`1
Wkh qryho uhfhlyhu lv vxlwdeoh wr surfhvv slorw0v|pero0dvvlvwhg FH wrr1 Lq
frqwudvw wr wkh frqyhqwlrqdo slorw0v|pero0dvvlvwhg fkdqqho hvwlpdwru/ doo pdwfkhg
owhu rxwsxw vdpsohv/ l1h1 dovr wkh gdwd v|perov/ duh xvhg iru FH1 Dowkrxjk qrw
wuhdwhg ixuwkhu/ wkh sursrvhg uhfhlyhu pd| dovr eh h{whqghg wr dffhsw d sulrul
lqirupdwlrq dqg wr gholyhu vriw0rxwsxwv +zklfk lv qhfhvvdu| iru lwhudwlyh sur0
fhvvlqj,1 Ixuwkhupruh/ wkh uhfhlyhu pd| eh jhqhudol}hg iru iuhtxhqf|0vhohfwlyh
fkdqqhov1
7 Dqdo|wlfdo Shuirupdqfh Hydoxdwlrq

Lq wklv vhfwlrq zh pdnh dq dqdo|wlfdo dqdo|vlv ri wkh EHU xvlqj vxshulpsrvhg
slorwv1 Iluvw zh ghulyh wkh sdluzlvh huuru suredelolw| wkdw dq huurqhrxv k|srwk0
hvlv lv fkrvhq lqvwhdg ri wkh fruuhfw rqh1 Zh wkhq xvh wklv huuru suredelolw| wr
qg wkh prvw suredeoh huuru hyhqwv1 Xvlqj wkhvh zh qdoo| jhw dq dssur{lpd0
wlrq ri wkh ryhudoo EHU e| d Pdunry fkdlq dssurdfk1
FKDQQHO HVWLPDWLRQ XVLQJ VXSHULPSRVHG SLORWV <
Wr pdnh d ghflvlrq wkh uhfhlyhu fdofxodwhv wkh fxpxodwlyh phwulfv +<, iru
hdfk ri wkh k|srwkhvhv1 Wkh uhfhlyhu pdnhv d fruuhfw ghflvlrq li wkh phwulf
iru wkh fruuhfw sdwk lv vpdoohu wkdq doo ri wkh rwkhu sdwk phwulfv1 Ghqh wkh
a @
dgglwlyh ghflvlrq yduldeoh dv 1 Li a A 3 iru doo k|srwkhvhv wkhq wkh
fruuhfw sdwk zloo eh fkrvhq1 Wkh ghflvlrq yduldeoh fdq eh h{suhvvhg dv
a @ |K ^+L FD
F 4 ,K +L FD
F 4 , +L FDF4 ,K +L FDF4 ,`|=
+46,
Uhzulwh wkh uhfhlyhg vdpsohv dv | @ Gj zlwk
5 6
x4 . s4 3 3 4 3 3
9 3 x 5 . s5 3 3 4 3 :
9 :
G@9 11 1 11 11 11 : +47,
7 1 1 1 1 1 1 8
3 3 xQ . sQ 3 3 4
dqg

i
j@ = +48,
q
Wkh ghflvlrq yduldeoh fdq wkhq eh uhzulwwhq dv
a @ jK GK ^+L FD
F 4 ,K +L FD
F 4 , +L FDF4 ,K +L FDF4 ,`Gj
@ jK Nj=
+49,
Qrwh wkdw wkh pdwul{ N lv v|pphwulf dqg wkdw j lv d yhfwru ri fruuhodwhg

Jdxvvldq vdpsohv1 Wkh fryduldqfh pdwul{/ Uj > ri wkh }hur0phdq Jdxvvldq
yduldeohv j fdq eh fdofxodwhg dv

Ui 3
Uj @ > +4:,
3 5q L
zkhuh Ui lv wkh fryduldqfh pdwul{ ri wkh idglqj surfhvv1
Wkh h{suhvvlrqv iru wkh rswlpdo ghflvlrq yduldeoh duh vlplodu wr wkh dgglwlyh
rqhv dqg wkh vdph fdofxodwlrqv dsso|1 Wkh rswlpdo ghflvlrq yduldeoh lv
a rsw
@ jK GK ^+FU K K 5 4 K K 5 4
i F . q L, +FUi F . q L, `Gj
@ jK N
rsw j> +4;,
lqvwhdg
exw wr pdnh fruuhfw ghflvlrqv wklv kdv wr h{fhhg oq+ghw+!,,oq+ghw+!,,
ri }hur1
714 Sdluzlvh Huuru Suredelolw|

Pdq| dxwkruv kdyh vwxglhg wkh sdluzlvh huuru suredelolw| uhvxowlqj iurp txdg0
udwlf irupv olnh +49,/ h1j1 lq ^4<`^53`^54`1 Lq wkh vhtxho zh iroorz ^54` wr ghulyh wkh
glvwulexwlrq ri wkh ghflvlrq yduldeoh dqg fdofxodwh wkh sdluzlvh huuru suredelolw|1
Ghqh qhz Jdxvvldq yduldeohv dv j3 @ Ej> zkhuh E lv d udqn u pdwul{ ri vl}h
5Q u ixooolqj Uj @ EEK 1 Lq ^54` lw lv vkrzq wkdw wkh ghflvlrq yduldeoh
fdq eh uhjdughg dv d vxp ri u lqghshqghqw zhljkwhg "5 0yduldeohv/ zkhuh wkh
zhljkwv duh wkh hljhqydoxhv ri EK NE>
Qv
[
a @ jK Nj
@ q "55>q = +4<,
q@4
Wkh yhfwru ri Jdxvvldq vdpsohv/ j> lv frpsoh{ dqg wkh "5 0yduldeohv wkhuhe|
kdyh wzr ghjuhhv ri iuhhgrp1 Wkh phdq ri wkh ghflvlrq yduldeoh jlyhv d jrrg
lqglfdwlrq ri wkh huuru suredelolw|1 Lw fdq eh fdofxodwhg dv wkh wudfh ^54`
a @ wu+NU
H^` j ,= +53,
a zh vruw wkh zhljkwv vr

Wr fdofxodwh wkh suredelolw| ghqvlw| ixqfwlrq ri
wkdw wkh uvw p hljhqydoxhv duh srvlwlyh dqg wkh odvw Qv p hljhqydoxhv duh
qhjdwlyh1 Wkh ghflvlrq yduldeoh fdq wkhq eh h{suhvvhg dv
p
[ Qv
[
a@
q "55>q q "55>q= +54,
q@4 q@p.4
Dvvxplqj wkdw doo wkh hljhqydoxhv duh glvwlqfw/ zklfk wkh| iru doo sudfwlfdo fdvhv
a fdq eh fdofxodwhg dv ^54`
duh/ wkh suredelolw| ghqvlw| ixqfwlrq ri
; p
A S +q ,Qv 5 h|@+5q ,
A
A %
Q
& |A3
A Tv
? q@4 5
A +q n ,
n@4>n9@q
ia +|, @ Qv +55,
A S
A
A
+%q ,Qv 5 h|@+5q&, | 3=
A
A Qv
= q@p.4 5 T +n q ,
n@4>n9@q
Wkh sdluzlvh huuru suredelolw| iru wkh dgglwlyh ghflvlrq yduldeoh/ l1h1 wkh sure0
delolw| wkdw wkh ghflvlrq yduldeoh +49, ehfrphv qhjdwlyh vr wkdw zh zloo pdnh d
ghflvlrq lq idyru ri wkh huurqhrxv k|srwkhvlv lqvwhdg ri wkh fruuhfw rqh/ lv wkhq
] 3
+q ,Qv 4
Qv
[
Sdgg @ ia +|,g| @ Q = +56,
4 q@p.4
Tv
+n q ,
n@4>n9@q
Wkh fruuhvsrqglqj huuru suredelolw| iru wkh rswlpdo ghflvlrq yduldeoh ghshqgv
rq wkh vljq ri wkh ghwhfwlrq olplw { @ oq+ghw+!,, oq+ghw+!,,>

;
Qv p
A
A
A
S +q ,Qv 4 . S +q ,Qv 4 +4 h 5{q , { A 3
A
A Q
T v Q
T v
? q@p.4 +n q , q@4 +q n ,
n@4>n9@q n@4>n9@q
Srsw +{, @ Q
A
A
A
S v +q ,Qv 4 +4 h 5{q , { 3=
A Q
= q@p.4 Tv +n q ,
A
n@4>n9@q
+57,
Qrwh wkdw wkh hljhqydoxhv q duh ghshqghqw rq wkh wudqvplwwhg gdwd zklfk
phdqv wkdw wkh huuru suredelolw| ydulhv ghshqglqj rq wkh wudqvplwwhg gdwd
vhtxhqfh1
715 Prvw Suredeoh Huuruv

Wkh sdluzlvh huuru suredelolwlhv duh/ ri frxuvh/ ghshqghqw rq wkh vshflf sd0
udphwhuv xvhg/ h1j1 wkh srzhu ri wkh slorw vhtxhqfh/ wkh vljqdo wr qrlvh udwlr
+VQU, dqg wkh eorfn ohqjwk1 Dv dq h{dpsoh zh zloo qrz vwxg| wkh wkhruhwl0
fdo sdluzlvh huuru suredelolwlhv iru d vpdoo eorfn1 Wkh eorfn lv wrr vpdoo iru d
sudfwlfdo dssolfdwlrq/ exw lw lv lqvwuxfwlyh wr dqdo|}h wkh shuirupdqfh zkhq wkh
wrwdo qxpehu ri k|srwkhvhv/ l1h1 wkh qxpehu ri glhuhqw sdwkv lq wkh wuhoolv/ lv
vpdoo hqrxjk wr pdnh wkh h{dpsoh pdqdjhdeoh1 Dvvxph wkdw ; ESVN v|p0
erov duhsvhqw +P @ 5,1 Wkh gdwd dqg wkh slorw vhtxhqfhv duh udqgrpo| fkrvhq
s
dv x @ 4 +4> 4> 4> 4> 4> 4> 4,W dqg s @ +4> 4> 4> 4> 4> 4> 4> 4,W >
uhvshfwlyho|1 Iru wkh h{dpsoh zh kdyh 8( slorw srzhu/ owhu ohqjwk 6 dqg d
Grssohu udwh ri 4( + @ 3=38> O @ 6> iGpd{ Wv @ 3=34,= Lq Ilj1 6 zh sorw wkh
sdluzlvh huuru suredelolwlhv iru doo wkh 588 huurqhrxv k|srwkhvhv zkhq xvlqj wkh
rswlpdo ghflvlrq yduldeoh dqg wkh rswlpdo owhu frh!flhqwv1 Wkh k|srwkhvhv duh
rughuhg e| wkhlu wkhruhwlfdo sdwk suredelolw| dw He @Q3 @ 48 gE= Lq Ilj1 7 zh
sorw wkh vdph suredelolwlhv iru wkh dgglwlyh phwulf zkhq xvlqj wkh {hg owhu
frh!flhqwv/ l1h1 owhu frh!flhqwv zkhuh wkh ydu|lqj qrlvh yduldqfh gxh wr wkh
glhuhqw k|srwkhvhv lv qrw wdnhq lqwr dffrxqw1 Dowkrxjk qrw vkrzq khuh/ wkh
suhvhqwhg wkhruhwlfdo ydoxhv duh yhulhg e| vlpxodwlrqv1 Dv vhhq lq wkh jxuhv
wkh glhuhqfh lq huuru suredelolw| lv vpdoo ehwzhhq wkh rswlpdo phwulf xvlqj
wkh rswlpdo owhu frh!flhqwv dqg wkh dgglwlyh phwulf xvlqj wkh {hg owhu fr0
h!flhqwv1 Wkh vorsh ri wkh fxuyhv lv vrphzkdw orzhu iru wkh dgglwlyh phwulf/
exw iru orz dqg phglxp ydoxhv ri wkh VQU wkh shqdow| lv vpdoo1 Dw kljk VQU
ydoxhv krzhyhu/ wkh shqdow| iru wkh dgglwlyh phwulf ehfrphv pruh hylghqw iru
vrph vshflf k|srwkhvhv1
Wkhuh duh lq sulqflsoh wkuhh nlqgv ri frpprq huuruv iru vpdoo eorfnv= Huuruv
zkhuh wkh fkdqqho hvwlpdwru pdnhv d frqvwdqw skdvh huuru ri 4;3 ghjuhhv/ huuruv
zkhuh wkhuh lv d 4;3 ghjuhh skdvh vkliw lq wkh ghprgxodwhg gdwd vhtxhqfh/
dqg huuruv zlwk d vlqjoh skdvh vkliwhg v|pero vrphzkhuh lq wkh plggoh ri wkh
vhtxhqfh1 Wkh uvw jurxs lv dyrlghg e| lqfuhdvlqj wkh eorfn ohqjwk ru wkh slorw
0
10
PEP, optimal
-2
10
-4
10
-6
10
0
10
20 50
150 100
Eb/N0 30 250 200
hypothesis
Iljxuh 6= Sdluzlvh huuru suredelolwlhv +SHS, iru doo 588 huurqhrxv k|srwkhvhv lq
dq ;0v|pero eorfn xvlqj wkh rswlpdo phwulf dqg wkh rswlpdo owhu frh!flhqwv1
srzhu1 Wkh huuruv zlwk d 4;3 ghjuhh skdvh vkliw lq wkh plggoh ri wkh vhtxhqfh
duh dovr dyrlghg e| lqfuhdvlqj wkh eorfn ohqjwk1 Wkh phwulf iru wkh huurqhrxv
sdwkv lq wkh wuhoolv ehfrphv kljk diwhu d ihz v|perov vr iru orqj eorfnv wkhvh
k|srwkhvhv zloo qrw eh fkrvhq1 Krzhyhu/ dovr iru orqjhu eorfnv wkhuh duh vwloo
frpprq huuruv zlwk d 4;3 ghjuhh skdvh vkliw lq wkh ehjlqqlqj ru qhdu wkh hqg
ri wkh eorfn1 Wklv lv ehfdxvh wkhuh lv qr lqlwldo fkdqqho hvwlpdwh iru wkh uvw
v|perov dqg qr nqrzq hqg0vwdwh iru wkh odvw rqhv1 Li wkh uvw ru odvw v|perov
duh khdylo| glvwxuehg e| qrlvh wkh fkdqqho hvwlpdwhv fdq eh dgmxvwhg wr wkhvh
huurqhrxv v|perov dqg lw wdnhv vrph wlph ehiruh wkh fkdqqho hvwlpdwh ehfrphv
fruuhfw djdlq1 Ilqdoo|/ wkh prvw frpprq jurxs ri huuruv lv zkhuh zh kdyh d
vlqjoh skdvh vkliwhg v|pero lq wkh plggoh ri wkh vhtxhqfh1 Wkhvh huuruv fdq ri
frxuvh eh fruuhfwhg e| xvlqj dq| nlqg ri fkdqqho frglqj diwhu wkh ghwhfwru ru
e| xvlqj mrlqw ghwhfwlrq dqg ghfrglqj1 Wkh huuruv zlwk d vlqjoh skdvh vkliwhg
v|pero phdqv wkdw wkh fkdqqho hvwlpdwh lq wkh qhljkerukrrg ri wkh huuru zloo
eh edvhg rq dq lqfruuhfw k|srwkhvlv1 Exw vlqfh rqo| rqh ri wkh O v|perov xvhg
iru wkh hvwlpdwh lv zurqj/ wkh fkdqqho hvwlpdwlrq huuru dq|krz ehfrphv txlwh
vpdoo1
Iru odujh ru lqqlwh eorfn ohqjwkv wkh huuruv dsshdu dv vlqjoh v|pero huuruv ru
dv d jurxs ri huurqhrxv v|perov1 Wkh odwwhu lv hvshfldoo| wuxh iru orz Grssohu
iuhtxhqflhv zkhuh wkh idglqj glsv lq Ud|ohljk idglqj fkdqqhov kdyh gxudwlrq
0
10
PEP, additive
-2
10
-4
10
-6
10
0
10
20 50
150 100
Eb/N0 30 250 200
hypothesis
Iljxuh 7= Sdluzlvh huuru suredelolwlhv +SHS, iru doo 588 huurqhrxv k|srwkhvhv
lq dq ;0v|pero eorfn xvlqj wkh dgglwlyh phwulf dqg {hg owhu frh!flhqwv1
ryhu vhyhudo v|perov1 Lq d idglqj gls wkh suredelolw| lv kljk wkdw wkh zurqj
sdwkv lq wkh wuhoolv zloo eh fkrvhq/ zklfk phdqv wkdw wkh fkdqqho hvwlpdwhv duh
ohvv uholdeoh wkhuh1 Krzhyhu/ li dq huurqhrxv sdwk lv fkrvhq wkh phwulf ehfrphv
kljk diwhu d ihz v|perov dqg wkhuhiruh wkh fruuhfw sdwk zloo eh uhdfkhg djdlq
diwhu d fhuwdlq qxpehu ri v|perov1 Wkh uhfhlyhu fdq eh vhhq dv d Pdunry
fkdlq zkhuh wkh wudqvlwlrq suredelolw| dqg wkh suredelolw| ri pdnlqj d fruuhfw
ghflvlrq duh ghshqghqw rq wkh vwdwh/ l1h1 wkh sduwlfxodu v|perov +wkdw pljkw eh
huurqhrxv, xvhg wr irup wkh fkdqqho hvwlpdwh1
Wkhuh lv d kljk fruuhodwlrq ehwzhhq wkh phdq ri wkh ghflvlrq yduldeoh dqg wkh
fruuhvsrqglqj sdluzlvh huuru suredelolw|/ dqg wkhuhiruh vwxg|lqj wkh phdq lv dq
hdv| zd| wr qg wkh prvw suredeoh huuru sdwkv1 Wkh prvw suredeoh huurqhrxv
k|srwkhvhv dovr kdyh wkh orzhvw phdq ri wkh ghflvlrq yduldeoh1 Wkhuh lv dovr
d vwurqj ghshqghqfh ehwzhhq wkh udqn ri wkh pdwul{ UK
j N dqg wkh sdluzlvh
huuru suredelolw|1 Li wkh udqn lv orz/ wkh phdq ri wkh ghflvlrq yduldeoh lv orz dv
zhoo dqg wkhuh lv d kljk suredelolw| iru wkdw vshflf huuru hyhqw1
716 Dssur{lpdwlyh EHU Fdofxodwlrq

Wr fdofxodwh dq dssur{lpdwlyh EHU zh xvh d Pdunry fkdlq dssurdfk1 Iru rwkhu
ghwhfwlrq vfkhphv lw lv riwhq srvvleoh wr dfklhyh d wljkw xqlrq erxqg xvlqj rqo|
Iljxuh 8= Wudqvlwlrq suredelolwlhv lq wkh wuhoolv dqg wkh fruuhvsrqglqj vwdwh

suredelolwlhv1
wkh vlqjoh huuru hyhqwv/ vhh ^55` iru EHU fdofxodwlrqv ri wuhoolv frghg prgxodwlrq1
Iru wkh vxshulpsrvhg slorw whfkqltxh/ krzhyhu/ wkhuh lv d ghshqghqfh ehwzhhq
wkh huuruv dqg d xqlrq erxqg grhv qrw frqyhujh1 Lqvwhdg zh xvh wkh sdluzlvh
huuru suredelolwlhv wr fdofxodwh wkh wudqvlwlrq suredelolwlhv lq wkh wuhoolv/ l1h1 lq
wkh Pdunry fkdlq1 Dvvxph wkdw wkh huuru hyhqw kdv d olplwhg gxudwlrq1 Iru d
jlyhq wudqvplwwhg vhtxhqfh dqg d fhuwdlq vwdwh lq wkh wuhoolv zh uvw lghqwli| wkh
wzr prvw suredeoh k|srwkhvhv zlwk dqg zlwkrxw dq huuru dw srvlwlrq n= Zh wkhq
xvh rqo| wkhvh wzr vhtxhqfhv wr fdofxodwh wkh suredelolw| wkdw wkh ghflvlrq lv
pdgh lq idyru ri wkh fruuhfw v|pero dqg wkhuhe| wkh wudqvlwlrq suredelolwlhv wr
wkh qh{w vwdwhv lq wkh wuhoolv1 Dvvxplqj wkdw wkh gdwd vhtxhqfh lv 4> = = = > 4>
zh fdq xvh wkhvh wudqvlwlrq suredelolwlhv wr fdofxodwh wkh suredelolw| ri ehlqj lq
d fhuwdlq vwdwh lq vwhdg| vwdwh/ l1h1 lq wkh plggoh ri d orqj eorfn1 Wkhvh vwdwh
suredelolwlhv duh wkhq pxowlsolhg e| wkh suredelolw| ri pdnlqj dq huurqhrxv
ghflvlrq iru hdfk ri wkh vwdwhv wr jhw wkh dssur{lpdwlyh EHU1 Iru h{dpsoh/
dvvxph wkdw O @ 6/ @ 3=38> iGpd{ Wv @ 3==34 dqg He @Q3 @ 48 gE1 Ixuwkhu/
dvvxph wkdw wkh wudqvplwwhg slorw vhtxhqfh lv d vhtxhqfh ri rqhv pdsshg wr
s W
s @ +4> s 4> 4> 4> 4> 4> 4, dqg wkh gdwd vhtxhqfh lv wkh }hur vhtxhqfh pdsshg
wr x @ 4 +4> 4> 4> 4> 4> 4> 4,W = Iru vwdwh 33 wkh prvw suredeoh
k|srwkhvhv zlwk dqg zlwkrxw dq huuru dw srvlwlrq n @ 8 lv 3333433 dqg 3333333/
uhvshfwlyho|1 Khuh zh kdyh wzr v|perov wr jhw lqwr vwhdg| vwdwh/ wzr v|perov wr
uhdfk wkh dfwxdo vwdwh/ wkh huurqhrxv ru fruuhfw v|pero dqg wkhq wzr v|perov
wr uhdfk wkh fruuhfw vwdwh djdlq1 Iru vwdwh 34 wkh fruuhvsrqglqj k|srwkhvhv duh
3343433 dqg 3343333/ uhvshfwlyho|1 Wkh sdluzlvh huuru suredelolw| ehwzhhq wkh
vhtxhqfhv 3343433 dqg 3343333 zkhq wkh gdwd vhtxhqfh lv 3333333 lv 3135: iru
vwdwh 34 dqg wkhuhiruh wkh suredelolw| ri pdnlqj dq huurqhrxv ghflvlrq lv 3135:
lq wklv vwdwh/ wkh wudqvlwlrq suredelolw| wr vwdwh 43 lv 3135: dqg wkh wudqvlwlrq
suredelolw| wr vwdwh 33 lv 31<:61 Lq Ilj1 8 wkh sdwkv ri wkh vwhdg| vwdwh wuhoolv
lv suhvhqwhg zlwk lwv fruuhvsrqglqj wudqvlwlrq dqg vwdwh suredelolwlhv1 Wkh
0
10
fDmaxTs=10-1
fDmaxTs=10-3
calculations
-1 perfect CE
10
BER
-2
10
-3
10
-4
10
0 5 10 15 20 25 30
Eb/N0, dB
Iljxuh 9= EHU iru glhuhqw vljqdo wr qrlvh udwlrv1 Wkh grwwhg olqhv duh fdofx0
odwlrqv dvvxplqj wkh }hur gdwd vhtxhqfh dqg dq p0vhtxhqfh dv slorw vhtxhqfh1
vwdwh suredelolwlhv duh dfklhyhg e| wkh hljhqyhfwru ri wkh wudqvlwlrq pdwul{

fruuhvsrqglqj wr wkh hljhqydoxh rqh ^56`1 Dq dssur{lpdwlrq ri wkh ryhudoo
EHU lv wkhq wkh vxp ri wkh surgxfwv ehwzhhq wkh vwdwh suredelolwlhv dqg wkh
wudqvlwlrq suredelolwlhv fruuhvsrqglqj wr d v|pero huuru/ lq wklv fdvh Se r 3=345=
Wkh phwkrg ghvfulehg deryh zrunv zhoo iru kljk dqg phglxp VQU ydoxhv1
Lq Ilj1 9 zh sorw wkh dssur{lpdwlyh EHU xvlqj wkh phwkrg deryh wrjhwkhu
zlwk vlpxodwhg ydoxhv ri wkh EHU iru glhuhqw VQU ydoxhv1 Iru wkh fdofxod0
wlrqv dqg vlpxodwlrqv zh xvhg @ 3=38 dqg O @ 9= Lq Vhfwlrq 8 zh prwlydwh
wklv vshflf fkrlfh ixuwkhu1 Wkh fdofxodwlrqv zhuh edvhg rq wkh dvvxpswlrq
wkdw wkh wudqvplwwhg gdwd zdv wkh }hur vhtxhqfh dqg wkh slorw vhtxhqfh zdv
dq p0vhtxhqfh1 Iru orz VQU wkh fdofxodwhg ydoxhv duh vrphzkdw orzhu wkdq
wkh vlpxodwhg rqhv1 Vlqfh zh dvvxphg wkdw wkh ohqjwk ri wkh huuru hyhqw zdv
olplwhg/ wkh wzr prvw suredeoh k|srwkhvhv duh wkh k|srwkhvhv uhwxuqlqj edfn
wr wkh fruuhfw k|srwkhvlv gluhfwo| diwhu v|pero n> wkh v|pero ri lqwhuhvw1 Wklv
lv wuxh dovr iru orz VQU/ exw wkhq wkh suredelolw| ri rwkhu k|srwkhvhv/ zklfk
zh gr qrw wdnh lqwr dffrxqw/ duh kljk dv zhoo1 Dovr/ wkh EHU lv ghshqghqw rq
wkh uhodwlrq ehwzhhq wkh gdwd vhtxhqfh dqg wkh slorw vhtxhqfh1 Zh dvvxphg
dq p0vhtxhqfh iru wkh slorw vhtxhqfh dqg wkh }hur vhtxhqfh iru wkh gdwd vh0
txhqfh1 Wklv dvvxpswlrq uhvxowv lq d orz fruuhodwlrq ehwzhhq wkh slorw dqg
gdwd vhtxhqfhv dqg wkhuhiruh wkh fdofxodwhg ydoxhv duh vrphzkdw orzhu wkdq
wkh vlpxodwhg ydoxhv1 Wklv ghshqghqfh lv hvshfldoo| odujh iru odujh Grssohu
iuhtxhqflhv1
Iljxuh := Srvvleoh vwdwh wudqvlwlrqv lq wkh xqfrghg +gdvkhg olqh, dqg frghg
+vrolg olqh, fdvh1
717 Lq xhqfh ri Fkdqqho Frglqj

Li fkdqqho frglqj lv xvhg wkh qxpehu ri srvvleoh sdwkv lq wkh wuhoolv lv uhgxfhg1
Wklv phdqv d odujhu suredelolw| ri fkrrvlqj wkh fruuhfw sdwk dqg wkhuhe| d
odujhu srvvlelolw| ri xvlqj wkh fruuhfw k|srwkhvhv zkhq fdofxodwlqj wkh fkdqqhov
hvwlpdwhv1 Khqfh/ ehvlghv wkh huuru fruuhfwlrq fdsdelolw|/ wkh frgh dovr lpsolhv
wkdw ghwhfwlrq lv lpsuryhg1 Lq Ilj1 : zh vkrz wkh srvvleoh sdwkv lq dq hljkw0
vwdwh wuhoolv xvhg zlwkrxw ru zlwk d frqyroxwlrqdo frgh ri udwh U @ 4@5/ phpru|
5/ dqg jhqhudwru sro|qrpldo +8/:,1
Lq wkh frghg fdvh doo rxwsxw frpelqdwlrqv duh qrw doorzhg dqg wkhuhiruh
wkhuh lv/ h1j1/ qr ihdvleoh sdwk ehwzhhq vwdwh 333 dqg vwdwh 433 lq wkh vhfrqg
vwhs1 Lw lv srvvleoh wr xvh wkh vdph wuhoolv iru ghwhfwlrq dqg iru ghfrglqj1 Lqvwhdg
ri wkh frqyhqwlrqdo kdug ru vriw ghfrglqj phwulf zh xvh qrz wkh phwulf jlyhq
e| +<,1 Qrwh wkh srvvleoh sdwkv duh wkh rqhv lq wkh frqyhqwlrqdo irxu0vwdwh
wuhoolv iru wkh +8/:,0frgh1 Iru wkh irxu0vwdwh wuhoolv zh mxvw xvh wkh frgh rxwsxwv
dv k|srwkhvhv dqg qrw wkh vwdwhv gluhfwo|1
8 Shuirupdqfh Hydoxdwlrq e| Vlpxodwlrqv

Zh kdyh yhulhg rxu uhfhlyhu jlyhq wkh iroorzlqj vhw0xs= Zh xvhg ESVN prgx0
odwlrq iru wkh gdwd dqg slorw v|perov +qr vwdjjhulqj/ qr skdvh rvhw,/ l1h1 uhdo0
ydoxhg v|perov1 Iru wklv sduwlfxodu prgxodwlrq vfkhph/ wkh shuirupdqfh frxog
eh lpsuryhg e| wudqvplwwlqj wkh slorw vhtxhqfh dv wkh txdgudwxuh frpsrqhqw1
Krzhyhu/ khuh zh zdqw wr ghprqvwudwh wkh ihdvlelolw| ri xvlqj d frpprq fkdq0
qho lqvwhdg ri wzr lqghshqghqw fkdqqhov1 Wkh slorw vhtxhqfh zdv d orqj SQ
FKDQQHO HVWLPDWLRQ XVLQJ VXSHULPSRVHG SLORWV 4:
0
10
optimal filter
fixed filter
fDmaxTs=10-1
fDmaxTs=10-3
BER
-1
10
-2
10
0 0.2 0.4 0.6 0.8 1
pilot power
Iljxuh ;= Uhvxowlqj EHU zkhq xvlqj wkh rswlpdo dqg {hg owhu frh!flhqwv1
vhtxhqfh nqrzq wr wkh uhfhlyhu/ dv rssrvhg wr d vkruw vhtxhqfh dssolhg lq

^;`1 Gdwd dqg slorw v|perov zhuh jhqhudwhg e| lqghshqghqw svhxgr0udqgrp
jhqhudwruv1 Zh dyhudjhg rxu uhvxowv ryhu vhyhudo wkrxvdqgv ri srvvleoh slorw vh0
txhqfhv1 Wkh eorfn ohqjwk zdv fkrvhq wr eh 5333 v|perov dqg wkh shuirupdqfh
zdv vwxglhg iru d zlgh udqjh ri glhuhqw idglqj udwhv iGpd{ Wv 1
Lq wkh uhfhlyhu/ wkh sursrvhg Ylwhuel edvhg ghwhfwru zlwk d vx!flhqwo| orqj
ghflvlrq ghod| zdv dssolhg1 Wkh Grssohu vshfwuxp dqg wkh VQU zhuh dvvxphg
wr eh nqrzq lq wkh fkdqqho hvwlpdwlrq owhu ghvljq1 Wklv dvvxpswlrq grhv qrw
dsshdu wr eh fulwlfdo dv lqglfdwhg lq uhodwhg zrun1 Wkh EHU zdv fkrvhq dv
shuirupdqfh fulwhuld1 Irfxv zdv rq wkh wudfnlqj skdvh/ l1h1 zh glg qrw wdnh wkh
uvw sduw ri wkh eorfn lqwr dffrxqw1
Lq d uvw vhw ri Prqwh Fduor vlpxodwlrqv zh rswlpl}hg wkh qrupdol}hg srzhu/
/ ri wkh vxshulpsrvhg slorw v|perov1 Lq Ilj1 ; wkh uhvxowlqj EHU lv sorwwhg
iru glhuhqw slorw srzhu ydoxhv zkhq wkh rswlpdo owhu frh!flhqwv +zkhuh wkh
fkdqqho hvwlpdwlrq owhu frh!flhqwv duh dgmxvwhg wr wkh srzhu ri wkh vshflf
v|perov, ru wkh {hg owhu frh!flhqwv duh xvhg1 Wkh vljqdo wr qrlvh udwlr shu
lqirupdwlrq elw zdv He @Q3 @ 48 gE dqg wkh owhu rughu zdv fkrvhq wr eh O @ 9/
zklfk fruuhvsrqgv wr 5O4 @ 65 vwdwhv1 Iru wkh lqwhuhvwlqj srzhu udqjh/ l1h1
zkhuh wkh EHU dqg wkh slorw srzhu lv orz/ wkh shuirupdqfh lv vlplodu iru erwk
owhuv1 Wkh {hg owhu frh!flhqwv fdq wkhuhiruh eh xvhg/ uhvxowlqj rqo| lq d
vpdoo shuirupdqfh ghjudgdwlrq frpsduhg wr wkh rswlpdo owhu frh!flhqwv1 Lq
Ilj1 < d ghwdlohg sorw ri wkh orz slorw srzhuv lv vkrzq1 Lw dsshduv wkdw wkh EHU
lv qrw vhqvlwlyh zlwk uhvshfw wr ryhu d zlgh udqjh1 Wkh rswlpxp qrupdol}hg
4; FKDQQHO HVWLPDWLRQ XVLQJ VXSHULPSRVHG SLORWV
0
10
fDmaxTs=10-1
fDmaxTs=10-2
fDmaxTs=10-3
BER
-1
10
-2
10 -2 -1 0
10 10 10
pilot power
Iljxuh <= Rswlpl}dwlrq ri wkh srzhu/ > ri wkh vxshulpsrvhg slorw vhtxhqfh1
srzhu lv derxw @ 3=35 = = = 3=38 iru doo Grssohu iuhtxhqflhv ri lqwhuhvw= Wkh
srzhu ri slorw v|perov vkrxog rqo| eh derxw 5 ( wr 8 ( ri wkh srzhu ri wkh
gdwd v|perov1 Lq wkh iroorzlqj/ @ 3=38 lv dssolhg1
Lq d vhfrqg vhw ri vlpxodwlrqv zh lqyhvwljdwhg wkh lq xhqfh ri wkh owhu
ohqjwk O/ dqg wkh uhvxowv duh suhvhqwhg lq Ilj1 431 Djdlq/ He @Q3 @ 48 gE1 Lw
dsshduv iurp wkh fxuyhv wkdw d owhu ohqjwk ri derxw O @ 9 lv vx!flhqw iru doo
Grssohu iuhtxhqflhv ri lqwhuhvw1 Dowkrxjk qrw vkrzq khuh/ wkh phdq vtxduhg
huuru ri wkh owhu/ zklfk fdq eh fdofxodwhg dqdo|wlfdoo|/ vkrzv wkh vdph sulqflsoh
ehkdylru dv wkh EHU fxuyh lq wkh jxuh1
Jlyhq wkhvh rswlpl}dwlrqv/ zh vlpxodwhg wkh EHU yhuvxv He @Q3 iru wkh
vfkhph xqghu lqyhvwljdwlrq1 Wkh uhvxow lv vkrzq lq Ilj1 91 Dv d ehqfkpdun/
wkh EHU shuirupdqfh ri 50SVN rq d dw Ud|ohljk idglqj fkdqqho jlyhq shuihfw
fkdqqho hvwlpdwlrq lv sorwwhg dv zhoo1 Iru idglqj udwhv xs wr 435 wkh orvv lv ohvv
wkdq 5 gE/ dqg hyhq iru idvw idglqj +iGpd{ Wv @ 3=4, qr huuru rru lv ylvleoh lq
wkh lqwhuhvwlqj udqjh1 Wklv lv lq frqwudvw wr frqyhqwlrqdo slorw0v|pero0dvvlvwhg
FH1 Iru kljk Grssohu iuhtxhqflhv slorw0v|pero0dvvlvwhg FH vkrzv dq huuru rru
dqg wkh vxshulpsrvhg slorw vhtxhqfh whfkqltxh vljqlfdqwo| rxwshuirupv wklv
phwkrg1
Lq Ilj1 44 zh frpsduh wkh EHU iru glhuhqw Grssohu ydoxhv zkhq xvlqj
frqyhqwlrqdo slorw0v|pero0dvvlvwhg FH dqg vxshulpsrvhg slorw vhtxhqfhv1 Dv
d uhihuhqfh zh dovr jlyh wkh EHU iru shuihfw fkdqqho hvwlpdwlrq1 Lq erwk fdvhv
wkh slorw srzhu lv 8(1 Wklv phdqv iru wkh slorw0v|pero0dvvlvwhg vfkhph wkdw
hyhu| 53wk v|pero lv d slorw v|pero dqg wkdw wkh hqhuj| shu lqirupdwlrq elw lv
FKDQQHO HVWLPDWLRQ XVLQJ VXSHULPSRVHG SLORWV 4<
0.1
fDmaxTs=10-1
fDmaxTs=10-3
perfect CE
BER
0.01
2 3 4 5 6 7 8 9 10
filter length
Iljxuh 43= Lq xhqfh ri wkh fkdqqho hvwlpdwlrq owhu ohqjwk1
ghfuhdvhg e| 8(1 Dv h{shfwhg/ wklv phwkrg vkrzv odujh shuirupdqfh ghjudgd0

wlrq dw Grssohu ydoxhv deryh 31358 gxh wr wkh ylrodwlrq ri wkh Q|txlvw vdpsolqj
wkhruhp1 Wkhuh wkh uhfhlyhu lv qrw deoh wr wudfn doo fkdqqho fkdqjhv dqg wkh
EHU shuirupdqfh lv ghjudghg1 Wkh vxshulpsrvhg slorw vfkhph/ krzhyhu/ xvhv
wkh O odwhvw uhfhlyhg v|perov iru fkdqqho hvwlpdwlrq dqg wkhuhiruh gr qrw vxhu
iurp d wrr odujh glvwdqfh ehwzhhq wkh slorw v|perov1
9 Frqfoxvlrqv
Lq wklv sdshu/ zh h{soruhg wkh vxshulpsrvhg slorw vhtxhqfh whfkqltxh iru wkh
sxusrvh ri fkdqqho hvwlpdwlrq1 Gxh wr wkh uhgxqgdqf|/ wuxo| frkhuhqw ghprg0
xodwlrq lv dfklhyhg1 Wkh pdlq frqwulexwlrq zdv wkh ghulydwlrq dqg dqdo|vlv ri d
uhfhlyhu edvhg rq wkh Ylwhuel dojrulwkp1 Zh ghulyhg erwk wkh rswlpdo phwulf/
dqg d vxe0rswlpdo phwulf vxlwdeoh iru uhfxuvlyh ghwhfwruv1 Wkh shuirupdqfh ri
wkh sursrvhg uhfhlyhu zdv wkhq dqdo|}hg erwk dqdo|wlfdoo| dqg e| vlpxodwlrqv1
Iru wkh dqdo|wlfdo dqdo|vlv zh ghulyhg wkh sdluzlvh huuru suredelolwlhv dqg edvhg
rq wkhvh zh pdgh dssur{lpdwlrqv ri wkh ryhudoo EHU1
Frpsduhg wr slorw0v|pero0dvvlvwhg FH/ zklfk lv fxuuhqwo| vwdwh0ri0wkh0duw/
glvwlqfw dgydqwdjhv duh dv iroorzv=
Qr edqgzlgwk h{sdqvlrq
Ehwwhu srzhu h!flhqf| lq idvw idglqj hqylurqphqwv

PSA CE
SPP
perfect CE
-1
10
BER
-2 15 dB
10
25 dB
-3
10
-2 -1
10 10
Doppler rate
Iljxuh 44= EHU iru shuihfw fkdqqho hvwlpdwlrq/ vxshulpsrvhg slorw vhtxhqfhv
+VSS, ru slorw0v|pero0dvvlvwhg FH +SVD FH,1 Lq wkh wzr odwwhu fdvhv 8( ri wkh
srzhu duh vshqw rq slorwv1
Wkh whfkqltxh lv pruh xqlyhuvdoo| dssolfdeoh= Wkh vdph vhtxhqfh pd|

eh xvhg iru wlph/ skdvh/ iuhtxhqf|/ dqg iudph v|qfkurql}dwlrq dqg dv d
xqltxh zrug zlwkrxw dgglwlrqdo ryhukhdg1 Wkh frqfhsw ri vxshulpsrvhg
slorw vhtxhqfhv pd| eh xvhg dv dq dgg0rq lq h{lvwlqj v|vwhpv1
Slorw0v|pero0dvvlvwhg FH/ zlwk lvrodwhg slorw v|perov/ idlov zkhq wkh slorw
v|perov duh dhfwhg e| iuhtxhqf|0vhohfwlyh idglqj
Gdwd v|perov duh htxdoo| vhqvlwlyh djdlqvw wudqvplvvlrq huuruv zkhq df0
txlvlwlrq lv qrw frqvlghuhg1
Wkh sursrvhg Ylwhuel uhfhlyhu lv/ krzhyhu/ pruh frpsoh{ wkdq d vlqjoh lqwhu0
srodwlrq owhu gxh wr wkh shu0vxuylyru surfhvvlqj1 Wkhuhiruh/ vlpsolhg vwuxf0
wxuhv edvhg rq uhgxfhg0vwdwh vhtxhqfh hvwlpdwlrq zhuh sursrvhg dv zhoo1 Dv
rxwolqhg lq wkh dqdo|vlv/ wkh shuirupdqfh ghshqgv rq wkh eorfn ohqjwk dqg rq
wkh fkrlfh ri wkh SQ vhtxhqfh1
Ixwxuh zrun pd| eh ghyrwhg wr rswlpl}dwlrqv ri wkh slorw vhtxhqfh/ dq
hydoxdwlrq ri P 0du| prgxodwlrq vfkhphv/ pxowl0fduulhu prgxodwlrq vfkhphv
+h1j1/ RIGP zlwk 50G slorw duud|,/ dqg uhfhlyhu vwuxfwxuhv zlwk ryhuvdpsolqj1
D Ghulydwlrq ri wkh phwulf

Wkh suredelolw| ghqvlw| ixqfwlrq ri wkh uhfhlyhg v|pero yhfwru lv jlyhq e|
,K !4 +| |
4 4
i +| m F, @ h{s+ +| | ,,> +58,
+5,Q@5 ghw4@5 +!, 5
zkhuh | ghqrwhv wkh phdq ri wkh yhfwru | dqg ! ghqrwhv wkh fryduldqfh pdwul{
ri wkh uhfhlyhg vdpsohv/ ! @H^+| | ,+| |,K `= Wkh phdq |
lv qrw nqrzq d
wkh phdq lv jlyhq e| wkh fkdqqho hvwlpdwhv
sulrul/ exw jlyhq wkh k|srwkhvlv F
dv

@F F 4
|
@ H^| m F` i @ FD |= +59,
Wkh fryduldqfh pdwul{ fdq qrz eh fdofxodwhg dv
F
! @ H^+L FD 4 ,||K +L FD
F 4 ,K ` +5:,
F
@ +L FD 4 ,+FU
K K 5 4 K
i F . q L,+L FDF ,
zkhuh Ui lv wkh fryduldqfh pdwul{ ri wkh fkdqqho1 Vxevwlwxwlqj +59, dqg +5:,
lqwr +58, dqg wdnlqj wkh qdwxudo orjdulwkp |lhogv wkh orj0olnholkrrg ixqfwlrq
@ 4 oq++5,Q ghw+!,, 4 +| FD
oq i+| m F, F 4 +| FD
4 |,K ! F 4 |,
5 5
4 4 K K
@ oq++5, ghw+!,, | +FUi F . 5q L,4 |
Q +5;,
5 5
Pd{lpl}lqj wklv lv htxlydohqw wr plqlpl}lqj
rsw @ oq+ghw+!,, . +| FD
F 4 |,K !
4 +| FD
F 4 |, +5<,
. |K +FU
@ oq+ghw+!,, K K 5 4
i F . q L, |
zklfk lv wkh rswlpdo phwulf1

Wkh jrdo lv qrz wr ghulyh vxlwdeoh phwulf lqfuhphqwv vxfk wkdw wkh Ylwhuel
dojrulwkp fdq eh xvhg1 Li wkh srzhu ri wkh slorw vhtxhqfh lv vpdoo/ wkhq wkh
glhuhqfh ehwzhhq wkh ghwhuplqdqwv iru wkh glhuhqw k|srwkhvhv lv vpdoo dqg
wkhuhiruh wklv whup lv qhjohfwhg1 Ixuwkhu/ dvvxplqj wkdw wkh lqyhuvh ri wkh
fryduldqfh pdwul{ lv gldjrqdoo| grplqdqw/ wkhq lw fdq eh dssur{lpdwhg dv dq
lghqwlw| pdwul{ dqg zh jhw wkh ghvluhg dgglwlyh phwulf dv
@ +| FD
F 4 |,K +| FD
F 4 |,= +63,
Uhihuhqfhv
^4` M1 J1 Surdnlv1 Gljlwdo frppxqlfdwlrqv1 Suhqwlfh0Kdoo/ 6ug hglwlrq/ 4<<81
^5` K1 Ph|u/ P1 Prqhfodh|/ dqg V1 Ihfkwho1 Gljlwdo Frppxqlfdwlrq Uhfhlyhuv1

Zloh|/ Qhz \run/ XVD/ 4<<;1
^6` P1 O1 Prkhu dqg M1 K1 Orgjh1 WFPS D prgxodwlrq dqg frglqj vwudwhj|

iru Ulfldq0idglqj fkdqqhov1 LHHH M1 Vhohfw1 Duhdv Frppxq1/ :+<,=467:
4688/ Ghf1 4<;<1
^7` P1 Prkhu dqg M1 Orgjh1 D wlph glyhuvlw| prgxodwlrq vwudwhj| iru wkh
vdwhoolwh0preloh fkdqqho1 Lq 46wk Elhqqldo V|ps1 Frppxq1/ Nlqjvwrq/
Fdqdgd/ Mxqh 4<;91
^8` D1 Djkdprkdppdgl/ K1 Ph|u/ dqg J1 Dvfkhlg1 D qhz phwkrg iru skdvh

v|qfkurql}dwlrq dqg dxwrpdwlf jdlq frqwuro ri olqhduo|0prgxodwhg vljqdov
rq iuhtxhqf|0 dw idglqj fkdqqhov1 LHHH Wudqv1 Frppxq1/ 6<+4,=585</
Mdq1 4<<41
^9` M1 N1 Fdyhuv1 Dq dqdo|vlv ri slorw0v|pero dvvlvwhg prgxodwlrq iru Ud|ohljk0

idglqj fkdqqhov1 LHHH Wudqv1 Yhklf1 Whfkqro1/ 73+7,=9;99<6/ Qry1 4<<41
^:` G1 Pdnudnlv dqg N1 Ihkhu1 D qryho slorw lqvhuwlrq0h{wudfwlrq phwkrg edvhg

rq vsuhdg vshfwuxp whfkqltxhv1 Lq Pldpl Whfkqlfrq/ Pldpl/ XVD/ 4<;:1
^;` W1 Kroghq dqg N1 Ihkhu1 D vsuhdg vshfwuxp edvhg v|vwhp whfkqltxh iru
Eurdgf1/ sdjhv 4;84<7/ Vhsw1 4<<31
^<` N1 M1 vwup dqg E1 Zlwwhqpdun1 Dgdswlyh Frqwuro1 Dgglvrq0Zhvoh|/

Uhdglqj/ Pdvvdfkxvhwwv/ vhfrqg hglwlrq/ 4<<81
^43` D1 Vwhlqjdvv/ D1 Zlmqjddughq/ dqg Z1 Whlfk1 Iudph v|qfkurql}dwlrq xvlqj

vxshulpsrvhg vhtxhqfhv1 Lq Surf1 LHHH Lqwhuq1 V|ps1 Lqi1 Wkhru|/ sdjh
7;</ Xop/ Jhupdq|/ Mxqh 4<<:1
^44` I1 Wxiyhvvrq/ P1 Idxonqhu/ S1 Krhkhu/ dqg R1 Hgiruv1 RIGP wlph

dq iuhtxhqf| v|qfkurql}dwlrq e| vsuhdg vshfwuxp slorw whfkqltxh1 Lq
Hljkwk Frppxqlfdwlrq Wkhru| Plql Frqihuhqfh lq frqmxqfwlrq zlwk LHHH
LFF *<</ sdjhv 44844</ Ydqfrxyhu/ Fdqdgd/ Mxqh 4<<<1
^45` H1 Ehmmdql/ M10F1 Ehoruh/ dqg S1 Ohfodlu1 Frkhuhqw ghwhfwlrq iru wudqvplv0
vlrq ryhu vhyhuo| wlph dqg iuhtxhqf| glvshuvlyh pxowlsdwk idglqj fkdqqhov1
Lq Surf1 LHHH Lqwhuq1 V|ps1 Lqi1 Wkhru|/ sdjh 545/ Zklvwohu/ Fdqdgd/
Vhsw1 4<<81
^46` H1 Gdkopdq/ S1 Ehplqj/ M1 Nqxwvvrq/ I1 Ryhvm/ P1 Shuvvrq/ dqg

F1 Urrero1 ZFGPD 0 wkh udglr lqwhuidfh iru ixwxuh preloh pxowlph0
gld frppxqlfdwlrqv1 LHHH Wudqv1 Yhklf1 Whfkqro1/ 7:+7,=4438444;/ Qry1
4<<;1
vhtxhqfh dssolhg wr pxowl0fduulhu v|vwhpv1 Lq Lqwhuqdwlrqdo Zrunvkrs rq
Pxowl0Fduulhu Vsuhdg0Vshfwuxp dqg Uhodwhg Wrslfv/ sdjhv 5<8635/ Rehus0
idhqkrihq/ Jhupdq|/ Vhsw1 4<<<1
^48` M1 Orgjh dqg P1 Prkhu1 Pd{lpxp olnholkrrg vhtxhqfh hvwlpdwlrq ri
FSP vljqdov wudqvplwwhg ryhu Ud|ohljk dw0idglqj fkdqqhov1 LHHH Wudqv1
Frppxq1/ 6;+9,=:;::<7/ Mxqh 4<<31
^49` U1 Udkhol/ D1 Sro|grurv/ dqg F10N1 W}rx1 Shu0vxuylyru surfhvvlqj= D
jhqhudo dssurdfk wr POVH lq xqfhuwdlq hqylurqphqwv1 LHHH Wudqv1 Frp0
pxq1/ 76+52627,=687697/ Ihe2Pdu2Dsu 4<<81
^4:` P1 Y1 H|xerjox dqg V1 X1 Txuhvkl1 Uhgxfhg0vwdwh vhtxhqfh hvwlpd0
wlrq zlwk vhw sduwlwlrqlqj dqg ghflvlrq ihhgedfn1 LHHH Wudqv1 Frppxq1/
69+4,=4653/ Mdq1 4<;;1
^4;` D1 Gxho0Kdoohq dqg F1 Khhjdug1 Ghod|hg ghflvlrq0ihhgedfn vhtxhqfh hvwl0
pdwlrq1 LHHH Wudqv1 Frppxq1/ 6:+8,=75;769/ Pd| 4<;<1
^4<` P1 M1 Eduuhw1 Huuru suredelolw| iru rswlpdo dqg vxerswlpdo txdgudwlf uh0
fhlyhuv lq udslg Ud|ohljk idglqj fkdqqhov1 LHHH M1 Vhohfw1 Duhdv Frppxq1/
VDF08+5,=635637/ Ihe1 4<;:1
^53` F1 Vfkohjho1 Huuru suredelolw| fdofxodwlrq iru pxowlehdp Ud|ohljk fkdqqhov1
LHHH Wudqv1 Frppxq1
^54` D1 Pdwkdl dqg V1 E1 Suryrvw1 Txdgudwlf irupv lq udqgrp yduldeohv= Wkhru|
dqg dssolfdwlrqv1 Pdufho Ghnnhu/ Qhz \run/ X1V1D1/ 4<<51
^55` M1 N1 Fdyhuv dqg S1 Kr1 Dqdo|vlv ri wkh huuru shuirupdqfh ri wuhoolv0
frghg prgxodwlrq lq Ud|ohljk0idglqj fkdqqhov1 LHHH Wudqv1 Frppxq1/
73+4,=:7;6/ Mdq1 4<<51
^56` J1 Vwudqj1 Olqhdu Dojheud dqg Lwv Dssolfdwlrqv1 Dfdghplf Suhvv/ 6ug
hglwlrq/ 4<;;1
Paper G
Paper G
This part is a revised version of:
Fredrik Tufvesson and Ove Edfors, "A Comparaitve Analysis Between Different
Detection Schemes in Flat Rayleigh-fading Channels", Proceedings of IEEE Vehicu-
lar Technology Conference, Tokyo, Japan, pp. 1205-1209, May 2000.
D Frpsdudwlyh Dqdo|vlv Ehwzhhq Glhuhqw
Ghwhfwlrq Vfkhphv lq Iodw Ud|ohljk0idglqj
Fkdqqhov
Iuhguln Wxiyhvvrq dqg Ryh Hgiruv

Ghsduwphqw ri Dssolhg Hohfwurqlfv/ Oxqg Xqlyhuvlw|/
Er{ 44;/ VH0554 33 Oxqg/ Vzhghq
Skrqh= .79079 555 63 54/ Id{= .79079 45 << 7;/
H0pdlo= Iuhguln1WxiyhvvrqCwgh1owk1vh
Devwudfw
Zh frpsduh wkh shuirupdqfh ri glhuhqw ghwhfwlrq vfkhphv lq whupv

ri uhvxowlqj sdluzlvh huuru suredelolw| lq Ud|ohljk0idglqj fkdqqhov1 Zh
lqwurgxfh d xqlhg dqdo|vlv iudphzrun xvhg wr frpsduh vhqvlwlylw| wr
Grssohu dqg qrlvh1 Wkh dqdo|}hg phwkrgv lqfoxgh= slorw v|pero dvvlvwhg
prgxodwlrq/ glhuhqwldo ghwhfwlrq/ pxowlsoh0v|pero glhuhqwldo ghwhfwlrq
dqg ghwhfwlrq xvlqj vxshusrvhg slorw v|perov1 Zh dovr lqwurgxfh dq lp0
suryhg k|eulg ghwhfwlrq vfkhph/ frpelqlqj wkh vwuhqjwkv ri slorw v|pero
dvvlvwhg prgxodwlrq dqg vxshusrvhg slorw v|perov1
4 Lqwurgxfwlrq
Lw lv zhoo nqrzq wkdw frkhuhqw ghwhfwlrq uhvxowv lq jrrg shuirupdqfh1 Lw dovr
hqdeohv wkh xvh ri pxowlohyho vljqdoolqj exw uhtxluhv uholdeoh skdvh dqg2ru fkdq0
qho hvwlpdwlrq1 Glhuhqwldo ghwhfwlrq/ rq wkh rwkhu kdqg/ lv urexvw dqg uhtxluhv
qr fkdqqho hvwlpdwlrq/ exw wkhuh lv d shuirupdqfh orvv frpsduhg wr frkhuhqw
ghwhfwlrq dqg lw lv prvwo| xvhg iru skdvh vkliw nh|lqj +SVN,1 Pdq| qhz vfkhphv
kdyh ehhq suhvhqwhg wr hqkdqfh wkh shuirupdqfh ri erwk frkhuhqw dqg glhuhq0
wldo phwkrgv1 Lq wklv sdshu zh pdnh dq dqdo|wlfdo frpsdulvrq ehwzhhq vrph ri
wkhvh ghwhfwlrq vfkhphv1 Wkh dqdo|vlv lv edvhg rq d xqlhg iudphzrun zkhuh
zh fdofxodwh wkh sdluzlvh suredelolw| wkdw dq huurqhrxv vhtxhqfh lv ghwhfwhg
lq idyru ri wkh fruuhfw rqh1 Wkh dqdo|}hg phwkrgv lqfoxgh= frqyhqwlrqdo gli0
ihuhqwldo ghwhfwlrq +Gl,/ pxowlsoh0v|pero glhuhqwldo ghwhfwlrq +PVGG,/ slorw
v|pero dvvlvwhg prgxodwlrq +SVDP,/ ghwhfwlrq xvlqj vxshusrvhg slorw v|perov
+VSS,/ dqg d qryho frkhuhqw k|eulg ghwhfwlrq vfkhph +K|eu,1
Wkh PVGG zdv lqwurgxfhg iru DZJQ fkdqqhov lq ^4` dqg h{whqghg wr idg0
lqj fkdqqhov lq ^5` dqg ^6`1 Wkh lghd ehklqg wklv whfkqltxh lv wr vlpxowdqhrxvo|
4
5 D FRPSDUDWLYH DQDO\VLV
ghwhfw d jurxs ri skdvh vkliwv lq rughu wr kdyh pruh wkdq rqh v|pero dv d uhi0
huhqfh1 Lq wklv zd| wkh uhihuhqfh lv ohvv dhfwhg e| qrlvh dqg wkh shuirupdqfh
lv wkhuhe| hqkdqfhg1 Wkh dqdo|vlv lq ^6` xvhv h{suhvvlrqv iru wkh sdluzlvh hu0
uru suredelolwlhv dqg d xqlrq erxqg wr fdofxodwh dq dssur{lpdwlyh elw0huuru
udwh1 Lq ^7` d ghwhfwru edvhg rq wkh Ylwhuel dojrulwkp zdv suhvhqwhg/ zklfk zh
xvh lq rxu dqdo|vlv1 Wklv Ylwhuel0edvhg ghwhfwru lv hvvhqwldoo| wkh vdph dv wkh
ghflvlrq0gluhfwhg frkhuhqw ghwhfwru lqwurgxfhg lq ^8`/ h{fhsw wkdw lq wkh odwwhu
rqh v|pero0e|0v|pero ghflvlrqv duh pdgh lqvwhdg ri vhtxhqfh ghflvlrqv1
SVDP zdv ruljlqdoo| sursrvhg lq ^9` dqg odwhu dqdo|}hg lq ^:`1 Lq wklv
vfkhph nqrzq slorw v|perov duh pxowlsoh{hg lqwr wkh gdwd vhtxhqfh1 Wkh uh0
fhlyhg slorw v|perov pdnhv lw srvvleoh wr revhuyh wkh lq xhqfh ri wkh fkdqqho
dw wkh slorw srvlwlrqv1 Wkh fkdqqho hvwlpdwhv iru wkh gdwd v|perov duh wkhq
dfklhyhg e| lqwhusrodwlrq1 Vlqfh wkh slorw v|perov frqwdlq qr gdwd wkhuh lv d
vpdoo ryhukhdg fdxvlqj d edqgzlgwk h{sdqvlrq dqg dq hqhuj| orvv1 Krzhyhu/
wkh phwkrg kdv vkrzq wr eh urexvw lq idglqj hqylurqphqwv zlwk rqo| d vpdoo
orvv frpsduhg wr lghdo frkhuhqw ghwhfwlrq1 Lq dgglwlrq/ lw dovr doorzv pxowlohyho
vljqdoolqj1
Wkh xvh ri vxshusrvhg slorw vhtxhqfhv zdv lqwurgxfhg lq ^;` iru fduulhu
skdvh wudfnlqj1 Dv rssrvhg wr SVDP/ wkh slorw vhtxhqfh lv dgghg wr wkh
gdwd vhtxhqfh +vxshusrvhg, dqg wkhuhe| wudqvplwwhg lq wkh vdph iuhtxhqf|
edqg dqg dw wkh vdph wlph1 Vlqfh wkh slorw vhtxhqfh dqg wkh gdwd vhtxhqfh
duh vxshusrvhg wkhuh lv qr lqfuhdvh lq edqgzlgwk1 Lq ^<` wklv whfkqltxh zdv
sursrvhg iru frkhuhqw ghwhfwlrq ri gdwd v|perov1 Wkh VSS vfkhph lv urexvw
djdlqvw kljk Grssohu iuhtxhqflhv dqg lw hqdeohv pxowlohyho vljqdoolqj1 Krzhyhu/
wkh frpsoh{lw| jurzv zlwk wkh qxpehu ri vljqdoolqj dowhuqdwlyhv1
Wkh vwuhqjwkv ri SVDP dqg vxshusrvhg slorw v|perov fdq eh frpelqhg
lqwr d qhz k|eulg ghwhfwlrq vfkhph zkhuh qrw rqo| nqrzq slorw v|perov/ exw
dovr k|srwkhvhv ri wkh xqnqrzq gdwd v|perov/ duh xvhg iru fkdqqho hvwlpdwlrq1
Lq ^43` d vlplodu vfkhph zdv suhvhqwhg xvlqj ghflvlrqv ri wkh xqnqrzq gdwd
v|perov iru frqvhfxwlyh skdvh fruuhfwlrqv ri wkh fkdqqho hvwlpdwh1 Rxu qhz
vfkhph xvhv k|srwkhvhv ri wkh gdwd v|perov dv yluwxdo slorw v|perov dqg lw lv
wkhuhiruh ohvv vhqvlwlyh wr ghflvlrq huuruv1
Wkh frqwulexwlrqv ri wklv sdshu lqfoxgh erwk wkh qhz k|eulg ghwhfwlrq
vfkhph dqg d xqlhg dqdo|vlv iudphzrun iru sdluzlvh huuru hyhqwv/ wkdw fdq
eh dssolhg wr doo ri wkh deryh ghwhfwlrq vfkhphv1 Wkrxjk eh|rqg wkh vfrsh ri
wklv sdshu/ wkh sdluzlvh huuru suredelolwlhv suhvhqwhg fdq eh xvhg wr fdofxodwh
erxqgv iru wkh uhvxowlqj elw huuru udwh +EHU, xvlqj wkh prvw suredeoh huuru
vhtxhqfhv lq wkh vdph zd| dv lq ^6`1
5 Ghwhfwlrq Vfkhphv
Wkh dqdo|}hg ghwhfwlrq phwkrgv/ lqfoxglqj wkh glhuhqwldo rqhv/ uhtxluh vrph
nlqg ri fkdqqho hvwlpdwh iru ghwhfwlrq1 Frqyhqwlrqdo glhuhqwldo ghwhfwlrq
D FRPSDUDWLYH DQDO\VLV 6
~
Metric Decision
calculation
~
f
Channel A
estimator
^
f ~
C
Remodulation
y
Iljxuh 4= Eorfn vfkhph ri wkh ghwhfwru xvhg iru wkh dqdo|vlv1
xvhv h1j1 wkh odvw uhfhlyhg v|pero dv d fkdqqho hvwlpdwh1 Lq SVDP wkh uh0
fhlyhg slorw v|perov duh xvhg wr fdofxodwh fkdqqho hvwlpdwhv1 Zkhq xvlqj VSS
wkh fkdqqho hvwlpdwhv duh edvhg rq wkh nqrzq slorw vhtxhqfh wrjhwkhu zlwk
k|srwkhvhv ri wkh xqnqrzq gdwd vhtxhqfh1 Wkh jhqhudo uhfhlyhu xvhg iru wkh
dqdo|vlv ri doo ghwhfwlrq phwkrgv lv suhvhqwhg lq Ilj141
Lq wkh iroorzlqj/ zh xvh frpsoh{ edvhedqg qrwdwlrq dqg dvvxph shuihfw
wlph/ iuhtxhqf| dqg iudph v|qfkurql}dwlrq1 Wkh wudqvplwwhg gdwd v|perov
duh ghqrwhg dv fn 5 F1 Wkh v|pero fn lvshlwkhu d gdwd v|pero/ xn > d slorw
s
v|pero/ sn > ru d frpelqdwlrq ri erwk/ fn @ 4 n xn . n sn > zkhuh n lv
wkh dprxqw ri hqhuj| vshqw rq slorw lqirupdwlrq dw wlph n= Ohw |n ghqrwh wkh
uhfhlyhg v|pero vdpsohv/
|n @ fn in . qn > +4,
zkhuh in 5 F lv d pxowlsolfdwlyh +wlph0vhohfwlyh, idglqj surfhvv zlwk H^min m5 ` @
4/ dqg qn 5 F duh }hur0phdq zklwh Jdxvvldq qrlvh vdpsohv zlwk rqh0vlghg
srzhu vshfwudo ghqvlw| Q3 1 Zlwkrxw orvv ri jhqhudolw|/ wkh dyhudjh hqhuj| ri wkh
uhfhlyhg v|perov lv Hv @ H^mfn in m5 ` @ 4= Lw lv frqyhqlhqw wr xvh pdwul{ qrwdwlrq
dqg zh ghqrwh wkh uhfhlyhg vdpsoh yhfwru e| | @ +|4 > |5 > ===> |Q ,W @ Fi . q/
zkhuh i dqg q duh yhfwruv ri wkh idglqj surfhvv dqg wkh qrlvh/ uhvshfwlyho|/ dqg
F lv d gldjrqdo pdwul{ zlwk wkh wudqvplwwhg v|perov rq lwv gldjrqdo1
Wkh jrdo lv wr frpsxwh wkh pd{lpxp0olnholkrrg hvwlpdwh ri wkh wudqvplw0
whg vhtxhqfh1 Wkh uhfhlyhu whvwv doo srvvleoh k|srwkhvhv dqg ghflghv iru wkh
vhtxhqfh uhvxowlqj lq wkh orzhvw phwulf1 Iru Frqvwdqw0hqyhorsh prgxodwlrq/
wkh rswlpdo phwulf lv jlyhq ^6` dv
@ |K F+
. L,4 F K |> +5,
lv d gldjrqdo pdwul{ ri wkh k|srwkhvlv wr eh whvwhg +ghqhg odwhu rq,
zkhuh F
dqg lv wkh fryduldqfh pdwul{ ri wkh fkdqqho1 Vlqfh lw lv dgydqwdjhrxv wr kdyh
dq dgglwlyh phwulf lqfuhphqw iru hdfk qhz v|pero rqh riwhq xvh dssur{lpdwlrqv
lqvwhdg1 Wkh qrlvh lv dvvxphg wr eh Jdxvvldq/ dqg wkhuhiruh d vxlwdeoh phwulf
lqfuhphqw lv
fn > fnS > = = = > fnS , @ m|n fn in m5 >
n +
3 O +6,
zkhuh fn lv d k|srwkhvlv iru wkh n0wk gdwd v|pero dqg in lv d fkdqqho hvwlpdwh1
Wkh odwwhu lv ghulyhg iurp wkh gdwd k|srwkhvlv ^ xnS3 > = = = > x
nSO ` dqg2ru slorw
v|perov ^snS3 > = = = > snSO `1 Wkh sdudphwhu O lv wkh qxpehu ri v|perov xvhg
iru wkh fkdqqho hvwlpdwh dqg So ghqrwhv wkh wlph lqglfhv ri wkh uhfhlyhg v|perov
xvhg iru wkh hvwlpdwh1
Lq rughu wr pdnh ghflvlrqv/ wkh uhfhlyhu ghulyhv fkdqqho hvwlpdwhv dqg xvlqj
wkhvh lw whvwv doo srvvleoh k|srwkhvhv djdlqvw wkh uhfhlyhg v|perov1 Wkh fkdqqho
hvwlpdwhv duh fdofxodwhg dv iroorzv= D whqwdwlyh fkdqqho hvwlpdwh lv rewdlqhg
e| uh0prgxodwlrq/ ianSo @ |nSo @ fnSo = Jlyhq wkh k|srwkhvlv dqg wkh slorw
v|perov/ wkh fkdqqho hvwlpdwh in fdq eh frpsxwhg e| olqhdu owhulqj ^44`=
O[4 O
[4 |
nSo
in @ dn>So ianSo @ do = +7,
fnSo
o@3 o@3
Wkh owhu frh!flhqwv dn>p duh ghshqghqw rq wkh ghwhfwlrq phwkrg xvhg1 Iru
frqyhqwlrqdo glhuhqwldo ghwhfwlrq rqo| wkh odvw v|pero lv xvhg iru wkh fkdqqho
hvwlpdwh vr O @ 4> S3 @ 4> d4 @ 4> dqg doo rwkhu owhu frh!flhqwv duh }hur1 Li
d Zlhqhu hvwlpdwru lv xvhg/ dv lq h1j1 SVDP ru VSS/ wkh owhu frh!flhqwv duh
rewdlqhg dv wkh vroxwlrq ri wkh Zlhqhu0Krsi htxdwlrqv1
Wkh fkdqqho hvwlpdwh dw wlph n> edvhg rq d fhuwdlq k|srwkhvlv/ lv jlyhq e|
wkh suhylrxv uhfhlyhg v|perov dv
in @ dn F 4 |> +8,
zkhuh dn lv d yhfwru ri wkh fkdqqho hvwlpdwlrq frh!flhqwv dn>p / h1j1 wkh Zlhqhu
owhu frh!flhqwv/ dqg F lv d gldjrqdo pdwul{ frqwdlqlqj wkh k|srwkhvlv ri wkh
s s
wudqvplwwhg v|perov rq lwv gldjrqdo/ l1h1 fn>n @ 4 n xn . n sn x n =
Wkh fruuhvsrqglqj yhfwru ri fkdqqho hvwlpdwhv iru d fhuwdlq k|srwkhvlv fdq
frqvhtxhqwo| eh fdofxodwhg dv

i @ DF 4 |= +9,
Wkh hohphqwv lq wkh fkdqqho hvwlpdwlrq pdwul{ D dqg wkh uhprgxodwlrq pdwul{
4 duh ghwhuplqhg e| wkh ghwhfwlrq phwkrg1 Wkhvh duh ghvfulehg ehorz1
F
Li v|pero0e|0v|pero ghwhfwlrq lv xvhg wkhq wkh phwulf lqfuhphqw +6, lv xvhg
iru wkh ghflvlrqv dqg li vhtxhqfh ghwhfwlrq lv xvhg wkhq d vxp ri wkhvh lv xvhg
lqvwhdg1 Dgglqj wkh phwulf lqfuhphqwv dqg vxevwlwxwlqj wkh fkdqqho hvwlpdwhv
e| +9, ohdgv wr wkh fxpxodwlyh phwulf
@ |K +L FD
F 4 ,K +L FD
F 4 ,|= +:,
Ehorz zh vkruwo| ghvfuleh wkh dqdo|}hg ghwhfwlrq phwkrgv dqg jlyh wkhlu uh0
vshfwlyh fkdqqho hvwlpdwlrq pdwulfhv dqg k|srwkhvlv pdwulfhv1
514 Slorw V|pero Dvvlvwhg Prgxodwlrq

Zkhq xvlqj SVDP zh wudqvplw nqrzq slorw v|perov dw fhuwdlq wlph vorwv dqg
rqo| wkhvh v|perov duh xvhg wr fdofxodwh wkh fkdqqho hvwlpdwhv1 Wkh fkdqqho
hvwlpdwlrq pdwul{ kdv wkh owhu frh!flhqwv/ dn>So > dw srvlwlrqv zkhuh wkh slorw
v|perov duh orfdwhg/ doo rwkhu hohphqwv duh }hur/ l1h1
5 6
d4>S3 3 3 d4>S5 3
9 d5>S3 3 3 d5>S5 3 :
9 :
DS VDP @ 9 11 11 :=
7 1 1 8
dQ>S3 3 3 dQ>S5 3
Wkh k|srwkhvlv pdwul{ lv gldjrqdo dqg kdv wkh nqrzq slorw v|pero ydoxhv dw
wkhlu fruuhvsrqglqj srvlwlrqv dqg wkh ydoxhv iru wkh gdwd k|srwkhvlv wr eh whvwhg
dw wkh rwkhu srvlwlrqv/ l1h1
S VDP @ gldj+s4 > x5 > = = = > x

F n4 > sn > xn.4 > = = = ,=
515 Vxshusrvhg Slorw V|pero Ghwhfwlrq

Zkhq xvlqj vxshusrvhg slorw v|perov wkh nqrzq slorw vhtxhqfh lv vxshusrvhg
rq wkh xqnqrzq gdwd vhtxhqfh1 Wkh k|srwkhvhv duh wkhuhiruh d frpelqdwlrq ri
erwk nqrzq dqg xqnqrzq ydoxhv dqg wkh k|srwkhvlv pdwul{ kdyh wkh irup
s s
VS S @ gldj
F 4 +
x4 > = = = > x
Q , . +s4 > = = = > sQ ,
zkhuhdv wkh fkdqqho hvwlpdwlrq pdwul{ kdyh wkh irup

5 6
d4>3 3 3 3
9 d5>4 d5>3 3 3 :
9 :
DVS S @ 9 1 11 :=
7 1 1 1 8
3 3 dQ>O4 dQ>3
516 Pxowlsoh0V|pero Glhuhqwldo Ghwhfwlrq

Lq pxowlsoh0v|pero glhuhqwldo ghwhfwlrq vhyhudo glhuhqwldoo| hqfrghg v|perov
duh ghwhfwhg dw wkh vdph wlph1 Wkh ghflvlrq lv wdnhq ryhu Q v|perov dqg wkh
uhfhlyhu dlpv wr qg d gdwd vhtxhqfh wkdw doljqv wkh uhfhlyhg skdvhv qrw rqo|
ehwzhhq frqvhfxwlyh v|perov/ exw dovr ehwzhhq qrq0qhljkerulqj rqhv1 Wkh
uhihuhqfhv duh fdofxodwhg xvlqj k|srwkhvhv lq wkh vdph zd| dv iru vxshusrvhg
slorw v|perov/ vr DPVGG @ DVS S = Wkhuh lv qr nqrzq lqirupdwlrq wudqvplwwhg
vr wkhuhiruh wkh k|srwkhvlv pdwul{ frqvlvwv ri wkh xqnqrzq gdwd v|perov rqo|/
PVGG @ gldj+
F x4 > = = = > x
Q ,=
517 K|eulg Ghwhfwlrq

Wkh k|eulg ghwhfwru xvhv erwk slorw v|perov dqg k|srwkhvhv wr irup wkh fkdqqho
hvwlpdwhv1 Wkh slorw v|perov hqvxuh wkdw wkhuh lv ohvv skdvh dpeljxlw| dqg
wkh k|srwkhvhv khosv wkh fkdqqho hvwlpdwru wr wudfn wkh fkdqqho yduldwlrqv
ehwzhhq slorw v|perov1 Lq wklv zd| lw lv srvvleoh wr/ h1j1/ pdqdjh odujh Grssohu
iuhtxhqflhv zlwkrxw kdylqj wrr pxfk ryhukhdg gxh wr slorw v|perov1 Khuh
djdlq/ wkh fkdqqho hvwlpdwlrq pdwul{ kdv wkh vdph irup dv iru vxshusrvhg slorw
v|perov/ DK|eu @ DVS S / exw wkh k|srwkhvlv pdwul{ lv wkh vdph dv iru SVDP/
K|eu @ F
l1h1/ F S VDP =
518 Glhuhqwldo Ghwhfwlrq

Lq frqyhqwlrqdo glhuhqwldo ghwhfwlrq wkh ghflvlrqv duh pdgh v|pero0e|0v|pero1
Wkh suredelolw| ri fruuhfw ghflvlrqv fdq ehfdofxodwhg
lq wkh xqlhg iudphzrun
e| wkh fkdqqho hvwlpdwlrq pdwul{ DGG @ 4 3 dqg wkh k|srwkhvlv pdwul{

xn4 3
FGG @ =
3 x
n
Dv rssrvhg wr wkh rwkhu phwkrgv/ wkh ghflvlrqv duh pdgh v|pero e| v|pero
zkhq xvlqj frqyhqwlrqdo glhuhqwldo ghwhfwlrq1 Wkhuhiruh/ wkh dqdo|vlv suh0
vhqwhg lq qh{w vhfwlrq lv xvhg wr vwxg| rqo| vlqjoh0huuru hyhqwv iru wklv vfkhph1
6 Ghwhfwru Shuirupdqfh
Lq wklv vhfwlrq zh ghulyh dqdo|wlfdo h{suhvvlrqv iru wkh suredelolw| ri sdluzlvh
huuru hyhqwv1 Wkh dqdo|vlv lv wr vrph h{whqw vlplodu wr wkh rqh lq ^6`/ exw zh dovr
surylgh forvhg irup h{suhvvlrqv iru wkh glvwulexwlrq ri wkh ghflvlrq yduldeoh1
Wkh uhfhlyhu pdnhv d fruuhfw ghflvlrq li wkh phwulf iru wkh fruuhfw sdwk lv
vpdoohu wkdq doo rwkhu sdwk phwulfv1 Ghqh wkh ghflvlrq yduldeoh a @

>zkhuh dqg duh wkh fxpxodwlyh phwulfv ri wkh k|srwkhvlv dqg fruuhfw
sdwk/ uhvshfwlyho|1 Li a A 3 iru doo k|srwkhvhv wkhq wkh fruuhfw sdwk zloo eh

fkrvhq1 E| uhzulwlqj wkh uhfhlyhg vdpsohv dv | @ Gj zlwk G @ F L dqg
W
j@ i q > wkh ghflvlrq yduldeoh fdq eh h{suhvvhg
a @ |K ^+L FD
F 4 ,K +L FD
F 4 , +;,

+L FDF ,K +L FDF 4 ,`|
4
@ jK Nj=

Wkh fryduldqfh pdwul{ ri wkh }hur0phdq Jdxvvldq yduldeohv lq j lv

Ui 3
Uj @ > +<,
3 5q L
D FRPSDUDWLYH DQDO\VLV :
zkhuh Ui lv wkh fryduldqfh pdwul{ ri wkh idglqj surfhvv1 Iru 50G lvrwurslf
vfdwwhulqj Ui kdv hohphqwv ul>m @ M3 +5iG pd{ Wv +l m,,/ zkhuh M3 +, lv wkh
}hurwk rughu Ehvvho ixqfwlrq/ iG pd{ lv wkh pd{lpxp Grssohu iuhtxhqf| dqg
Wv lv wkh v|pero wlph1
614 Sdluzlvh Huuru Suredelolw|

Pdq| dxwkruv kdyh vwxglhg wkh sdluzlvh huuru suredelolw| uhvxowlqj iurp txdgudwlf
irupv olnh +;,/ h1j1 lq ^45`^46`1 Iroorzlqj wkhp zh ghqh qhz Jdxvvldq yduldeohv
j3 @ Ej> zkhuh E lv d pdwul{ ri vl}h 5Q u ixooolqj Uj @ EEK dqg mEm @ 9 3>
zkhuh u lv wkh udqn ri Uj = Lq ^45` lw lv vkrzq wkdw wkh ghflvlrq yduldeoh fdq eh
uhjdughg dv d vxp ri u lqghshqghqw zhljkwhg "5 0yduldeohv/ zkhuh wkh zhljkwv
duh wkh hljhqydoxhv ri EK NE= Wkh yhfwru ri Jdxvvldq vdpsohv lv frpsoh{ dqg
wkh "5 0yduldeohv wkhuhe| kdyh 5 ghjuhhv ri iuhhgrp1
E| vruwlqj wkh zhljkwv vr wkdw wkh uvw p hljhqydoxhv duh srvlwlyh dqg wkh
odvw Qv p hljhqydoxhv duh qhjdwlyh wkh ghflvlrq yduldeoh fdq eh h{suhvvhg dv
p
[ Qv
[
a@
q "55>q q "55>q= +43,
q@4 q@p.4
Dvvxplqj wkdw doo hljhqydoxhv duh glvwlqfw/ wkh suredelolw| ghqvlw| ixqfwlrq ri
a fdq eh fdofxodwhg dv ^45`

; p
A S +q ,Qv 5 h|@+5q ,
A
A %
Q
& |3
A @4 Tv
A
? q 5 +
q n ,
n@4>n9@q
ia +|, @ Qv +44,
A
A S + q ,Qv 5 h|@+5q&,
A
A
%
Qv | ? 3=
A
= q@p.4 5 T +n q ,
n@4>n9@q
Wkh sdluzlvh huuru suredelolw|/ l1h1 wkh suredelolw| wkdw wkh ghflvlrq yduldeoh
ehfrphv qhjdwlyh vr wkdw wkh uhfhlyhu pdnhv d ghflvlrq lq idyru ri wkh huurqhrxv
k|srwkhvlv v lqvwhdg ri wkh fruuhfw rqh/ lv wkhq jlyhq e|
+q ,Qv 4
Qv
[
Sv @ Q = +45,
q@p.4
Tv
+n q ,
n@4>n9@q
7 Frpsdulvrq edvhg rq Huuru Hyhqwv

Wkh dqdo|vlv wrrov surylghg lq wkh suhylrxv vhfwlrq fdq eh xvhg wr ghwhuplqh ds0
sur{lpdwh elw0 dqg iudph0huuru udwhv iru wkh ydulrxv ghwhfwlrqv phwkrgv1 Krz0
hyhu/ khuh zh rqo| vwxg| wkh sdluzlvh huuru suredelolwlhv +45, iru vrph vhohfwhg
huuru hyhqwv1 Wklv zloo uhyhdo vrph ri wkh lqkhuhqw vwuhqjwkv dqg zhdnqhvvhv ri
wkh glhuhqw ghwhfwlrq phwkrgv1
; D FRPSDUDWLYH DQDO\VLV
Iluvw zh h{hpsoli| wkh suredelolw| ri huuru hyhqwv ri glhuhqw ohqjwkv lq d

eorfn0idglqj Ud|ohljk fkdqqho +iG Wv @ 3,1 Wklv vkrzv wkh dpeljxlw| ri dq0
wlsrgdo vhtxhqfhv iru vxshusrvhg slorwv lq frkhuhqw v|vwhpv dqg wkh lq xhqfh
ri d phpru| lq wkh ghwhfwru iru glhuhqwldo v|vwhpv1 Vhfrqgo|/ zh h{hpsoli|
wkh urexvwqhvv ri wkhvh phwkrgv djdlqvw yduldwlrqv lq dyhudjh VQU dqg Grssohu
hhfwv1 Wklv vlqjohv rxw wkh srru shuirupdqfh ri SVDP dqg glhuhqwldo gh0
whfwlrq dw odujh Grssohu vsuhdgv dqg kljk VQUv/ zkhq frpsduhg wr wkh rwkhu
phwkrgv1
Zh lqyhvwljdwh sdluzlvh huuru suredelolwlhv iru uxqv ri n frqvhfxwlyh wuhoolv
vwdjhv xvlqj ESVN1 Wkh eorfnv duh ri ohqjwk Q @ 84> zklfk lv dvvxphg wr
eh orqj hqrxjk wr dssur{lpdwh wkh vwhdg| vwdwh ehkdylru/ dqg wkh fkdqqho
hvwlpdwlrq owhu ohqjwkv duh O @ 81 Wkh n vwdjhv ri ghyldwlrq iurp wkh }hur0
sdwk ri wkh wuhoolv duh sodfhg lq wkh plggoh ri wkh eorfn dqg zh kdyh wzr nlqgv ri
lqwhuhvwlqj huuru hyhqwv= huuru vhtxhqfhv zkhuh n lv vpdoo vr wkdw wkh huurqhrxv
vhtxhqfh phujhv zlwk wkh fruuhfw vhtxhqfh diwhu n v|perov dqg huuru vhtxhqfhv
zkhuh wkh odvw +kdoi, sduw lv huurqhrxv1 Wkh odwwhu uhsuhvhqwv d vlqjoh v|pero
huuru diwhu glhuhqwldo ghwhfwlrq1 Zh vhsdudwh wkh frkhuhqw dqg glhuhqwldo
vfkhphv lq wkh h{dpsohv1 Iru wkh frkhuhqw phwkrgv hyhu| ghyldwlrq iurp wkh
wuxh vhtxhqfh zloo uhvxow lq d v|pero huuru1 Iru wkh glhuhqwldo phwkrgv rqo|
wkh wudqvlwlrqv ehwzhhq wkh fruuhfw dqg huurqhrxv vhtxhqfh fdxvh v|pero huuruv
diwhu glhuhqwldo ghfrglqj1 Wklv phdqv wkdw dq huuru hyhqw ri n frqvhfxwlyh
huuruv rqo| surgxfhv wzr qdo v|pero huuruv1
Lq Ilj1 5 zh looxvwudwh wkh sdluzlvh huuru suredelolw| iru huuru hyhqwv ri
ohqjwk 4/ 5/ 6/ 7/ dqg 59 +orqj, lq d eorfn0idglqj Ud|ohljk fkdqqho dqg dq
He @Q3 ri 43 gE4 1 Zh kdyh 8( slorw hqhuj| iru wkh vxshusrvhg slorw vfkhph
dqg hyhu| 43wk v|pero dv d slorw v|pero lq wkh rwkhu wzr frkhuhqw vfkhphv1
Wkh vfkhphv vkrz udwkhu vlplodu ehkdylru iru vkruw hyhqwv/ exw vhsdudwh iru
orqj hyhqwv1 Wkh srruhu shuirupdqfh ri wkh VSS dqg k|eulg vfkhphv iru orqj
huuru hyhqwv lv h{sodlqhg e| wkh lqkhuhqw dpeljxlw| ri dqwlsrgdo vhtxhqfhv vlqfh
wkhuh lv d srvvlelolw| wkdw wkh zurqj k|srwkhvlv lv xvhg iru fkdqqho hvwlpdwlrq1
Iru wkh PVGG vfkhph/ wkhuh duh rqo| wzr v|pero huuruv iru vkruw huuru
hyhqwv dqg rqh v|pero huuru iru wkh orqj hyhqw1 Dq lqwhuhvwlqj ehkdylru iru wklv
vfkhph lv wkdw wzr v|pero huuruv forvh wr hdfk rwkhu kdyh d kljkhu suredelolw|
wkdq wzr elw huuruv vhsdudwhg iurp hdfk rwkhu1 Wklv lv gxh wr wkh phpru| ri
wkh PVGG/ zkhuh wkh qhjdwlyh lq xhqfh ri hduolhu v|pero huuruv ghfuhdvhv zlwk
wlph1
Lq Ilj1 6 zh suhvhqw wkh sdluzlvh huuru suredelolw| ri hyhqwv fdxvlqj d vlqjoh
elw huuru lq wkh wudqvplvvlrq eorfn1 Iru wkh frkhuhqw phwkrgv wklv fruuhvsrqgv
wr dq huuru hyhqw ri ohqjwk rqh/ zkloh iru wkh glhuhqwldo vfkhphv lw lv wkh htxly0
dohqw ri d orqj huuru hyhqw1 Dprqj wkh frkhuhqw vfkhphv lq Ilj1 6 +wrs,/ wkh
glhuhqfh lv uhodwlyho| vpdoo iru wkh eorfn idglqj fkdqqho +ig Wv @ 3,1 Krzhyhu/
zkhq wkh Grssohu iuhtxhqf| lqfuhdvhv +ig Wv @ 3=38,/ wkh olplwdwlrq ri SVDP
4 Lq wkh ruljlqdo sxeolfdwlrq d fxuyh iru frqyhqwlrqdo glhuhqwldo ghwhfwlrq zdv dovr suh0
vhqwhg1 Lq wklv uhylvhg yhuvlrq wkh fxuyh lv rplwwhg dqg wkh wh{w kdv ehhq fkdqjhg dffruglqjo|1
D FRPSDUDWLYH DQDO\VLV <
04
43
SVDP
VSS
K|eu
PVGG
1
e
r
u
s

u
r
u
u 05
h

h
43
v
l
z
u
l
d
S
06
43 4 5 6 7 orqj
Ohqjwkrihuuruhyhqw
Iljxuh 5= Sdluzlvh huuru suredelolw| iru huuru vhtxhqfhv ri glhuhqw ohqjwkv1
1 3
e
r43
u
s 04

u43
r
u
u 05
h
43
h
v
l 06 SVDP
z43 VSS
u
l
d 07 K|eu
S43
1 3
e
r43
u
s 04

u43
r
u
u 05
h43

h
v
l 06
z43 Glii
u
l
d 07 PVGG
S43
3 8 43 48 53 58 63
He2Q3+gE,
Iljxuh 6= Sdluzlvh huuru suredelolw| iru glhuhqw VQU ydoxhv1 Vrolg olqhv
uhsuhvhqw qr Grssohu +iG Wv @ 3, zkhuhdv gdvkhg olqhv uhsuhvhqw idvw idglqj
+iG Wv @ 3=38,=
ehfrphv dssduhqw1 Gxh wr wkh olplwdwlrqv lq lfwhg e| wkh vdpsolqj wkhruhp/

wkh fkdqqho hvwlpdwlrq lq SVDP ehfrphv srru +rqo| rqh ri 43 v|perov lv d
slorw, dqg wkh uhvxowlqj huuru rru lv txlwh vhyhuh1 Zlwk vxshusrvhg slorwv zh
duh qrw forvh wr wkh olplw ri wkh vdpsolqj wkhruhp dw wkh kljkhu Grssohu iuh0
txhqflhv/ dqg wkh sdluzlvh huuru suredelolw| lv kdugo| dhfwhg1 Wklv fohduo|
vkrzv wkh vwuhqjwk ri vxshusrvhg slorwv dw kljk Grssohu iuhtxhqflhv/ nhhslqj
lq plqg wkdw wkh dprxqw ri hqhuj| vshqw rq slorw lqirupdwlrq lv hyhq ohvv lq
wklv fdvh frpsduhg wr SVDP1
Iru wkh glhuhqwldo vfkhphv lq Ilj1 6 +erwwrp, zh revhuyh wkh lqfuhdvhg
shuirupdqfh e| lqwurgxflqj d phpru| lq wkh glhuhqwldo ghwhfwlrq1 Wklv lv wuxh
iru erwk wkh eorfn0idglqj fkdqqho +ig Wv @ 3, dqg zkhq wkh fkdqqho lv idglqj
zlwklq wkh eorfn +ig Wv @ 3=38,1 Dw kljk Grssohu iuhtxhqf| dqg orz VQUv/
wkh huuru suredelolw| lv grplqdwhg e| wkh dgglwlyh qrlvh dqg wkh wzr ghwhfwlrq
phwkrgv kdyh udwkhu vlplodu shuirupdqfh1 Krzhyhu/ zkhq wkh VQU lqfuhdvhv/
wkh gudzedfn ri frqyhqwlrqdo glhuhqwldo ghwhfwlrq ehfrphv dssduhqw1 Gxh wr
skdvh fkdqjhv lq wkh fkdqqho ehwzhhq wzr frqvhfxwlyh v|perov/ wkh glhuhqwldo
ghwhfwru h{shulhqfh d txlwh vhyhuh huuru rru/ zkhuhdv wkh pxowlsoh0v|pero
glhuhqwldo ghwhfwru zlwk lwv phpru| fdq frpshqvdwh iru wkhvh1
Xvlqj d vlqjoh dqdo|wlfdo h{suhvvlrq iru sdluzlvh huuru hyhqwv/ zh kdyh pdq0
djhg wr fdswxuh pdq| ri wkh lqkhuhqw fkdudfwhulvwlfv ri d zkroh fodvv ri wudqv0
plvvlrq2ghwhfwlrq vfkhphv1 E| lghqwli|lqj vljqlfdqw huuru hyhqwv lw lv srvvleoh
wr ghulyh d +wuxqfdwhg, xqlrq erxqg iru wkh huuru udwhv/ dqg wkhuhe| hvwlpdwlqj
wkh ryhudoo shuirupdqfh1 Wklv lv/ krzhyhu/ rxw ri vfrsh iru wkh fxuuhqw sdshu
dqg ohiw iru ixuwkhu uhvhdufk1
8 Frqfoxvlrqv
Zh kdyh dqdo|}hg wkh shuirupdqfh dqg ehkdylru ri glhuhqw ghwhfwlrq vfkhphv
lq Ud|ohljk idglqj fkdqqhov1 Edvhg rq d xqlhg dqdo|wlfdo iudphzrun zh kdyh
fdofxodwhg wkh sdluzlvh huuru suredelolw| iru fhuwdlq frkhuhqw dqg qrq0frkhuhqw
ghwhfwlrq phwkrgv1 E| vwxg|lqj wkh yduldwlrq ri wkh sdluzlvh huuru suredelolwlhv
zh kdyh frpsduhg wkh urexvwqhvv ri wkhvh phwkrgv djdlqvw yduldwlrqv lq dyhudjh
VQU dqg Grssohu hhfwv1 Dv h{shfwhg/ wkh vfkhphv xvlqj d jurxs ri v|perov
forvh wr wkh v|pero wr eh ghwhfwhg wr ghulyh d fkdqqho hvwlpdwh duh vkrzq wr
eh yhu| urexvw lq idvw idglqj hqylurqphqwv dqg vkrz qr huuru rru1 Frqyhq0
wlrqdo glhuhqwldo ghwhfwlrq/ rq wkh rwkhu kdqg/ vxhuv iurp d qrlv| fkdqqho
hvwlpdwh zkhuhdv SVDP pd| vxhu iurp dq revrohwh fkdqqho hvwlpdwh lq idvw
idglqj fkdqqhov1 Wkh odwwhu fdq eh dyrlghg e| lqfuhdvlqj wkh qxpehu ri slorw
v|perov/ exw wkhq wkh ryhukhdg lqfuhdvhv1 Lq wklv sdshu zh kdyh dovr lqwurgxfhg
d qhz ghwhfwlrq vfkhph wkdw frpelqhv wkh vwuhqjwkv ri xvlqj slorw0 dqg gdwd
v|perov wr ghulyh fkdqqho hvwlpdwhv1 Wkh slorw v|perov hqvxuhv wkdw wkhuh lv
ohvv skdvh dpeljxlw| zkloh wkh xvh ri xqnqrzq gdwd v|perov pdnhv lw srvvleoh
wr wudfn fkdqqho yduldwlrqv ehwzhhq wkh slorw v|perov1
Wkh suhvhqwhg dqdo|vlv lv jhqhudo dqg fdq hdvlo| eh dssolhg wr rwkhu ghwhfwlrq
vfkhphv1 Iru d ixuwkhu dqdo|vlv/ wkh h{suhvvlrqv ri wkh sdluzlvh huuru suredelolw|
fdq eh xvhg wr jhw dssur{lpdwlyh h{suhvvlrqv ri wkh uhvxowlqj elw dqg iudph huuru
udwhv e| xvlqj xqlrq erxqg whfkqltxhv1
Dfnqrzohgjhphqw
Wkh dxwkruv zrxog olnh wr wkdqn Suri1 Shwhu Krhkhu dw Xqlyhuvlw| ri Nlho/
Jhupdq|/ iru lqvslulqj wklv zrun1
Uhihuhqfhv
^4` Gdulxvk Glyvdodu dqg Pduylq N1 Vlprq1 Pxowlsoh0v|pero glhuhqwldo gh0
whfwlrq ri PSVN1 LHHH Wudqv1 Frppxq1/ 6;+6,=63363;/ Pdufk 4<<31
^5` Gdulxvk Glyvdodu dqg Pduylq N1 Vlprq1 Pd{lpxp0olnholkrrg glhuhqwldo

ghwhfwlrq ri xqfrghg dqg wuhoolv0frghg dpsolwxgh skdvh prgxodwlrq ryhu
DZJQ dqg idglqj fkdqqhov Phwulfv dqg shuirupdqfh1 LHHH Wudqv1
Frppxq1/ 75+4,=:9;</ Mdqxdu| 4<<71
^6` Sdxo Kr dqg Grplqlf Ixqj1 Huuru shuirupdqfh ri pxowlsoh0v|pero glhu0

hqwldo ghwhfwlrq ri SVN vljqdov wudqvplwwhg ryhu fruuhodwhg Ud|ohljk idglqj
fkdqqhov1 LHHH Wudqv1 Frppxq1/ 73+43,=4899489</ Rfwrehu 4<<51
^7` Jlrujlr P1 Ylwhwwd dqg Ghvprqg S1 Wd|oru1 Ylwhuel ghfrglqj ri glhuhq0

wldoo| hqfrghg SVN vljqdov wudqvplwwhg ryhu Ud|ohljk iuhtxhqf|0 dw idglqj
fkdqqhov1 LHHH Wudqv1 Frppxq1/ 76+52627,=4589458</ 4<<81
^8` Duqh Vyhqvvrq1 Frkhuhqw ghwhfwru edvhg rq olqhdu suhglfwlrq dqg ghflvlrq
ihhgedfn iru GTSVN1 Hohfwurq1 Ohww1/ 63+53,=49754976/ Vhswhpehu 4<<71
^9` Plfkdho O1 Prkhu dqg Mrkq K1 Orgjh1 WFPS D prgxodwlrq dqg frg0
lqj vwudwhj| iru Ulfldq0idglqj fkdqqhov1 LHHH M1 Vhohfw1 Duhdv Frppxq1/
:+<,=467:4688/ Ghfhpehu 4<;<1
^:` Mdphv N1 Fdyhuv1 Dq dqdo|vlv ri slorw0v|pero dvvlvwhg prgxodwlrq iru

Ud|ohljk0idglqj fkdqqhov1 LHHH Wudqv1 Yhklf1 Whfkqro1/ 73+7,=9;99<6/
Qryhpehu 4<<41
^;` W1S1 Kroghq dqg N1 Ihkhu1 D vsuhdg vshfwuxp edvhg v|vwhp whfkqltxh iru
Eurdgf1/ sdjhv 4;84<7/ Vhswhpehu 4<<31
^<` Shwhu Krhkhu dqg Iuhguln Wxiyhvvrq1 Fkdqqho hvwlpdwlrq zlwk vxshulp0
srvhg slorw vhtxhqfh1 Lq Surf1 Jorehfrp/ sdjhv 54955499/ Ulr gh Mdqhlur/
Eud}lo/ Ghfhpehu 4<<<1
^43` Jduulfn W1 Luylqh dqg Shwhu M1 PfOdqh1 V|pero0dlghg soxv ghflvlrq0

gluhfwhg uhfhswlrq iru SVN2WFP prgxodwlrq rq vkdgrzhg preloh vdwhoolwh
idglqj fkdqqhov1 43+;,=45;<45<</ Rfwrehu 4<<51
^44` M1K1 Orgjh dqg P1O1 Prkhu1 Pd{lpxp olnholkrrg vhtxhqfh hvwlpdwlrq ri
FSP vljqdov wudqvplwwhg ryhu Ud|ohljk dw0idglqj fkdqqhov1 LHHH Wudqv1
Frppxq1/ 6;+9,=:;::<7/ Mxqh 4<<31
^45` D1P1 Pdwkdl dqg Vhujh E1 Suryrvw1 Txdgudwlf irupv lq udqgrp yduldeohv=
Wkhru| dqg dssolfdwlrqv1 Pdufho Ghnnhu/ Qhz \run/ X1V1D1/ 4<<51
^46` Fkulvwldq Vfkohjho1 Huuru suredelolw| fdofxodwlrq iru pxowlehdp Ud|ohljk
fkdqqhov1 LHHH Wudqv1 Frppxq1

GHVLJQ Ri Zluhohvv FRPPXQLFDWLRQ V - VWHPV LVVXHV RQ V - Qfkurql) DWLRQ/ Fkdqqho HVWLPDWLRQ DQG Pxowl0Fduulhu V - VWHPV

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

GHVLJQ Ri Zluhohvv FRPPXQLFDWLRQ V - VWHPV LVVXHV RQ V - Qfkurql) DWLRQ/ Fkdqqho HVWLPDWLRQ DQG Pxowl0Fduulhu V - VWHPV

Uploaded by

Copyright:

Available Formats

Ghvljq ri Zluhohvv Frppxqlfdwlrq V|vwhpv

 Lvvxhv rq V|qfkurql}dwlrq/ Fkdqqho

Ghsduwphqw ri Dssolhg Hohfwurqlfv

f Iuhguln Wxiyhvvrq/ 5333

Zkhq L vwduwhg dv d Sk1G1 vwxghqw p| dgylvru/ Wruohly Pdvhqj/ h{sodlqhg wr

D I1 Wxiyhvvrq dqg W1 Pdvhqj1 Slorw dvvlvwhg fkdqqho hvwlpdwlrq iru RIGP

E I1 Wxiyhvvrq dqg W1 Pdvhqj1 Rswlpl}dwlrq ri vxe0fkdqqho edqgzlgwk iru

F I1 Wxiyhvvrq/ P1 Idxonqhu/ dqg W1 Pdvhqj1 Suh0frpshqvdwlrq iru Ud|ohljk

G I1 Wxiyhvvrq dqg R1 Hgiruv1 Suhdpeoh0edvhg wlph dqg iuhtxhqf| v|q0

H I1 Wxiyhvvrq/ P1 Idxonqhu/ S1 Krhkhu/ dqg R1 Hgiruv1 RIGP wlph

L Ryhuylhz ri wkh Uhvhdufk Duhd 4

6 Vxppdu| dqg Frqwulexwlrqv 68

61514Slorw Dvvlvwhg Fkdqqho Hvwlpdwlrq iru RIGP lq Preloh

Ryhuylhz ri wkh Uhvhdufk

Udglr wudqvplvvlrq kdv doorzhg shrsoh wr frppxqlfdwh zlwkrxw dq| sk|vlfdo

ehhq dq lqfuhdvlqj lqwhuhvw lq kljk0vshhg zluhohvv orfdo duhd qhwzrunv +Z0

414 Gljlwdo Frppxqlfdwlrq V|vwhpv

source channel de-

Iljxuh 414= Ixqfwlrqdo eorfnv lq d frppxqlfdwlrq v|vwhp1

 RIGP v|vwhp ghvljq1 Ruwkrjrqdo iuhtxhqf| glylvlrq pxowlsoh{ +RIGP,

 Fkdqqho hvwlpdwlrq1 Lq rughu wr dfklhyh jrrg shuirupdqfh wkh uhfhlyhu

 V|qfkurql}dwlrq1 Wkh uhfhlyhu kdv wr ghwhuplqh zkhq wkhuh lv d vljqdo wr

Dv phqwlrqhg ehiruh/ wkh wudqvplwwhg lqirupdwlrq lv uhsuhvhqwhg e| glhuhqw

415 Zluhohvv V|vwhpv

vlghudwlrqv1 Lq d zluhg qhwzrun/ wkhuh duh qr idvw pryhphqwv ri whuplqdov ru

41514 Sdfnhw0Edvhg yv1 Flufxlw0Vzlwfkhg V|vwhpv

41515 Frkhuhqw yv1 Qrq0frkhuhqw V|vwhpv

Iljxuh 416= Eorfn ghvfulswlrq ri d frqyhqwlrqdo RIGP prgxodwru dqg ghprg0

41614 Vlqjoh0Fduulhu yv1 Pxowl0Fduulhu V|vwhpv

Lq wklv fkdswhu d ghwdlohg edfnjurxqg dqg vrph ri wkh pdlq sxeolfdwlrqv lq

514 Fkdqqho Hvwlpdwlrq

f @ duj pd{ i +umf> k,=

Lq RIGP v|vwhpv wkhuh lv qr LVL diwhu ghprgxodwlrq +xqghu lghdo frqgl0

51414 Slorw V|pero Dvvlvwhg Prgxodwlrq

l ,/ iru wkh wudqvihu ixqfwlrq fdq eh rewdlqhg e| edfn0urwdwlqj wkh

zn>o +S, @ 4  n>o > +519,

Mn>o @ 5K Wn>o 4 n>o > +51:,

dqg  5q ghqrwhv wkh yduldqfh ri wkh qrlvh1

observation time prediction time

pilot symbol data symbol

Iljxuh 514= Wkh sulqflsoh ri suh0frpshqvdwlrq vkrzlqj wzr RIGP eorfnv1

{+w, dujiKa +w,j

515 Pxowl0Fduulhu V|vwhp Ghvljq

51514 Vxe0Fkdqqho Edqgzlgwk

zkhuh %l lv wkh qrupdol}hg hhfwlyh wlph rvhw ghqhg dv

51515 Suhdpeoh0Edvhg V|qfkurql}dwlrq

Iljxuh 515= Vwuxfwxuh ri wkh frqyhqwlrqdo v|qfkurql}hu1 Wklv fruuhvsrqgv wr

Iljxuh 516= Vwuxfwxuh ri wkh suh0fruuhodwlqj v|qfkurql}hu1 Wklv fruuhvsrqgv wr

dyhudjhg ehiruh pxowlsolfdwlrq lv shuiruphg1 Wkhuhiruh/ wkh phdq dpsolwxgh ri

51516 Slorw Sdwwhuq Ghvljq

5iG Wv QW   pd{ ivf QI  4@5> +514;,

Iljxuh 517= H{dpsoh ri d uhfwdqjxodu slorw sdwwhuq1 QI lv wkh slorw glvwdqfh lq

516 Vxshulpsrvhg Slorw Vhtxhqfhv

51614 Wlph dqg Iuhtxhqf| V|qfkurql}dwlrq iru RIGP

51615 Fkdqqho Hvwlpdwlrq xvlqj Vxshulpsrvhg Slorw Vh0

rwkhu dqdo|}hg phwkrgv lqfoxgh SVDP/ pxowlsoh0v|pero glhuhqwldo ghwhfwlrq/

614 Jhqhudo Uhvxowv

615 Sdshu Frqwulexwlrqv

61515 Rswlpl}dwlrq ri Vxe0Fkdqqho Edqgzlgwk iru Pr0

Lq wklv sdshu wkh edqgzlgwk ri wkh vxe0fkdqqhov lv rswlpl}hg zlwk uhvshfw

61516 Suh0Frpshqvdwlrq iru Ud|ohljk Idglqj Fkdqqhov lq

51 skdvh dqg dpsolwxgh frpshqvdwlrq xs wr d fhuwdlq srzhu ohyho/ deryh

61 skdvh dqg dpsolwxgh frpshqvdwlrq xs wr d fhuwdlq srzhu ohyho/ deryh

Lvvxhv rq V|qfkurql}dwlrq/ Fkdqqho

f Iuhguln Wxiyhvvrq/ 5333

RIGP v|vwhp ghvljq1 Ruwkrjrqdo iuhtxhqf| glylvlrq pxowlsoh{ +RIGP,

Fkdqqho hvwlpdwlrq1 Lq rughu wr dfklhyh jrrg shuirupdqfh wkh uhfhlyhu

V|qfkurql}dwlrq1 Wkh uhfhlyhu kdv wr ghwhuplqh zkhq wkhuh lv d vljqdo wr

Dv phqwlrqhg ehiruh/ wkh wudqvplwwhg lqirupdwlrq lv uhsuhvhqwhg e| glhuhqw

zn>o +S, @ 4 n>o > +519,

Mn>o @ 5K Wn>o 4 n>o > +51:,

dqg 5q ghqrwhv wkh yduldqfh ri wkh qrlvh1

zkhuh %l lv wkh qrupdol}hg hhfwlyh wlph rvhw ghqhg dv

5iG Wv QW pd{ ivf QI 4@5> +514;,

rwkhu dqdo|}hg phwkrgv lqfoxgh SVDP/ pxowlsoh0v|pero glhuhqwldo ghwhfwlrq/

h{whqvlyh vlpxodwlrqv dqg wkh lq xhqfh ri owhu ohqjwkv/ qrq0rswlpdo owhuv/

6151: D Frpsdudwlyh Dqdo|vlv ehwzhhq Glhuhqw Ghwhf0

^4` ;35144e04<<< vxsssohphqw wr lqirupdwlrq whfkqrorj|whohfrppxqlfdwlrqv

^54` G1 Glyvdodu dqg P1 N1 Vlprq1 Pxowlsoh0v|pero glhuhqwldo ghwhfwlrq ri

^5<` S1 Kr dqg G1 Ixqj1 Huuru shuirupdqfh ri pxowlsoh0v|pero glhuhqwldo gh0