IST-2000-30148 I-METRA
D3.2
Implemen a ion o Rele an algo i hms
Con ac ual Da e o Deli e y o he CEC: 31 Oc obe , 2002
Ac ual Da e o Deli e y o he CEC: 31 Oc obe , 2002
Au ho (s): Ja ie R Fonollosa, Roge Gaspa, Ma kku Heikkilä, Jean Philippe
Ke moal, Ka i Majonen, Xa ie Mes e, Alba Pagès, Adam Polla d,
Lau en Schumache , Ami Wiesel, Juha Yli alo
Pa icipan (s): UPC, AAU, NET, NMP, VOD
Wo kpackage: WP3: Re-con igu able Algo i hm s udy
Es . pe son mon hs: 3
Secu i y: Public
Na u e: Repo
Ve sion: 0.1
To al numbe o pages: 33
Abs ac :
This deli e able p o ides a high le el desc ip ion o he so wa e de eloped wi hin he I-
METRA p ojec ollowing he selec ion epo ed in D3.1 "Design, Analysis and Selec ion o
Sui able Algo i hms".
Keywo d lis : Adap i e An ennas, MIMO Sys ems, Space-Time Coding, Adap i e
Modula ion and Coding, HSDPA, HSUPA, Tu bo Coding.
IST-2000-30148 I-METRA
Doc.: IST-2000-30148/UPC-WP3-D3.2-V1.2.doc
Ti le: I-METRA D3.1: Design, Analysis and
Selec ion o sui able algo i hms
3 July, 2002
File: UPC-WP3-D3.2-V1.2.doc
Page 2 o 33
0 EXECUTIVE SUMMARY
This deli e able p o ides a high le el desc ip ion o he so wa e de eloped wi hin he I-
METRA p ojec o he HSDPA and HSUPA o UMTS ollowing he selec ion epo ed in
D3.1 "Design, Analysis and Selec ion o Sui able Algo i hms".
Wi h espec o he HSDPA se ices, only he high le el desc ip ion o link le el simula ions
is p esen ed. The ansmi e ollows he HS-DSCH speci ica ions al eady epo ed in D3.1 in
such a way ha his deli e able is mainly ocused o desc ibe he so wa e o he space- ime
block codes implemen ed in he ansmi e as well as he ecep ion schemes.
The he ein epo ed HSDPA ansmission echniques include he ansmi an enna swi ching
echnique, ou di e en space- ime block codes based on he STTD (i.e., CSTD, T ombi,
DSTTD and DABBA), he well-known V-BLAST sys em and he mo e gene al Linea
Dispe sion Codes. The channel is implemen ed using he MIMO channel model de eloped in
WP2. Rega ding he ecei e , se e al schemes a e desc ibed: he In e e ence Cancelling and
Nulling echnique de eloped o V-BLAST, he LMMSE Space-Time Equalize echnique
and di e en subop imal ML ecei e s (such as, he “Tu bo algo i hm”).
Wi h espec o HSUPA, he simula ion model includes ull 3GPP ea u es and has been
designed o bo h a low-complexi y UE (1 an enna UE) and a dual-an enna wi h MIMO
capabili y. This app oach allows a common e e ence poin o ad anced MIMO concep as
well as ealis ic signal models and ecei e implemen a ion schemes. A s anda d Rake
ecei e s uc u e is assumed bu he model allows also he implemen a ion o mo e complex
ecei e s wi h in e e ence supp ession in a s aigh - o wa d manne .
IST-2000-30148 I-METRA
Doc.: IST-2000-30148/UPC-WP3-D3.2-V1.2.doc
Ti le: I-METRA D3.1: Implemen a ion o he
Sui able algo i hms
31 Oc obe , 2002
File: UPC-WP3-D3.2-V1.2.doc
Page 3 o 33
DISCLAIMER
The wo k associa ed wi h his epo has been ca ied ou in acco dance wi h he highes echnical s anda ds and
he I-METRA pa ne s ha e endea ou ed o achie e he deg ee o accu acy and eliabili y app op ia e o he
wo k in ques ion. Howe e since he pa ne s ha e no con ol o e he use o which he in o ma ion con ained
wi hin he epo is o be pu by any o he pa y, any o he such pa y shall be deemed o ha e sa is ied i sel as
o he sui abili y and eliabili y o he in o ma ion in ela ion o any pa icula use, pu pose o applica ion.
Unde no ci cums ances will any o he pa ne s, hei se an s, employees o agen s accep any liabili y
wha soe e a ising ou o any e o o inaccu acy con ained in his epo (o any u he consolida ion,
summa y, publica ion o dissemina ion o he in o ma ion con ained wi hin his epo ) and/o he connec ed
wo k and disclaim all liabili y o any loss, damage, expenses, claims o in ingemen o hi d pa y igh s.
IST-2000-30148 I-METRA
Doc.: IST-2000-30148/UPC-WP3-D3.2-V1.2.doc
Ti le: I-METRA D3.1: Implemen a ion o he
Sui able algo i hms
31 Oc obe , 2002
File: UPC-WP3-D3.2-V1.2.doc
Page 5 o 33
0 EXECUTIVE SUMMARY .......................................................................................................................... 2
1 INTRODUCTION ........................................................................................................................................ 6
2 HIGH SPEED DOWNLINK PACKET ACCESS (HSDPA) .................................................................... 7
2.1 HSDPA LINK-LEVEL SIMULATOR ............................................................................................................. 7
2.2 DIVERSITY MIMO TECHNIQUES ................................................................................................................ 9
2.2.1 STTD ................................................................................................................................................ 9
2.2.2 Ci cula Shi ed T ansmi Di e si y (CSTD) ................................................................................. 10
2.2.3 T ombi ............................................................................................................................................ 11
2.2.4 T ansmi An enna Swi ching .......................................................................................................... 11
2.3 LAYERED MIMO TECHNIQUES ................................................................................................................ 12
2.3.1 DSTTD ........................................................................................................................................... 12
2.3.2 V-BLAST ....................................................................................................................................... 13
2.3.3 T ans o med 4-An enna Double-Ra e Block Code (DABBA) ...................................................... 14
2.4 ADVANCED RECEIVER STRUCTURES ........................................................................................................ 15
2.4.1 Space-Time Equalize s ................................................................................................................... 15
2.4.2 Maximum Likelihood Recei e s .................................................................................................... 17
2.5 COMBINATION OF LINEAR DISPERSION CODES WITH TURBO SPACE TIME DECODING. ............................ 19
3 HIGH SPEED UPLINK PACKET ACCESS (HSUPA) ......................................................................... 25
3.1 INTRODUCTION ......................................................................................................................................... 25
3.2 SIMULATION MODEL ................................................................................................................................. 25
3.2.1 3GPP compa ibili y ........................................................................................................................ 25
3.2.2 High-le el modelling o uplink high-speed da a wi h MIMO app oach ........................................ 25
3.2.3 Signal s uc u e ............................................................................................................................... 27
3.2.4 T ansmi e s uc u e ...................................................................................................................... 28
3.2.5 Recei e s uc u e ........................................................................................................................... 29
3.2.6 MIMO channel p ope ies .............................................................................................................. 30
3.2.7 Pe o mance measu e ..................................................................................................................... 31
3.2.8 Simula ion cases ............................................................................................................................. 31
4 REFERENCES ........................................................................................................................................... 33
IST-2000-30148 I-METRA
Doc.: IST-2000-30148/UPC-WP3-D3.2-V1.2.doc
Ti le: I-METRA D3.1: Implemen a ion o he
Sui able algo i hms
31 Oc obe , 2002
File: UPC-WP3-D3.2-V1.2.doc
Page 6 o 33
1 INTRODUCTION
[UPC]
IST-2000-30148 I-METRA
Doc.: IST-2000-30148/UPC-WP3-D3.2-V1.2.doc
Ti le: I-METRA D3.1: Implemen a ion o he
Sui able algo i hms
31 Oc obe , 2002
File: UPC-WP3-D3.2-V1.2.doc
Page 7 o 33
2 HIGH SPEED DOWNLINK PACKET ACCESS (HSDPA)
2.1 HSDPA Link-Le el Simula o
Figu e 1 illus a es he HSDPA link-le el simula o which is applicable wi h mos o
he selec ed MIMO ansmission schemes. The simula o is implemen ed in CoCen ic
Sys em S udio and COSSAP en i onmen s.
T ansmi e
Da a bi s a e i s g ouped in o packe s which a e hen channel encoded. A he same
ime, he packe da a is s o ed in a memo y slo o possible la e e ansmissions.
A e punc u ing ( a e ma ching) and in e lea ing he signal is modula ed using he
selec ed modula ion scheme (QPSK o 16-QAM). The symbols a e ei he space- ime
encoded using he selec ed space- ime coding (di e si y) scheme o mul iplexed in o
se e al pa allel s eams acco ding o he chosen laye ed scheme. Each an enna symbol
s eam is sp ead using he selec ed numbe o sp eading codes. A his phase, common
pilo signal is added o he an enna signals as well as he selec ed numbe o
in e e ing speech o HSDPA code channels.
Channel
METRA MIMO channel model gene a es he desi ed co ela ion be ween he
an ennas and simula es he desi ed mul ipa h ading channel. Addi i e whi e Gaussian
noise (AWGN) is used o model in e cell in e e ence. I is also possible o gene a e
ac ual in e e ing base s a ion (BS) signal(s). This is especially impo an when space-
ime equalizing ecei e s a e applied since hey bene i om he s uc u ed
in e e ence compa ed o con en ional ecei e s.
Recei e
A mul ian enna ecei e de ec s he da a bi s using channel es ima es ob ained using
he common pilo channel. Only he complex channel gains a e es ima ed while he
channel delays a e always assumed o be known. Es ima ion is pe o med by
(mo ing-a e age) il e ing he common pilo based aw channel es ima es. One o he
ecei e algo i hms desc ibed in he ollowing sec ions is used. The ecei e has
pe ec knowledge o he index o he packe i has ecei ed and, in case o a
e ansmi ed packe , i combines he so symbol es ima es in o a combine memo y
slo . Chase-combining based hyb id-ARQ has been applied.
Depunc u ing ope a ion (o in e se a e ma ching) es o es he o iginal code ame
leng h o he channel decode . Cyclic edundancy check (CRC) bi s in he code ame
a e no used bu he co ec ness o inco ec ness o he decoded packe is de ec ed
ideally. This in o ma ion is applied in gene a ion o an ARQ command o he ecei e .
The ansmission o ARQ is no simula ed explici ly bu he ACK/NACK message is
gi en o he ansmi e wi hou e o s. BS ansmi e does no apply any scheduling
bu e ansmi s he eques ed packe a e a ixed, pa ame e ized delay.
IST-2000-30148 I-METRA
Doc.: IST-2000-30148/UPC-WP3-D3.2-V1.2.doc
Ti le: I-METRA D3.1: Implemen a ion o he
Sui able algo i hms
31 Oc obe , 2002
File: UPC-WP3-D3.2-V1.2.doc
Page 8 o 33
The main pa ame e s o he HSDPA link-le el simula o a e:
• Channel coding a e ( u bo code wi h punc u ing o a e ma ching)
• Modula ion me hod (QPSK o 16-QAM)
• Powe and numbe o HSDPA code channels o he desi ed use
• Powe and numbe o in e e ing HSDPA code channels
• Powe and numbe o in e e ing speech use s
• Common pilo channel powe
• Maximum numbe o e ansmissions pe packe
• Re ansmission delay
Va ious pe o mance da a is collec ed du ing he simula ion including:
• aw (non-decoded) bi e o a e
• decoded bi e o a e
• da a h oughpu (kbi /s) as a unc ion o bo h a e age geome y pa ame e (G)
and maximum allowable numbe o packe e ansmissions
• esidual a e age packe e o a e (PER) as a unc ion o bo h a e age
geome y pa ame e (G) and maximum allowable numbe o e ansmissions
• dis ibu ion o packe -wise signal- o-noise a io
• PER as a unc ion o packe -wise signal- o-noise a io (only i s ansmission
o each packe is included in his igu e)
The las pe o mance igu e is used o gene a ing inpu o sys em simula ions.
IST-2000-30148 I-METRA
Doc.: IST-2000-30148/UPC-WP3-D3.2-V1.2.doc
Ti le: I-METRA D3.1: Implemen a ion o he
Sui able algo i hms
31 Oc obe , 2002
File: UPC-WP3-D3.2-V1.2.doc
Page 9 o 33
Figu e 1. S uc u e o link-le el HSDPA simula o .
2.2 Di e si y MIMO Techniques
Di e si y MIMO echniques u ilize he mul iplici y o ansmi an ennas o ob ain
ansmi di e si y. In HSDPA, da a a e and sys em capaci y may inc ease indi ec ly
h ough a lowe packe e o a e, highe suppo ed modula ion alphabe , o highe
channel coding a e.
2.2.1 STTD
In STTD ansmission an o hogonal space- ime block code is applied o e wo
consecu i e da a symbols. STTD coding o symbols s1 and s2 is illus a ed in Figu e 2.
Columns o he code ma ix C(s1, s2) ep esen di e en symbol in e als and ows
di e en ansmi an ennas. The code is used as a basis o se e al o he mul ian enna
ansmission schemes.
Figu e 2. Space-Time T ansmi Di e si y scheme o HSDPA.
MIMO
channel
&
AWGN
Mul ian enna
modula o
&
sp eading
Bi sou ce
Packe gen.
Re ans.
memo y
Channel encode Punc u ing/
a e ma ching In e lea e
Pilo signal gen.
In e . use gen.
Mul ian enna
ecei e
Re ans.
combine
memo y
Channel decode Depunc u ing/
a e ma ching Dein e l.
ARQ command
gene a o
Pe o mance e alua o
Channel
es ima o
ú
û
ù
ê
ë
é
-
=**
12
21
21 ),( ss
ss
ssC
IST-2000-30148 I-METRA
Doc.: IST-2000-30148/UPC-WP3-D3.2-V1.2.doc
Ti le: I-METRA D3.1: Implemen a ion o he
Sui able algo i hms
31 Oc obe , 2002
File: UPC-WP3-D3.2-V1.2.doc
Page 16 o 33
whe e
s
m2 is a iance o dm(i) and he co a iance ma ix o he mul ian enna signal is
(2.9)
whe e E(×) deno es expec a ion. The co a iance ma ix is es ima ed as a ime a e age.
I is also possible o use adap i e echniques o implemen an LMMSE equalize .
Figu e 9 shows he s uc u e o he LMMSE space- ime equalize o he signal om
ansmi an enna m ollowed by a code co ela o . An M-an enna ansmission scheme
equi es M pa allel space- ime equalize s he ou pu s o which a e decoupled o code
co ela o s and ollowed by a p ope decode o de ec o o he applied MIMO
ansmission scheme. The p ocessing can assume ha he channel has only one
mul ipa h. This can also be seen in Figu e 9 whe e only a single code co ela o is
used o de ec ion o he symbol ansmi ed om an enna m.
Figu e 9. Basic s uc u e o LMMSE space- ime equalize ollowed by a
sp eading code co ela o .
A p ope y o he LMMSE ecei e s uc u e in Figu e 9 is ha he ecei e can be
unc ionally spli in o wo pa s: (i) so-called p e il e ing and (ii) con en ional RAKE
de ec ion including decoding o he possibly used space- ime block code. The
p e il e ing pe o ms he unc ion o he in e ed C (i) ma ix Figu e 9. The in e se
ma ix can be shown o app oach a block ma ix wi h Toepli z blocks. This implies
ha he ma ix mul iplica ion ope a ion can be eplaced by linea il e ing. Wi h
ï
ï
þ
ï
ï
ý
ü
ï
ï
î
ï
ï
í
ì
÷
÷
÷
÷
÷
ø
ö
ç
ç
ç
ç
ç
è
æ
=
)()()()()()(
)()()()()()(
)()()()()()(
E)(
HH
2
H
1
H
2
H
22
H
12
H
1
H
21
H
11
iiiiii
iiiiii
iiiiii
i
NNNN
N
N
C
(i)
hm,1(i)
Channel
es ima o
P/S
1(i)
N(i)
hm,N(i)
dm(i)Co ela o
Mul ipa h
combine
Mul ipa h
combine
Mul ipa h
combine
+
C
-1(i)
IST-2000-30148 I-METRA
Doc.: IST-2000-30148/UPC-WP3-D3.2-V1.2.doc
Ti le: I-METRA D3.1: Implemen a ion o he
Sui able algo i hms
31 Oc obe , 2002
File: UPC-WP3-D3.2-V1.2.doc
Page 17 o 33
mul iple RX an ennas his con e sion om block-wise ma ix ope a ion o con inuous
linea il e ing equi es a bank o pa allel linea il e s, as shown in Figu e 10.
The bene i o he p e il e -RAKE s uc u e is ha , ega dless o he numbe o he
o igin ( ansmi an enna, da a laye , base s a ion e c.) o he signals o be de ec ed, he
same p e il e bank can be used p io o con en ional RAKE- ype ecei e o con e
he ecei e in o an LMMSE space- ime equalize .
Figu e 10. P e il e -RAKE s uc u e o LMMSE space- ime equalize .
2.4.2 Maximum Likelihood Recei e s
Op imal maximum likelihood (ML) ecei e s canno be applied in p ac ice due o
complexi y easons. By conside ing in e symbol, mul iple access and in e cell
in e e ence as addi i e noise, i is possible o o mula e a subop imal ML ecei e
which is use ul especially wi h laye ed schemes (e.g. BLAST, DSTTD) when linea
space- ime equalize s do no supp ess he in e laye in e e ence e ec i ely enough.
The assump ions imply ha he ecei e can unc ion in one-sho mode de ec ing each
symbol in e al o decoding each space- ime code wo d sepa a ely.
Assuming a gene al L-pa h channel, and N ecei e an ennas, he ecei e desp eads all
NL signals using as many pa allel code co ela o s. The inpu signal o he ML
ecei e is o med by he co ela o ou pu s and can be simply w i en as
. (2.10)
Symbol ec o s(k) holds he unknown symbols and H(k) is he channel ma ix.
Vec o n(k) includes all such signal componen s ha a e conside ed as noise in he
es ima ion p ocess. (I should be no ed ha he abo e model applies di ec ly only o
he e y basic V-BLAST scheme. In mos cases, he e ec o space- ime block coding
ha e o be inco po a ed in o he signal model as shown e.g. in Sec ion 2.3.1.)
(i)
h1(i)
Channel
es ima o
P/S
1(i)
N(i)
hN(i)
P e il e 1
P e il e 2
P e il e N
b(k)
Con en ional
RAKE de ec o
g1(i)
g2(i)
gN(i)
)()()()( kkkk nsHy +=
IST-2000-30148 I-METRA
Doc.: IST-2000-30148/UPC-WP3-D3.2-V1.2.doc
Ti le: I-METRA D3.1: Implemen a ion o he
Sui able algo i hms
31 Oc obe , 2002
File: UPC-WP3-D3.2-V1.2.doc
Page 18 o 33
The unknown symbol ec o s(k) can be es ima ed using ML p inciple as
, (2.11)
whe e Cnn is he noise co a iance ma ix. The noise co a iance can ei he be assumed
o be a uni ma ix (I) o i can be es ima ed. ML solu ion is ound h ough exhaus i e
sea ch o e all possible combina ions o he unknown symbols.
Since he ecei e complexi y g ows exponen ially as a unc ion o he numbe o
unknown pa ame e s in ec o s, i is possible o apply he ML p inciple o a subse o
he unknown pa ame e s while conside ing he emaining pa ame e s as addi i e noise
(by including hem in Cnn). A subse can be selec ed based on he ins an aneous
signal- o-noise a ios o he pa ame e s. A e de ec ion, he subse is cancelled om
y(k), and hen he emaining unknown pa ame e s can be es ima ed.
Applica ion o Space-Time Block Coding
When ML ecep ion is applied e.g. o DSTTD ansmi ed signal, a p oblem is ha
block coding canno be p esen ed as a linea ans o ma ion o inpu symbol ec o s
(due o he ac ha complex conjuga ion is no a linea ope a ion). In his case, he
signal model ha e o be modi ied as shown in ea lie sec ions a e which he abo e
ML solu ion is applicable.
Two possible ML app oaches can be conside ed when he modi ied signal model is
used: (i) o apply signal y(k) di ec ly, o (ii) o apply signal HH(k)y(k) as an inpu o
he ML de ec o . The la e echnique has he bene i ha , by i s decoding he block
code, one o se e al in e e ing symbols cancel ou om he signal (since some
elemen s o HH(k)H(k) equal o ze o) hus simpli ying he ML sea ch. Howe e , a he
same ime, he noise e m becomes e y colou ed implying ha he noise co a iance
ma ix canno be neglec ed. The la e me hod also i s be e o cohe en combining
o packe e ansmissions o ob ain chase-combining gain.
( ) ( )
{ }
)()()()()()()(mina g)(
ˆ1
H
)(
kkkkkkkk
ksHyCsHys nn
s
--=-
IST-2000-30148 I-METRA
Doc.: IST-2000-30148/UPC-WP3-D3.2-V1.2.doc
Ti le: I-METRA D3.1: Implemen a ion o he
Sui able algo i hms
31 Oc obe , 2002
File: UPC-WP3-D3.2-V1.2.doc
Page 19 o 33
Figu e 11. Maximum likelihood ecei e s uc u e.
So Ou pu Gene a ion
ML de ec ion gene a es ha d decisions o he symbols. Bi -wise so ou pu s o he
channel decode can be gene a ed e.g. by compu ing a io
(2.12)
whe e e(k) is he bes o all such symbol ec o s which co espond o a bi ec o in
which he bi o he in e es is o he han in . This echnique au oma ically akes
in o accoun he applied G ay bi - o-symbol mapping.
2.5 Combina ion o Linea Dispe sion Codes wi h Tu bo Space Time
Decoding.
The ansmi e is based on a combina ion o a u bo encode , linea dispe sion space
ime codes [Hassibi01], and mul icode sp eading. Fi s , a ame o aw da a bi s is
encoded using a s anda d Tu bo Code wi h a e R. The coded bi s a e in e lea ed, and
modula ed using Mc bi s pe eal symbol. These symbols a e hen mul iplexed in o he
P pa allel ST encode s.
h1(i)
Channel
es ima o
1(i)
N(i)
hN(i)
Co ela o 1
Co ela o 2
Co ela o NL
s(k)
Maximum-
likelihood
es ima o
y1(k)
y2(k)
yNL(k)
2(i)
( ) ( )
( ) ( )
)(
ˆ
)()()()(
ˆ
)()(
)()()()()()()(
1
H
1
H
kkkkkkk
kkkkkkk
sHyCsHy
eHyCeHy
nn
nn
--
--
-
-
)(
ˆks
IST-2000-30148 I-METRA
Doc.: IST-2000-30148/UPC-WP3-D3.2-V1.2.doc
Ti le: I-METRA D3.1: Implemen a ion o he
Sui able algo i hms
31 Oc obe , 2002
File: UPC-WP3-D3.2-V1.2.doc
Page 20 o 33
Figu e 12. – T ansmi e s uc u e.
Each LDC ST encode b akes i s symbol sequence in o B blocks o 2Q symbols. Each
block will be ansmi ed h ough T ime slo s. Be o e ansmission, he ou pu o he
encode s a e sp eaded using Hadama d codes.
Figu e 13. – Space ime encode and sp eade .
The o e all ou pu o he ansmi e ’s M an ennas a he i’ h chip can be exp essed as
a linea unc ion o all he symbols in he associa ed block:
(2.13)
whe e is a ma ix ep esen ing he sp eading codes, is a ma ix
ep esen ing he LDC s uc u e, deno es he K onecke p oduc , is a ec o
ep esen ing he ansmi ed symbols (o all he pa allel encode s) a he b’ h
block, and we de ine he indexes , and ollow he ollowing s uc u e:
Encode In e lea e Mod
LDC
x
x
x
c(1)
LDC :
x
x
x
c(2)
LDC :
x
x
x
c(P)
•
•
•
•
•
•
MUX
#2
P b anches
#1
+
+
#M
+
M b anches
LDC
x
x
x
2Q symbols
. . .
. . .
T channel uses
M b anches
c(p)
MxT ma ix MxT SF ma ix
( )( ) ( )( )
( )
( )( )
BTSFiibi ici T××=Ä=,...,1)( sDcx
( )
cc
( )
D
Ä
( )
bs
2QP
()ci
() i
()bi
IST-2000-30148 I-METRA
Doc.: IST-2000-30148/UPC-WP3-D3.2-V1.2.doc
Ti le: I-METRA D3.1: Implemen a ion o he
Sui able algo i hms
31 Oc obe , 2002
File: UPC-WP3-D3.2-V1.2.doc
Page 21 o 33
The op imal ecei e o he p e ious p oblem o mula ion is imp ac ical. I in ol es a
Vi e bi Algo i hm o he e ec i e supe - ellis associa ed wi h he conca ena ed
encode s, and he equency selec i e channel. The complexi y o such a solu ion is
oo high o implemen a ion. We he e o e eso o a subop imal app oach, usually
e e ed o as he ``Tu bo Algo i hm''. The use o in e lea e s decouples he
demodula ion s age and he decoding s age. This allows o i e a i e exchange o so
in o ma ion be ween hem:
Figu e 14. – I e a i e ecei e .
The so in o ma ion is exp essed using log likelihood a ios (LLR). The inpu o each
s age is he a p io i LLRs deno ed by . Using hem, i calcula es he a pos e io i
LLRs, deno ed by , which a e supposedly ``be e ''. The di e ence be ween
hese LLRs is usually e e ed o as ex insic LLR and is deno ed by . The
ex insic LLRs o each s age a e used as he a p io i LLRs o he o he s age. This
p ocedu e is epea ed in an i e a i e ashion.
(2.14)
The decode uses he well known BCJR algo i hm [Polla a96]. In b e i y, his
algo i hm calcula es he a pos e io i LLR o each bi as a unc ion o he a p io i LLR
o all he o he bi s and he co ela ions be ween hem induced by he encode 's ellis.
This algo i hm is well known, he e o e we lea e i s de ails o e e ences.
The demodula ion algo i hm calcula es he ex insic LLR o each bi as a unc ion o
he a p io i LLR o all he o he bi s and he ecei ed samples. This algo i hm is based
on he so inpu so ou pu , linea mul iuse de ec o , e.g., [Wang99].
Bb
T
b
T
SFccSFccSFccSFcc
=
==
=
==
======== 1111
1
1
1
Demod Dein e lea e Decod
LE1
LE2
+
In e lea e
LA1
LA2
Ha d Decision
-
Recei ed da a
om he b h block
s(b)
^
Es ima ed
symbols
y(b)
()
A
Ld
()
D
Ld
()
E
Ld
( ) ( ) ( )
(1/) (1) (/ 1)
log log log
(1/) (1) (/ 1)
DAE
Ld Ld Ld
Pd Pd P d
Pd Pd P d
===
=+
=-=-=-
yy
yy
IST-2000-30148 I-METRA
Doc.: IST-2000-30148/UPC-WP3-D3.2-V1.2.doc
Ti le: I-METRA D3.1: Implemen a ion o he
Sui able algo i hms
31 Oc obe , 2002
File: UPC-WP3-D3.2-V1.2.doc
Page 22 o 33
Figu e 15. – So inpu so ou pu demodula o .
The demodula o is basically a weigh ed leas squa es linea es ima o . The es ima o
app oxima es he ecei ed ec o as a linea unc ion o he unknown symbols, and
hen uses a p io i s a is ical in o ma ion on hese symbols o de i e he linea
es ima o . The es ima es a e hen con e ed o ex insic LLRs.
The linea es ima o uses he ollowing app oxima ion o he ecei ed ec o
s uc u e:
(2.15)
wi h
, , ,
and
.
Due o he equency selec i e na u e o he channel, his s uc u e is only an
app oxima ion. Howe e , as will be explained in he sequel, using he DFE block
alida es his app oxima ion.
y(b)
LA1
DFE Linea
Es ima o s o LLR
^
LLR o s
^
y(b)
~
LA1
LE1
s(b-1)
^
A p io i s a is ical
in o ma ion on he
symbols
s(b)
^
() ()() () 1bbbbbB»+=yGsw
{ }
{ }
Re ( )
() Im ( )
b
b
b
éù
=êú
ëû
y
yy
{ }
{ }
Re ( )
Im ( )
b
b
b
éù
=êú
ëû
HD
G
HD
{ }
{ }
Re ( )
() Im ( )
b
b
b
éù
=êú
ëû
w
w
w
( )
( )
( )
( )
( )
( )
( )
( )
( )
( ) ( )
( ) ( )
( )
( ) ( )
( )
( ) ( )
( )
( )
( )
( )
(0)
(1) (0)
(1) (0)
() 1 1
11 0 0
12 120
()
00
0
11
() 1 1
T
TT
L
T
TT T
TT T
T
TT
b b SF T b SF T
b SF T
b SF T b SF T
b
b SF T b SF T
c SF T
b b SF T b SF T
-
éù
=-×+××
ëû
éù
-×+
êú
-×+-×+
êú
=êú
êú
êú
×× ××
ëû
éù
=Ä Ä Ä
êú
ëû
éù
=-×+××
ëû
yy y
H
HH
H
HH
Dc D c D c D
ww w
IST-2000-30148 I-METRA
Doc.: IST-2000-30148/UPC-WP3-D3.2-V1.2.doc
Ti le: I-METRA D3.1: Implemen a ion o he
Sui able algo i hms
31 Oc obe , 2002
File: UPC-WP3-D3.2-V1.2.doc
Page 23 o 33
Using he well known ma ix in e sion lemma, e.g., [Kay93], i is possible o show
ha he weigh ed leas squa es es ima e o he unknown symbols is ( o simplici y we
omi he block indexes):
(2.16)
whe e and a e he a p io i expec a ion and co a iance o he unknown
symbols. These a p io i symbols s a is ics can be easily de i ed om he a p io i bi s
LLRs:
(2.17)
whe e deno es he ec o o all he bi s associa ed wi h he symbol . Using
hese momen s i is easy o exp ess and :
(2.18)
The e o e, he demodula o de i es he a p io i s a is ics, and hen applies he
weigh ed leas squa es es ima o s. Finally, he las s age o he demodula o
ans o ms he symbols es ima es in o bi s LLRs using he ollowing o mula:
(2.19)
whe e in he las equa ion, deno es he ec o o all he associa ed bi s excluding
he cu en bi .
As p e iously explained, he ecei e assumes a block channel s uc u e. In o de o
alida e his assump ion, we p opose he use o a decision eedback equalize (DFE),
which will mi iga e he in e block in e e ence. Thus, be o e applying he es ima o
o each block, he DFE ies o elimina e he in e e ence o he p e ious block:
( )
( )
( )
1
2
11
ˆ
11
H
SS
H
SS
HH
KK
diag
s
-
=-+
=+
éù éù
=ëû ëû
sΣG W y E E
W GC G I
Σ G WG G WG
S
E
SS
C
( )
( )
( ) ( )
[ ]
( ) ( )
( ) ( )
/2
/2 /2
2
2
Ai
Ai Ai
ki
Ld
i
iLd Ld
ki
i
Es s Pd
e
Pd
ee Es s Pd
b
b
-+
éù
=êú
ëû
== Þ
+éù
éù
=êú
ëû ëû
åÕ
åÕ
d
d
d
d
d
( )
sd
S
E
SS
C
[ ]
[ ]
[ ]
[ ]
22
11
1
22
00
00
00
S SS
KKK
Es E s
Es
Es Es E s
éù
éù
-
éù ëû
êú
êú
==
êú
êú êú
êú éù
-
ëû êú
ëû
ëû
EC
( ) ( )
( )
( )
( )
21
2
2
ˆ
21
2
2
ˆ
1ˆ
log exp 1,
2
1ˆ
log exp 1,
2
jk
jk
T
Dj k j j A
s
T
kj j A
s
Ld s sd
ssd
s
s
¹
¹
¹
¹
ìü
éù
ïï
=--=++
êú
íý
êú
ïï
ëû
îþ
ìü
éù
ïï
---=-+
êú
íý
êú
ïï
ëû
îþ
å
å
d
d
ddL
ddL
j¹
d
j
d
IST-2000-30148 I-METRA
Doc.: IST-2000-30148/UPC-WP3-D3.2-V1.2.doc
Ti le: I-METRA D3.1: Implemen a ion o he
Sui able algo i hms
31 Oc obe , 2002
File: UPC-WP3-D3.2-V1.2.doc
Page 24 o 33
(2.20)
o and , whe e is ob ained om equa ion (¡E o ! No se
encuen a el o igen de la e e encia.) by subs i u ing he symbols wi h hei
p e ious es ima es.
( )
( )
( )
( )
( )
( )
( )
( )
1
() ˆ
11 1 1
L
l
li
bBi bBi bBil bBil
-
=
-+= -+- -+- -+-
å
yy H x
2bB=
1iL=
ˆ
x
IST-2000-30148 I-METRA
Doc.: IST-2000-30148/UPC-WP3-D3.2-V1.2.doc
Ti le: I-METRA D3.1: Implemen a ion o he
Sui able algo i hms
31 Oc obe , 2002
File: UPC-WP3-D3.2-V1.2.doc
Page 25 o 33
3 HIGH SPEED UPLINK PACKET ACCESS (HSUPA)
3.1 In oduc ion
In he ollowing we conside he simula ion models o bo h a low-complexi y UE (1-
an enna UE) and a dual-an enna UE wi h MIMO capabili y. The single an enna case
se es as a e e ence case wi h "s anda d" 1-code ansmission bu mul icode
ansmission is also conside ed. In addi ion, a mul i-an enna UE wi h single Tx-chain
may be in e es ing om iewpoin o implemen a ion complexi y and cos . This
app oach allows an enna swi ching in an open loop mode and an enna selec ion using
eedback om Node B.
A e conside ing he esul s om he ini ial s udy and aking in o accoun he
complexi y issues o he Node B, he ollowing cases we e selec ed o de ailed
simula ion s udy. This app oach was chosen also because i ep esen s a s aigh -
o wa d e olu ion pa h om he UE capabili ies o he cu en 3GPP speci ica ion.
The e o e, in his s udy, we assume ha he signal s uc u e o Uplink Common
Packe Access using he Random Access Channel (RACH) is employed. I mus be
no ed, howe e , ha concep p oposals o uplink high-speed packe access has no
been conside ed ye in 3GPP. Fo example, i is no possible o ob ain uplink channel
s a e in o ma ion a Node B because no e e ence channel exis s o uplink. Mo eo e ,
he e a e no eedback channels om Node B o UE o ca y he channel s a e
in o ma ion.
3.2 Simula ion model
3.2.1 3GPP compa ibili y
The simula ion model has been designed so ha i model includes ull 3GPP ea u es.
This app oach allows a common e e ence poin o ad anced MIMO concep as well
as ealis ic signal models and ecei e implemen a ion schemes. In a i s s age a
s anda d Rake ecei e s uc u e is assumed bu he model allows also he
implemen a ion o mo e complex ecei e s wi h in e e ence supp ession in a s aigh -
o wa d manne .
3.2.2 High-le el modelling o uplink high-speed da a wi h MIMO app oach
The de ailed simula ion model includes ea u es such as:
• packe mode da a wi h ARQ
• channel coding (con olu ional/ u bo ) wi h di e en code a es
• in e lea ing
• a e ma ching / punc u ing
• as powe con ol
• ealis ic in e e ence modelling (64 kb/s da a use s)
IST-2000-30148 I-METRA
Doc.: IST-2000-30148/UPC-WP3-D3.2-V1.2.doc
Ti le: I-METRA D3.1: Implemen a ion o he
Sui able algo i hms
31 Oc obe , 2002
File: UPC-WP3-D3.2-V1.2.doc
Page 32 o 33
- Known channel s a e
- Adap i i y on he link le el: ansmission du ing a o able channel s a e
- An enna swi ching w/wo ARQ
- Scheduling no conside ed
Case 4 ( en a i e):
• Mul i-code ia 1 Tx-an enna s. mul i-codes ia 2 Tx an ennas
- Re e ence case: 1x2, 1x4
- E alua ing uplink peak da a a e wi h MIMO app oach
IST-2000-30148 I-METRA
Doc.: IST-2000-30148/UPC-WP3-D3.2-V1.2.doc
Ti le: I-METRA D3.1: Implemen a ion o he
Sui able algo i hms
31 Oc obe , 2002
File: UPC-WP3-D3.2-V1.2.doc
Page 33 o 33
4 REFERENCES
[I-METRA D3.1] IST-2000-30148 I-METRA, Deli e able 3.1, "Design, Analysis
and Selec ion o Sui able Algo i hms", a ailable a www.is -
ime a.o g
[T ombi] 3GPP TSG RAN WG1#19, “T ombi – an ex ension o STTD o
mo e han wo an ennas”.
[DSTTD] Texas Ins umen s (2001) Double-STTD Scheme o HSDPA
Sys ems wi h Fou T ansmi An ennas: Link Le el Simula ion
Resul s. In TSG-RAN WG1 mee ing #20.
[Ho inen] A. Ho inen and O. Ti kkonen, ”Non-O hogonal Space-Time Block
Code wi h Symbol Ra e Two”, Con e ence on In o ma ion Science
and Sys ems, P ince on Uni e si y, Ma ch 20-22, 2002.
[Wang99] X. Wang and H. V. Poo . “I e a i e ( u bo) so in e e ence
cancella ion and decoding o coded CDMA”, IEEE T ansac ions on
communica ions, ol. 47:pp. 1046-1061, July 1999.
[Polla a96] S. Benede o, D. Di sala , G. Mo o si, and F. Polla a. “A so inpu
so ou pu maximum pos e io i module o decode pa allel and se ial
conca ena ed codes”, TDA P og ess Repo 42-127, No embe 1996.
[Kay93] S. M. Kay, “Fundamen als o S a is ical Signal P ocessing”, P en ice
Hall, 1993.
[Hassibi01] B. Hassibi and Be and M. Hochwald. “High a e codes ha a e
linea in space and ime”, Submi ed o IEEE T ans. In o. Theo y,
Augus 2000. Re ised Ap il 2001.
