SPACE-

SPACE-TIME CODING AND SIGNAL PROCESSING

FOR HIGH DATA RATE WIRELESS
COMMUNCATIONS

Ayman F. Naguib, Vahid Tarokh,

Tarokh, Nambi Seshadri,
Seshadri, and
A. R. Calderbank
AT&T Labs - Research
180 Park Avenue
Florham Park, NJ 07932
naguib@research.att.com

AT&T Labs-Research

Tutorial Overview
AT&T Labs

! PART 1:: Preliminaries.

! PART 2:: Space-
Space-Time Coding.
! PART 3:: Applications of Space-
Space-Time Coding.
! Summary.

1
 PART 1: Preliminaries
AT&T Labs

Wireless channels.
! Outage capacity of fading channels.
! Diversity Options: time, frequency, and space.
Receive and transmit diversity techniques.

Wireless Channels
AT&T Labs

Remote Dominant
Reflector

Local Scatterers
to Base

Co-Channel Mobile

Base Station Local Scatterers

to Mobile
Local Scattering Fading
Multipath Propagation Intersymbol Interference
Mobile Motion Time Varying Channel
Local Scatterers Cellular Spectrum Reuse Co-channel Interference
to Base
Remote Dominant
Reflector

2
 Signal Level in Wireless Channels
AT&T Labs

Short Term Fading

Mean Path Loss

Signal Level (dB)

Long Term Fading

Distance (dB)

! Slow fading (shadowing) is caused by large obstructions between

transmitter and receiver.
! Fast fading is due to scattering of the signal by object near transmitter.
! Path loss is proportional to 1/r, is between 2.5 and 5.

Classification of Wireless Channels

AT&T Labs

! The multipath channel:

h(t, , z) = i (t ) e j i ( t , i ,z ) (t i )
i
i

c
i (t, i , z) = 2 fc i (t ) fd ,it + ( z / ) sin i h z

! The three spreads:

spreads
Channel Channel Measure of
Spread Selectivity Selectivity
Delay Spread Frequency Selective Coherence Bandwidth
Doppler Spread Time Selective Coherence Time

Angle Spread Space Selective Coherence Distance

3
 Space-Time Fading
AT&T Labs

10log10 h(t,z)

Di
sta
nc
e(
in
w av
ele s
em
ng
th Tim
)

Angle Spread d = 0, Doppler Spread fd = 200 Hz

Space-Time Fading .
AT&T Labs
10log10 h(t,z)

Di
sta
nc
e(
in
w av s
ele em
ng
th Tim
)

Angle Spread d = 5, Doppler Spread fd = 200 Hz

4
 PART 1: Preliminaries
AT&T Labs

! Wireless channels.
Outage capacity of fading channels.
! Diversity Options: time, frequency, and space.
Receive and transmit diversity techniques.

Fundamental Limits: Outage Capacity

AT&T Labs

Tx 1 Rx 1
LM 11 12 ! 1 N OP
H =M PP
21 22 ! 2 N
Tx 2 Rx 2
MM " " # "
PQ
N M1 M 2 ! MN

Tx N Rx M

! Multiple input multiple output (MIMO

MIMO) channel: N transmitters, M receiver
! ij is the complex channel gain from i-th transmit antenna to j-th receive
antenna.
! H is the NM channel matrix.

5
 Fundamental Limits: Outage Capacity
AT&T Labs

! Channel capacity for SISO channel:

C = lo g 2 (1 + ) bits/sec/use, is the SNR

! Channel capacity for MIMO channel:

C = log2 det I + LM OP
n
HH*
N bits/sec/use
Q
H is the n m channel matrix
! Outage capacity C x :

z

Pr C > Cx l q= C ( H ) fH ( H ) d H = x
H :C ( ) = C x

Fundamental Limits: Outage Capacity

AT&T Labs
Achievable Data Rates with Multiple Antennas
Achievable Data Rate (kbits/sec) per 30 kHz Channel

225

200 We can be here

175 1 Tx 1 Rx Antennas
1 Tx 2 Rx Antennas
150 2 Tx 1 Rx Antennas
2 Tx 2 Rx Antennas
125

100

75

50
IS-136+

25
IS-136
We are here now
0
0 5 10 15 20
SNR per Rx Antenna (dB)

It is dangerous to put limit on wireless

Guglielmo Marconi (1932)

6
 PART 1: Preliminaries
AT&T Labs

! Wireless channels.
! Outage capacity of fading channels.
Diversity Options: time, frequency, and space.
Receive and transmit diversity techniques.

Diversity in Wireless Radio

AT&T Labs

! Receive the signal via independently fading channels.

! Improves the performance in a fading environment.
Space diversity: Multiple antennas are used to receive the signal.
Antenna spacing must be such that the fading at each antenna is
independent (coherence
coherence distance).
distance
Frequency diversity: Signal is transmitted in several frequency bands
(coherence
coherence BW).
BW Not effective on flat (non-frequency selective)
channels. Techniques that exploit frequency diversity include: RAKE
Receivers, OFDM, equalization.
Time diversity: Signal is transmitted in different time slots (coherence
coherence
time).
time Channel coding plus interleaving is used to provide time
diversity. Not effective over slow fading channels.
Polarization diversity: use two antennas with different polarization for
reception and/or transmission.

7
 Receive Diversity
AT&T Labs

! Make use of a number of receive antennas that are well

separated (>
> coherence distance)
distance to generate independent
receptions of the transmitted signal.
Selection diversity: choose received signal with largest
received power, S/N, etc.
Switched diversity: choose alternate antenna if signal falls
below a certain threshold.
Linear combining: linearly combine a weighted replica of all
received signals.
! There is a dramatic improvement even with two branch
selection diversity.
! The outage with MRC improves linearly with the number of
diversity branches M, but complexity becomes prohibitive.
prohibitive

Rx: Selection and Switched Diversity

AT&T Labs

Front End Data

1 Detector 1
Selection
Diversity Selector Data

Front End Data

2 Detector 2

Front End
1
Switched Switch
Diversity Data Data
Detector

Front End
2

8
 Receive Diversity: Linear Combining
AT&T Labs

1
r1 Front End r1 = A1e j1 d + n1
1 r2 = A2e j2 d + n2

Data
Data
Detector

r2 Front End
2

Equal gain: 1 = e j1 , 2 = e j2
Maximal ratio: 1 = A1e j1 , 2 = A2e j2
MMSE: training c h
1 , 2 = arg min 1r1 + 2 r2 d
1 , 2
2

decoding c
d = S r + *2 r2
*
1 1 h

Receive Diversity Performance

AT&T Labs

9
 Transmit Diversity
AT&T Labs

! Provide diversity benefit to a mobile using base station

antenna array for frequency division duplexing (FDD
FDD)
schemes. Cost is shared among different users.
! Order of diversity can be increased when used with other
conventional forms of diversity.
! Two kinds of transmit diversity techniques:
Transmit diversity with feedback from receiver
Transmit diversity without feedback from receiver:
No training.
Feedforward information.

Transmit Diversity with Feedback

AT&T Labs

Tx 1
w 2 (t )
Rx
r1 ( t ) r1 ( t )
S2 ( t ) r (t )
Tx 2 r (t )
Demod r2 ( t )
Mod
r2 ( t ) w1 ( t ), w 2 ( t )
variations
S1 ( t )
w1 ( t ) + w 2 ( t ) =1
2 2
w1 ( t )

! w1(t) and w2(t) are varied such that |r(t)|2 is maximized.

maximized
! w1(t) and w2(t) are adapted with feedback information from
the receiver.

10
 Tx Diversity with Frequency Weighting
AT&T Labs

Tx 1
r2 ( t )
Rx r1 ( t )
(A) Constructive Interference
Tx 2
Channel Channel
Encoder r2 ( t )
Decoder r1 ( t )
(B) Distructive Interference

( kT ) = 2 fm kT
e j ( t )

! Use frequency weighting to mitigate the harm of scenario B.

! Simulate fast fading can use conventional channel coding
and interleaving techniques.

Tx Diversity with Antenna Hopping

AT&T Labs
Tx 1

Tx 2 Rx

s Repetition s1 s2 ! s N ML
Code Detection
T

Tx N

! At time i, 1 i N , transmit s from antenna i.

! Achieves a diversity order of N using ML detection or MRC
at the receiver.
! Bandwidth efficiency is 1/N.

11
 Tx Diversity with Channel Coding
AT&T Labs
Tx 1

Rx
Tx 2

s Channel c1c 2 ! c N ML
Code Decoding
T

Tx N

! The channel code has a minimum Hamming distance dmin N.

! Transmit code symbol i from antenna i.
! After receiving the N symbols, the decoder performs ML
decoding to decode the received codeword.

Transmit Diversity via Delay Diversity

AT&T Labs

Tx 1

s(t )
Tx 2

Information Channel Delay

Encoder T s(t T )
Tx N

Delay
( N 1) T s ( t ( N 1) T )

! Provide diversity benefit by introducing intentional multipath.

! Receiver uses an equalizer or MLSE for detection.
! Provides a diversity order of N. No loss of BW efficiency.

12
 Transmit Diversity Options
AT&T Labs

p1 ( t )

a1 ( t ) e j 1 ( t )

Information
bits p2 ( t )

a 2 ( t ) e j6 2 ( t )
Channel Encoder

! Antenna hopping
! Frequency weighting pN (t )
! Delay diversity
! Space-
Space-Time Coding a N ( t ) e j N ( t )
Waveform Shaping

Part 2: Space-Time Coding

AT&T Labs

Space-
Space-Time coded modulation (STCM):
Probability of error analysis and design criterion.
Space-
Space-time trellis codes examples.
Decoding of space-
space-time trellis codes.
Comparison with delay-
delay-diversity.
! Space-
Space-Time block codes (STBC):
Basic idea and decoding of STBC.
Incoherent detection of STBC.
STBC for more than 2 transmit antennas.
! Interference suppression with STBC:
Zero forcing IC and ML decoding of STBC.
MMSE IC and ML decoding of STBC.
Iterative IC and ML decoding of STBC.

13
 Space-Time Coded Modulation
AT&T Labs

Space-Time Encoder
Information Receiver
Source

! For each input symbol, the space-time encoder chooses the

constellation points to simultaneously transmit from each
antenna so that coding and diversity gains are maximized.
maximized
Space-Time trellis codes: coding and diversity gain.
Space-Time block codes: diversity gain + some coding gain
(depending on the rate of the code).

Space-Time Coding: The Model

AT&T Labs

! N transmit and M receive antennas.

! The overall channel is made up of NM slowly varying sub-
channels.
! Each sub-channel is Rayleigh fading (same analysis applies
to Rician fading too).
! At any time interval, N signals are transmitted simultaneously,
simultaneously
one from each transmit antenna.
! The sub-channels undergo independent fading.
! The fade coefficients are assumed to be fixed during a slot
and independent from slot to another.

14
 STC: The Model .
AT&T Labs

! Transmitted Code Vector:

c l = [ c1 ( l ), c2 ( l ), ! , c N ( l ) ]T
! Channel Matrix: LM 11 12 ! 1 N OP
H =M PP
21 22 ! 2 N
MM " " # "
PQ
N M1 M 2 ! MN
! Received Signal Vector:

r (l ) = H c l + n (l )

STC: Probability of Error Analysis

AT&T Labs

! Transmitted code vector sequence: C = {c1,c2, , cL}.

! Probability of error: assuming perfect knowledge of CSI,

o ~
t o
~
Pr C C H exp d 2 (C, C ) t 2
M L
~
d (C, C ) = , j1 c1 (l ) - c~1 (l ) +!!+, jN c N (l ) - c~N (l )
2

j =1 l =1
M
~
= Gs h j A (C, C ) h *j , where
j =1

Gs = Es / 4 N o and h j = [ , j1 , j 2 ! , jN ]T
~ ~ ~
A (C, C ) = B(C, C )B* (C, C )

15
 STC: Probability of Error Analysis .
AT&T Labs

! The matrix B is the error matrix between the transmitted~

code vector sequence C and the decoded code vector C
sequence .

LM c (1) c~ (1)
1 1 c1 (2) c~1 (2 ) ! c1 ( L ) c~1 ( L ) OP
B= M c (1) c~ (1)
MM "
2 2 c2 (2) c~2 (2 )
"
! c2 ( L ) c~2 ( L )
# "
PP
Nc (1) c~ (1)
N N cN (2) c~N (2) ! cN ( L ) c~N ( L )
PQ

STC: Probability of Error Analysis .

AT&T Labs

! The matrix A=BB* is Hermitain A can be written as U

U* ,
where = diag{l1,l2,,lN} and U is orthonormal,
orthonormal U* U=I,
and its columns are the eigenvectors of A.

! Let j = U*h*j , then we will have

M M N
~
d 2 (C, C ) = s - *j L - j = s i | ij |2
j =1 j =1 i =1

The random variables ij = | ij | are i.i.d. , each has a 2

2
!
distribution with 2 degrees of freedom, that is

ij ~ fv ( ) = e v for > 0 and 0 otherwise

16
 STC: Probability of Error Analysis .
AT&T Labs

! Probability of error: average over all | ij |2

t FGH 1 + 1 IJK
M

o
N
~
Pr C C
i =1 i s

! Let r be the rank of the matrix A and l1,l2,,lr be the

nonzero eigenvalues of A. Then

t FGH IJK FGH 1 IJK

M rM

o
r
~
Pr C C i
i =1 s

! Thus a diversity gain of rM and a coding gain of (l1l2lr)1/r

are achieved.

STC: Design Criteria

AT&T Labs

Rank Criterion: In order to achieve the maximum diversity NM,

the matrix B(C1,C2) has to be full rank for any two code vector
sequences C1 and C2. If B(C1,C2) has a minimum rank r over the
set of two tuples of distinct code vector sequences, then a
diversity rM is achieved.
Determinant Criterion: The minimum of the r-th roots of the sum
of determinants of all rr principal cofactors of A(C1,C2) taken
over all pairs of distinct code vector sequences C1 and C2
corresponds to coding gain, r being the rank of A(C1,C2) . The
target of code design is making this sum as large as possible. If a
code is designed to give a diversity gain of NM, for a better
coding gain, the minimum of the determinant of A(C1,C2) taken
over all pairs of distinct code vector sequences C1 and C2 must be
maximized.

17
 STC: Design Criteria .
AT&T Labs

! Space-time trellis codes were designed using this criteria.

! The trade-off between diversity, rate, and trellis complexity
has also been studied.
! It is proved that our designs are optimal in terms of the
trade-off between complexity, constellation size, diversity,
and rate.
! It is not known whether our codes provide the optimum
coding gains at a given complexity, diversity, and rate.

STC: 4-PSK Example

AT&T Labs

0 00 01 02 03

1 10 11 12 13

2 20 21 22 23

Input: 0 1 2 3 2 2
Tx 1: 0 0 1 2 3 2
Tx 2: 0 1 2 3 2 2 3 30 31 32 33

4-PSK 4-State Space-Time Code with 2 Tx Antennas

18
 STC: 8-PSK Example 1
AT&T Labs

2
3 1
Input: 0 1 5 7 6 4
Tx 1: 0 0 5 1 3 6 4 0
Tx 2: 0 1 5 7 6 4
5 7
6
0 00,01,02,03,04,05,06,07
1 50,51,52,53,54,55,56,57
2 20,21,22,23,24,25,26,27
3 70,71,72,73,74,75,76,77
4 40,41,42,43,44,45,46,47
5 10,11,12,13,14,15,16,17
6 60,61,62,63,64,65,66,67
7 30,31,32,33,34,35,36,37

8-PSK 8-State Space-Time Code with 2 Tx Antennas

STC: 8-PSK Example 2

AT&T Labs
0 00 01 02 03 04 05 06 07
1 51 52 53 54 55 56 57 50
2 22 23 24 25 26 27 20 21
3 73 74 75 76 77 70 71 72
4 44 45 46 47 40 41 42 43
5 15 16 17 10 11 12 13 14
6 66 67 60 61 62 63 64 65
7 37 30 31 32 33 34 35 36
Input: 0 1 5 7 6 4 8 37 30 31 32 33 34 35 36
Tx 1: 0 0 5 4 6 3 9 00 01 02 03 04 05 06 07
Tx 2: 0 1 6 3 4 3 10 51 52 53 54 55 56 57 50
11 22 23 24 25 26 27 20 21
12 73 74 75 76 77 70 71 72
13 44 45 46 47 40 41 42 43
2 14 15 16 17 10 11 12 13 14
3 1 15 66 67 60 61 62 63 64 65
16 22 23 24 25 26 27 20 21
17 73 74 75 76 77 70 71 72
18 44 45 46 47 40 41 42 43
4 0 19 15 16 17 10 11 12 13 14
20 66 67 60 61 62 63 64 65
21 37 30 31 32 33 34 35 36
5 7 22 00 01 02 03 04 05 06 07
6 23 51 52 53 54 55 56 57 50
24 51 52 53 54 55 56 57 50
25 22 23 24 25 26 27 20 21
26 73 74 75 76 77 70 71 72
27 44 45 46 47 40 41 42 43
28 15 16 17 10 11 12 13 14
29 66 67 60 61 62 63 64 65
30 37 30 31 32 33 34 35 36
31 00 01 02 03 04 05 06 07

8-PSK 32-State Space-Time Code with 2 Tx Antennas

19
 STC: 16-QAM Example
AT&T Labs

0 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
1 B0 B1 B2 B3 B4 B5 B6 B7 B8 B9 BA BB BC BD BE BF

2 20 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F
Input: 0 1 A C 5 7
Tx 1: 0 0 B A C F 3 90 91 92 93 94 95 96 97 98 99 9A 9B 9C 9D 9E 9F
Tx 2: 0 1 A C 5 7
4 40 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F

5 F0 F1 F2 F3 F4 F5 F6 F7 F8 F9 FA FB FC FD FE FF

6 60 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F
0 1 2 3
7 D0 D1 D2 D3 D4 D5 D6 D7 D8 D9 DA DB DC DD DE DF

4 5 6 7 8 80 81 82 83 84 85 86 87 88 89 8A 8B 8C 8D 8E 8F

9 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F
8 9 A B A0 A1 A2 A3 A4 A5 A6 A7 A8 A9 AA AB AC AD AE AF
A
B 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F
C D E F
C C0 C1 C2 C3 C4 C5 C6 C7 C8 C9 CA CB CC CD CE CF

D 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F

E E0 E1 E2 E3 E4 E5 E6 E7 E8 E9 EA EB EC ED EE EF

F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F

16-QAM 16-State Space-Time Code with 2 Tx Antennas

STC: Maximum Likelihood Decoder

AT&T Labs

! Given the received vector sequence R = {r1,r2, , rL}, the

~
receiver chooses the code vector sequence C as the one
that was transmitted such that the a posteriori probability
~
o
Pr C R , H ( l ) , l = 1, ! , L t
! is maximized.
maximized
! The space-time maximum likelihood decoder (STMLD
STMLD):
L 2
~
C = arg m~in r ( l ) E s H ( l ) ~c l
C
l =1

! The STMLD is implemented as a vector Viterbi algorithm (v

v-
VA)
VA where the trellis path with the smallest accumulated
metric is chosen.

20
 STC: Performance with Perfect CSI
AT&T Labs

Performance of 4-PSK Space-Time Codes with 2 Transmit

and 1 Receive Antennas
100

Frame Length = 130 Symbols

Frame Error Rate

10-1

4-State Code
8-State Code
16-State Code
32-State Code
10-2 64-State Code

10 12 14 16 18 20
SNR (dB) per Receive Antenna

STC: Performance with Perfect CSI .

AT&T Labs

Performance of 8-PSK Space-Time Codes with 2 Transmit

and 1 Receive Antennas
100

Frame Length = 130 Symbols

Frame Error Rate

10-1

8-State Code
16-State Code
32-State Code
64-State Code

10-2
10 12 14 16 18 20
SNR (dB) per Receive Antenna

21
 STC: Performance with Perfect CSI .
AT&T Labs

Performance of 16-QAM Space-Time Codes with 2 Transmit

and 2 Receive Antennas
100

Frame Length = 130 Symbols

16-State Code
64-State Code
Frame Error Rate

10-1

10-2

10-3
10 12 14 16 18 20
SNR (dB) per Receive Antenna

Delay Diversity
AT&T Labs

Signal 2

Constellation
Mapper

Information Rate
Source Repetition Code One
Symbol Constellation
Delay Mapper

Signal 1

Delay Diversity

22
 Delay-Diversity as a ST Code
AT&T Labs
2
3 1
Input: 0 1 5 7 6 4
Tx 1: 0 0 1 5 7 6 4 0
Tx 2: 0 1 5 7 6 4
5 7
6
0 00,01,02,03,04,05,06,07
1 10,11,12,13,14,15,16,17
2 20,21,22,23,24,25,26,27
3 30,31,32,33,34,35,36,37
4 40,41,42,43,44,45,46,47
5 50,51,52,53,54,55,56,57
6 60,61,62,63,64,65,66,67
7 70,71,72,73,74,75,76,77

8-PSK 8-State Delay Diversity Code with 2 Tx Antennas

ST Code as a Delay-Diversity
AT&T Labs

Signal 2

Constellation
Mapper

Best Block Code

Information of Length 2
Source One
C Symbol Constellation
Mapper
Delay

Signal 1

Space-Time Coding as a Delay Diversity

! For the 8-PSK 8-state ST code:

k
C = 0 0 ,1 5 ,2 2 , 3 7 , 4 4 , 5 1, 6 6 , 7 3 p

23
 Part 2: Space-Time Coding
AT&T Labs

! Space-
Space-Time coded modulation (STCM):
Probability of error analysis and design criterion.
Space-
Space-time trellis codes examples.
Decoding of space-
space-time trellis codes.
Comparison with delay-
delay-diversity.
Space-
Space-Time block codes (STBC):
Basic idea and decoding of STBC.
Incoherent detection of STBC.
STBC for more than 2 transmit antennas.
! Interference suppression with STBC:
Zero forcing IC and ML decoding of STBC.
MMSE IC and ML decoding of STBC.
Iterative IC and ML decoding of STBC.

Space-
Space-Time Block Codes
AT&T Labs
Signal 2

Constellation
Mapper

ST Block Code
Information
c1 -c2*
Source c1 c 2
c2 c *1
Constellation
Mapper

Signal 1

Lc OP
! Idea:
c2
c1 c2 M 1

Nc 2 c1 Q
! Assumption: channel is quasi-
quasi-static.
static

24
 Decoding of STBC
AT&T Labs

! Received Signal:
r1 = h1c1 + h2 c2 + n1
r2 = h1c2* + h2 c1* + n2

r=
LM r OP = LM h
1 1 h2 c1
+
OPLM OP LM OP
n1
= Hc + n
Nr Q N h
2
* *
2 h1* c2 QN Q N Q
n2*
! H is orthogonal:

e
H*H = h1 + h2 I = h I
2 2
j 2

~r = H*r = h c + n
~
2

Decoding of STBC .
AT&T Labs

!
~ is still white c and c are detected
The noise term n 1 2
independently lower complexity.
complexity
! With M receive antennas:

~r =
M

H*i ri =
FG h IJ c + n~
M
2

i =1 H i =1K i

! With M receive antennas, a diversity order of 2M is

achieved.
! Only simple linear processing at the receiver is required.
! Complete CSI is required at the receiver. In practice channel
estimation is used to obtain CSI.

25
 Decoding of STBC .
AT&T Labs

H1 soft decision for c1

~ ML Decision
r1 r1 c$1
Linear Combiner
c1

~r = H* H r1 LM OP
1 2
r2 NQ ML Decision
r2
~
r2 c2
c$2

soft decision for c2

H2

STBC: Incoherent Detection

AT&T Labs

! Received Signal:
LM OP LM OPLM OP LM OP
~r = r1 = c1 c2 h1 + n1 = C h + n
NQ N
r2 c2* c1* h2 n2 QN Q N Q
! Given c1 and c2, h1 and h2 can be estimated as

~
~
LM OP
h1
h= ~ = 2
1
*
=
h1 + 1 LM OP
h2N Q
c1 + c2
2 C r
h2 + 2 N Q
c1*n1 + c2 n2 c2*n1 + c1n2
1 = 2 and 2 = 2
c1 + c2 c1 + c2
2 2

26
 STBC: Incoherent Detection .
AT&T Labs

! If the channel is quasi-

quasi-static,
static then the received signal for the
next code symbol is:
r3 = h1c3 + h2 c4 + n3
r4 = h1c4* + h2c3* + n4
~
! Use the channel estimate h for detecting c3 and c4.
! Need to transmit two known symbols at the beginning of
each time slot.
! If the channel variation is very slow, then at each time we
average the channel estimate less noisy estimates.
! This scheme will perform within 3 dB of the case when ideal
CSI is available at the receiver.

Space-
Space-Time Block Codes for N > 2
AT&T Labs

! Is there a rate 1 STBC with simple linear processing for

more than two antennas (N N >2) ?
! Answer: theory of generalized orthogonal designs.
Complex constellation: NO.
Real constellation: YES.
! Rate 1 codes with simple linear processing for arbitrary
number of transmit antennas and real constellation.
! Rate 1/2 codes with simple linear processing for arbitrary
number of transmit antennas and complex constellation.
! Rate 3/4 codes with simple linear processing for N=3 3 and
N=4
4 transmit antennas and complex constellation.

27
 Space-
Space-Time Block Codes for N > 2 .
AT&T Labs

! Example 1: Complex constellation, rate=1/2, 4 transmit

antennas (N=4)

LMc OP
1 LM c 1 c2 c3 c4 c1* c2* c3* c4* OP
MMcc PP
2
MMcc2 c1 c4 c3 c2* c1* c4* c3* PP
c4 c1 c2 c3* c4* c1* c2*
MNc PQ
3

4
MNc4
3

c3 c2 c1 c4* c3* c2* c1*

PQ
! Will provide 4 branch diversity performance + 3 dB coding
gain.

Space-Time Block Codes for N > 2 .

AT&T Labs

! Example 2: Complex constellation, rate=3/4, 4 transmit

antennas (N=4)

LM c c2*
c3* c3* OP
MM PP
1
2 2
LMc OP
1
MM c
2 c1*
c *
3
3
c *

PP
MMc PP
2
MM c2 c3 *
2
c c c2 c2 c2 c2 c1 c1*
+ + *
+ *
+
2
PP
Nc Q
3
3
2
1 1
2 2
MM c
c3

c2 + c2* c1 + c1*

c1 + c1* + c2 c2* PP
N2 Q
3
2 2 2

! Will provide 4 branch diversity performance + 1.25 dB

coding gain.

28
 Space-
Space-Time Block Codes for N > 2 .
AT&T Labs

! Example 3: Real constellation, rate=1, 4 transmit antennas

(N=4)

LMc OP
1 LM c1 c2 c3 c4 OP
MMcc PP
2
MMcc
2 c1 c4 c3
c4 c1 c2
PP
MNc PQ
3

4
MNc
3

4 c3 c2 c PQ
1

! Will provide 4 branch diversity performance and no coding

gain.

Part 2: Space-Time Coding

AT&T Labs

! Space-
Space-Time coded modulation (STCM):
Probability of error analysis and design criterion.
Space-
Space-time trellis codes examples.
Decoding of space-
space-time trellis codes.
Comparison with delay-
delay-diversity.
! Space-
Space-Time block codes (STBC):
Basic idea and decoding of STBC.
Incoherent detection of STBC.
STBC for more than 2 transmit antennas.
Interference suppression with STBC:
Zero forcing IC and ML decoding of STBC.
MMSE IC and ML decoding of STBC.
Iterative IC and ML decoding of STBC.

29
 Interference Cancellation with STBC
AT&T Labs
Burst 2

Information Space-Time
Source Block Encoder

Interference Cancellation Terminal 1

Burst 1 Information
and
Terminal 1 ML Decision

Burst 2 c1

Terminal 2
c2 Information
Information Space-Time
Source Block Encoder

Burst 1

Terminal 2

! K users, N transmit antennas per user. Classical IC techniques need N(K-1)

+1 receive antennas to suppress interference from K-1 co-channel users.
Exploit code structure to suppress interference only K receive antennas
are required. Assumption: full synchronization between terminals.
Can be implemented using adaptive LMS or RLS.
Can be used for increasing system capacity or increasing data rate.

Zero Forcing IC with STBC

AT&T Labs

! Received Signal:

r1 = H1 c + G1 s + n1

LMr OP = LMH
1 1 G1 c
+
OP LM OP LM OP
n1
r2 = H 2 c + G 2 s + n2 Nr Q NH
2 2 G2 s n2QNQ N Q
r = H ~c +
! Zero forcing solution:

LM H 1 G1 OP LM r OP = LM r$ OP = LMcOP + LM n$ OP
-1
1 1 1

NH 2 G2 Q Nr Q Nr$ Q N s Q Nn$ Q
2 2 2

30
 Zero Forcing IC with STBC .
AT&T Labs

LMH G1 OP = LMH~
1 1
OP L I G1G 21 OP
Q MNH H
1 0
NH 2 G2 Q N0 ~ 1
G 2
1
1 I Q
~ ~
H = H1 G1G 21H 2 , G = G 2 H 2 H11G1

LM I G1G 21 OP Lr O = L ~r O = LMH~ ~0 OP LcO + Ln~ O

Q MNr PQ MN~r PQ N 0 GQ MNs PQ MNn~ PQ
1 1 1

N H H2
1
1 I 2 2 2

~ ~
& H and G are orthogonal and have the same structure and
n1 and ~
noise ~ n2 are still white can detect c and s as before..

Zero Forcing Interference Cancellation .

AT&T Labs

! ZF Interference Cancellation:

b
( c, c ) = ZFIC r1 , r2 , H1 , H 2 , G1 , G 2 g
{
~r = r G G 1r
1 1 2 2
~
H = H1 G1G 21H 2
~ 2
c$ = arg min ~r H c
c C

~ 2
c = ~r H c$
}

31
 ZFIC Performance
AT&T Labs

FER Performance of 8-PSK with STBC and Zero Forcing

Interference Cancellation
100

10-1
Frame Error Rate

10-2

10-3

10-4
STBC( 2 Tx, 2 Rx) + ZFIC, SIR = 0 dB
STBC( 2 Tx, 1 Rx)
STBC( 2 Tx, 2 Rx) , No Interference
10-5

10-6
10 15 20 25 30

SNR per Rx Antenna (dB)

MMSE Interference Cancellation

AT&T Labs

! Error Criteria:
2
D(w, - 1 , - 2 ) = w*r - 1 c1 - - 2 c2

! MMSE IC: find a linear combiner w, 1, and 2 such that

E{(w, 1, 2)}is minimized.
& 1 or 2 should be set equal to 1 or else we will get a zero
solution.

- 1 = 1 D1 (w1 , - 2 ) = w1*r c1 - - 2c2

2 ~* ~r c
= w
2
1 1
2 ~* ~r c
- 2 = 1 D 2 (w2 , - 1 ) = w2*r c2 - - 1c1 = w
2
2 2

32
 MMSE Interference Cancellation .
AT&T Labs

! MMSE IC solution:

LM
~ = M h2 OP LMh OP , w~ = LMM h OP LMh OP
-1
1 1
-1
2
w1
N
h*2 1 Q N0Q Nh 1 Q N 0 Q
2 *
1

M = HH* + 1
SNR
I
h1 = first column of H
h2 = second column of H

MMSE Interference Cancellation .

AT&T Labs

! MMSE solution:

h*2 M -1h1
(
w1 = M - h 2 h 2)
* -1
h1 , - = *

1 - h*2 M -1h1
2

h1*M -1h 2
(
w2 = M - h h * -1
1 1 ) h2 , - = *

1 - h1*M -1h 2
1

! Using the structure of M and the fact that h1 and h2 are

orthogonal we can show that either 1=0 and 2 =1 or 1=1
orthogonal,
and 2 =0.
Two different combiners w1 and w2 for c1 and c2, respectively.

33
 MMSE Interference Cancellation .
AT&T Labs

b
( c , c ) = M M SE_ IC r1 , r2 , H 1 , H 2 , G 1 , G 2 , g
{
~r = r T r T T
1 1

~
H=
LM
H1 G1 OP
H2 G2N Q
1
M = HH* + I

T T
h 1 = h 11
T
h T21 , h 2 = h 12
T
h T22
1* = M 1h 1 , *2 = M 1h 2
c1 = arg min 1* ~r c$1 , c2 = arg min 1* ~r c$2
2 2

c$1 C c$2 C

c = 1* ~r c$1 + 1* ~r c$2
2 2

MMSE IC Performance
AT&T Labs

FER Performance of 8-PSK with STBC and MMSE

Interference Cancellation
100

10-1
Frame Error Rate

10-2

10-3

10-4
STBC( 2 Tx, 1 Rx)
STBC( 2 Tx, 1 Rx) + MMSE IC, SIR = 0 dB
STBC( 2 Tx, 1 Rx) + MMSE IC, SIR = 10 dB
10-5
STBC( 2 Tx, 1 Rx) + MMSE IC, SIR = 20 dB
STBC( 2 Tx, 2 Rx), No Interference

10-6
10 15 20 25 30

SNR per Rx Antenna (dB)

34
 Iterative Interference Cancellation
AT&T Labs

b
( c$ , s$ ) = IT _ IC_ STDEC r1 , r2 , H1 , H 2 , G 1 , G 2 g
{
b
( c$ 0 , c ,o ) = IC_ STDEC r1 , r2 , H 1 , H 2 , G1 , G 2 g
x 1 = r1 H 1 c$ o , x 2 = r2 H 2 c$ o
s$ o = arg min
s S
ex 1 G1 s + x 2 G 2 s
2 2
j
s ,o = x 1 G 1 s + x 2 G 2 s
2 2

b
( s$ 1 , s ,1 ) = IC_ STDEC r1 , r2 , G 1 , G 2 , H1 , H 2 g
y1 = r1 G 1 s$ 1 , y$ 2 = r2 G 2 s$ 1
c$ 1 = arg min
c C
e y H c
1 1
2
+ y2 H 2 c
2
j
c ,1 = y1 H 1 c + y 2 H 2 c
2 2

If ( c ,o + s ,o ) < ( c ,1 + s ,1 ) ( c$ , s$ ) = ( c$ o , s$ o )
Else ( c$ , s$ ) = ( c$ 1 , s$ 1 )
}

Iterative MMSE IC Performance

AT&T Labs

FER Performance of 8-PSK with STBC and MMSE

Iterative Interference Cancellation
100

10-1
Frame Error Rate

10-2

10-3

10-4

STBC( 2 Tx, 1 Rx)

10-5 STBC( 2 Tx, 2 Rx) + Iterative MMSE IC, SIR =0 dB
STBC( 2 Tx, 2 Rx) , No Interference

10-6
10 15 20 25 30

SNR per Rx Antenna (dB)

35
 Iterative MMSE IC Performance .
AT&T Labs

FER Performance of 8-PSK with STBC and MMSE

Iterative Interference Cancellation
100

STBC( 2 Tx, 2 Rx) + MMSE IC, SNR = 20 dB

STBC( 2 Tx, 2 Rx) + ITMMSE IC, SNR = 20 dB
Frame Error Rate

10-1

10-2

10-3
-30 -20 -10 0 10 20 30

SIR (dB)

Part 3: Space-Time Coding Applications

AT&T Labs

Narrowband TDMA Cellular.

! Broadband Wireless OFDM.
! Applications of STBC with Interference Suppression.
Suppression

36
 Base-Station Transmitter
AT&T Labs

Burst 2
Concatenated Space-Time Encoder
Pulse
Burst Shaper
Interleaving
ST Encoder Building
Information Block Encoder
Source (Reed Solomon)
Pulse
Burst Shaper
Interleaving
Building

Burst 1

! High rate outer block code will clean up burst errors of the ST code and
will provide an error detection mechanism.
! Same interleaving for both bursts.
! Synchronization and pilot sequences are inserted in both bursts for
timing and frequency synchronization as well as channel estimation at
the receiver.

Slot Structure
AT&T Labs

Lb
Burst 2
Training Training
Pilot Pilot Pilot
Sequence Data Data Data Sequence
S2 P2 P2 P2 S2

Lt Lp Ls

Training Training
Sequence Pilot Pilot Pilot Sequence
Data Data Data
S1 P1 P1 P1 S1
Burst 1
Lb

! Training sequences S1 and S2 as well as pilot sequences P1 and P2 are

orthogonal sequences. In addition these sequences are constant energy
sequences, i.e. from M-PSK constellation (this will give best MSE
performance).

37
 Mobile Receiver
AT&T Labs

Channel Estimation
and
Matched Timing Interpolation Deinterleaving
Filter Synchronization

Space-Time
Vector Viterbi
Decoder Reed Solomon
Decoder

Matched Timing
Filter Synchronization Concatenated Space-Time Decoder

Channel Estimation Deinterleaving

and
Interpolation

! Design assumes flat fading channel.

! Frequency offset is compensated for as part of the channel estimation.
! Channel estimates at the pilot and training symbol positions are
interpolated to obtain CSI for the whole slot.

Channel Estimation
AT&T Labs

! Complex channel gain:

~ ( t ) = ( t ) e j 2 fo t , f is the frequency offset
ij ij o

~
m ~
ij r
R ( ) = E ij ( t ) ( t + ) = J ( 2 f ) e j 2 fo
o d

! Pilots and synch sequences should be positioned in the slot

such that the channel sampling frequency fs satisfies 2(fd +fo)
fs = LsTs.
! Channel Estimation:
~ (l ) + z (l )
Y j ( l ) = A P1 P2 ! P N a j = 1, 2 , ! , M
j j

Pi* Y j ( l )
$ ij ( l ) = i = 1, 2 , ! , N
A Pi
2

38
 Channel Interpolation
AT&T Labs

! Channel estimates are interpolated to obtain complete CSI

for the whole time slot.
! Wiener interpolation filter (WIF):
Optimum interpolation filter needs the knowledge of fd , fo, and
SNR.
SNR
! A different filter for each symbol in the burst.
! Poor performance at low Doppler with worst case design.
design
! Optimum low-
low-pass time-
time-invariant interpolation filter:
! Similar to WIF, except that we use same filter for all symbols
in the burst.
! Poor CSI near the ends of the slot when only pilot symbols in
that slot are used.
! Lower complexity than WIF, performance is, however, poorer.
poorer

Channel Interpolation ...

AT&T Labs

Matched Channel Estimation

Filter Channel Estimates
and
Interpolation

Select Interpolation
Interpolation Filter Estimate
Filters Doppler Range

Matched
Channel Estimation
Filter
and
Interpolation Channel Estimates

Quasi-adaptive interpolation

! Doppler range is divided into different regions and pre-designed WIF is

used for each range.
! Receiver selects the filter use based on the quality of the interpolated
CSI for the previous slot

39
 Modem Simulation
AT&T Labs

! STCM with 16-QAM and 8-PSK.

! Timing and carrier frequency offset synchronization.
! fc = 1.9 GHz, IS-136 basic channelization and slot structure.
! /4-DQPSK modulation for synchronization and pilot symbols.
! Oversampling factor of 8.
! SR raised cosine pulse shaping with 0.35% excess BW.
! FER Performance:
! FER vs. SNR and Doppler Spread.
! Space-Time code vs. Delay-Diversity code.
! Effect of interpolation filter and transmit antenna correlation.
! Effect of delay spread.

Slot Structure
AT&T Labs

Burst 2
6.67 ms=162 Symbols

S2 O2 D2 P2 D2 P2 D2 P2 D2 P2 D2 P2 D2 P2 D2 O2 S2
14 5 9 2 21 2 21 2 21 2 21 2 21 2 12 5 14

S1 O1 D1 P1 D1 P1 D1 P1 D1 P1 D1 P1 D1 P1 D1 O1 S1
14 5 9 2 21 2 21 2 21 2 21 2 21 2 12 5 14

6.67 ms=162 Symbols

Burst 1

S1,S2 : Synchronization Sequence (S1 and S2 are orthogonal, S1 as in IS-136)

P1,P2 : Pilot Symbols (P1 and P2 are orthogonal)
D1,D2 : Data
O1,O2 : Overhead Symbols

40
 Error Distribution
AT&T Labs

Error Histogram for the 16-QAM 16-State ST Code with

2 Tx 2 Rx Antennas and Optimized WIF @ fd = 180 Hz
1.0
Fraction of Total Number of Byte Errors per Frame

SNR = 12.5 dB
0.8 SNR = 15.0 dB
SNR = 17.5 dB
SNR = 20.0 dB

0.6

0.4

0.2

0.0
1-2 3-4 5-6 7-8 9-10 11-12 >12
Number of Byte Errors per Frame

Frame Error Rate Performance

AT&T Labs

16-QAM 16-State ST Code with 2 Tx and 2 Rx Antennas @ fd = 180 Hz

with Optimized WIF
100
PF, Frame Error Rate (FER)

10-1

10-2

ST code only
10-3 ST code + RS1
ST code + RS3
ST code + RS5

10-4
14 16 18 20 22 24
SNR (per Rx antenna) dB

41
 Frame Error Rate Performance .
AT&T Labs

8-PSK 32-State ST Code with 2 Tx and 2 Rx Antennas @ fd = 180 Hz

and Optimized WIF
100

PF, Frame Error Rate (FER)

10-1

10-2

ST Code Only
10-3 ST Code + RS1
ST Code + RS3
ST Code + RS5

10-4
8 9 10 11 12 13 14 15
SNR (per Rx antenna) dB

Frame Error Rate Performance .

AT&T Labs

16-QAM 16-State Space-Time Code with 2 Tx and 2 Rx Antennas

17

ST Code + RS5, LPIF

ST Code + RS5, Optimized WIF
SNR for 0.1 Frame Error Rate (dB)

16

15

14

13

12
20 40 60 80 100 120 140 160 180
Maximum Doppler Frequency fd (Hz)

42
 Frame Error Rate Performance .
AT&T Labs

8-PSK 32-State Space-Time Code with 2 Tx and 1 Rx Antennas

20
ST code + RS5, LPIF
ST code + RS5, Optimized WIF
SNR for 0.1 Frame Error Rate (dB)

19

18

17

16
20 40 60 80 100 120 140 160 180
Maximum Doppler Frequency fd (Hz)

Frame Error Rate Performance .

AT&T Labs

8-PSK 32-State Space-Time Code with 2 Tx and 2 Rx Antennas

13

ST code + RS5 and LPIF

ST code + RS5 and Optimized WIF
Es/No for 0.1 Frame Error Rate

12

11

10

9
20 40 60 80 100 120 140 160 180
Doppler Spread fd (Hz)

43
 Frame Error Rate Performance .
AT&T Labs

8-PSK 32-State ST Code vs. Delay Diversity with 2 Tx

and 1 Rx Antennas
30

SNR for 0.1 Frame Error Rate (dB) DD Code + RS5, LPIF
28 ST Code + RS5, LPIF
DD Code + RS5, Optimized WIF
ST Code + RS5, Optimized WIF
26

24

22

20

18

16
20 40 60 80 100 120 140 160 180
Maximum Doppler Frequency fd (Hz)

Frame Error Rate Performance .

AT&T Labs

8-PSK 32-State ST Code vs. Delay Diversity with 2 Tx

and 2 Rx Antennas
20

19
DD Code + RS5, LPIF
SNR for 0.1 Frame Error Rate (dB)

ST Code + RS5, LPIF

18
DD Code + RS5, Optimized WIF
ST Code + RS5, Optimized WIF
17

16

15

14

13

12

11

10

9
20 40 60 80 100 120 140 160 180
Maximum Doppler Frequency fd (Hz)

44
 Frame Error Rate Performance .
AT&T Labs

8-PSK 32-State Space-Time Code with Optimized WIF

20

18
SNR for 0.1 Frame Error Rate

16

2 Tx 1 Rx, = 0
14 2 Tx 1 Rx, = 0.75
2 Tx 2 Rx, = 0
2 Tx 2 Rx, = 0.75
12

10

8
20 40 60 80 100 120 140 160 180
Maximum Doppler Frequency fd (Hz)

Delay Spread Performance

AT&T Labs

8-PSK 32-State ST Code with 2 Tx and 2 Rx Antennas and Optimized

WIF @ fd = 180 Hz and 5 s delay spread (TU model)
100
PF , Frame Error Probability

10-1

10-2

10-3 ST code, 0 s delay spread

ST code, 5 s delay spread
ST code+RS5, 0 s delay spread
ST code+RS5, 5 s delay spread
10-4
8 9 10 11 12 13 14 15
SNR (per Rx Antenna) dB

45
 Delay Spread Performance .
AT&T Labs

8-PSK 32-State Space-Time Code with Optimized WIF

@ fd = 180 Hz and 5 s delay spread (TU model)
18

SNR for 0.1 Frame Error Rate

16

2 Tx 1 Rx, 0 s delay spread

14 2 Tx 1 Rx, 5 s delay spread
2 Tx 2 Rx, 0 s delay spread
2 Tx 2 Rx, 5 s delay spread

12

10

8
20 40 60 80 100 120 140 160 180
Maximum Doppler Frequency fd (Hz)

Performance Summary
AT&T Labs

DIVERSITY MODULATION DOPPLER SNR FOR SNR FOR SNR FOR

& CODING SPREAD 55.8 KBPS 50.4 KBPS 46.8 KBPS
(ST) (ST+RS3) (ST+RS5)
2Tx 1Rx 8PSK/CST 10 Hz 18.3 dB 17.3 dB 16.9 dB

2Tx 2Rx 8PSK/CST 10 Hz 11.2 dB 10.3 dB 10.1 dB

2Tx 1Rx 8PSK/CST 180 Hz 20.1 dB 17.6 dB 17.1 dB

2Tx 2Rx 8PSK/CST 180 Hz 11.8 dB 10.6 dB 10.2 dB

DIVERSITY MODULATION DOPPLER SNR FOR SNR FOR SNR FOR

& CODING SPREAD 74.4 KBPS 67.2 KBPS 62.4 KBPS
(ST) (ST+RS3) (ST+RS5)
2Tx 1Rx 16-QAM/CST 10 Hz 22.4 dB 21.1 dB 20.5 dB

2T x 2Rx 16-QAM/CST 10 Hz 15.7 dB 13.8 dB 13.2 dB

2Tx 1Rx 16-QAM/CST 180 Hz 30.6 dB 25 .0 dB 23.0 dB

2T x 2Rx 16-QAM/CST 180 Hz 18.1 dB 15.3 dB 14.3 dB

46
 Part 3: Space-Time Coding Applications
AT&T Labs

! Narrowband TDMA Cellular.

Broadband Wireless OFDM.
! Applications of STBC with Interference Suppression.
Suppression

Broadband Wireless OFDM

AT&T Labs

Burst 2
Space-Time Coded OFDM Tx

ST Encoder IFFT RF
Information Block Encoder
Source (Reed Solomon)

IFFT RF

Burst 1

Space-Time
RF IFFT
Vector Viterbi
Decoder Reed Solomon
Decoder

RF IFFT

Space-Time Coded OFDM Rx

47
 Performance of ST-
ST-Coded OFDM
AT&T Labs

! 1 MHz bandwidth. Frame Error Probability for Concatenated Space-Time Coded OFDM
with 4-PSK 16-State ST Code and 2 Tx and 2 Rx Antennas
! 256 tones.
100
! QPSK 16-state ST code
with 2 transmit antennas.
0 s Delay Spread
! 2-Ray fading channel 5 s Delay Spread
model with equal 40 s Delay Spread

Frame Error Probability

strength.
! Ideal CSI.
CSI
1.5 Mbps data rate (1.5 10-1
bits/sec/Hz)
For 10% FER,
FER an Eb/No
between 2.7-
2.7-4 dB is
required (depending on
delay spread).
With 16-QAM, 4 Tx and
10-2
2 Rx antennas, 6 bits/sec 5 6 7 8 9 10
/Hz is achievable. Can be SNR per Rx Antenna (dB)
used for downlink.

Part 3: Space-Time Coding Applications

AT&T Labs

! Narrowband TDMA Cellular.

! Broadband Wireless OFDM.
Applications of STBC with Interference Suppression.

48
 Increase Capacity
AT&T Labs

Space-Time Modulator
Channel
Encoder Block Code

Concatenated Encoder Modulator

Space-Time Processor: Soft

ST Decoder, Interference Channel
Decision
Suppression, and MLSE Decoder
Generator

Concatenated Decoder

! Outer FEC code provides immunity against channel errors.

errors
Inner STBC provides interference suppression and fading
mitigation through transmit diversity.

Performance
AT&T Labs

FER Performance of 8-PSK with FEC + STBC and MMSE

Interference Cancellation at 0 dB SIR
100

10-1
Frame Error Rate

10-2

10-3 STBC Only

STBC + RS(57,63) over GF(64)
STBC + RS(53,63) over GF(64)

10-4
10 15 20 25 30

SNR per Rx Antenna (dB)

49
 Increase Data Rate
AT&T Labs

Channel Space-Time Modulator

Encoder Block Code
R1

Serial Concatenated Encoder Modulator

to
Parallel

Channel Space-Time Modulator

Encoder Block Code
R2

Concatenated Encoder Modulator

Throughput: b g
7 = 0.5 R1 1 - FER1 + R2 1 - FER2 b g

Performance
AT&T Labs

Throughput Performance of 8-PSK with STBC and Unequal

Error Protection with RS(53,63)

1.0

0.9

0.8

0.7
Throughput

0.6

0.5

0.4

0.3
R1= 1 and R2 =1
0.2 R1 = 53/63 and R2 = 1

0.1

0.0
10 15 20 25

SNR per Rx Antenna (dB)

50
 Summary
AT&T Labs

! Space-Time coding: good idea with great potential.

! Current work in space-time coding:
Systematic design of space-time trellis codes.
Multi-channel equalization for space-time coding.
Applications of space-time coding to WCDMA.

51