Base Station Cooperation

School of ECS, Univ. of Southampton, UK. http://www-mobile.ecs.soton.ac.
uk [1] - [10] 1/ 137 [

Mobile-station and Base-station Cooperation
Presented by
Lajos Hanzo
School of Electronics and Computer Science,
University of Southampton, SO17 1BJ, UK.
http://www-mobile.ecs.soton.ac.uk
School of ECS, Univ. of Southampton, UK. http://www-mobile.ecs.soton.ac.uk [1] - [10] 2/ 137 [
Acknowledgements
Sincere thanks are due to the team back at base in Southampton, in particular to
Lie-Liang Yang, Soon-Xin Ng, Wei Fang, Nan Wu, Rong Zhang, Lingkun Kong, Li
Wang, Fasih Butt, Mohammed El-Hajjar, KyungChun Lee, SeungHwan Won, Shinya
Sugiura.
The Sponsors: EPSRC, the EU, the VCE
Background Reading:
L. Hanzo, O. Alamri, M. El-Hajjar, N. Wu: Near-Capacity
Multi-Functional MIMO Systems: Sphere-Packing, Iterative
Detection and Cooperation, IEEE Press - John Wiley, 2009
1990
2000
1980
1970
1960
1950
Shannon limit (1948)
Elias, Convolutional codes
Viterbi algorithm
Bahl, MAP algorithm
Hamming codes
PGZ algorithm
LDPC Codes
Wolf, trellis block codes
Chase algorithm
Reed Solomon codes
BCH codes
algorithm
Berlekamp-Massey
RRNS codes
Ungerboeck, TCM
Berrou, turbo codes
Robertson, Log-MAP algorithm
Koch, Max-Log-MAP algorithm
Hagenauer, SOVA algorithm
Nickl, turbo Hamming code
Hagenauer, turbo BCH code
Alamouti, space-time
block code
Tarokh, space-time trellis code
Robertson, TTCM
Pyndiah, SISO Chase
algorithm
Convolutional Codes
Block Codes
Acikel, punctured turbo code
Figure 1: A glimpse of post-Shannon history
0 5 10 15 20 25 30 35
E
b
/ N
o
(dB)
10
0
2
3
4
5
6
7
8
9
10
1
N
o
r
m
a
l
i
s
e
d
C
a
p
a
c
i
t
y
(
B
i
t
s
/
s
/
H
z
)
BPSK
4QAM
16QAM
64QAM
Shannon Limit
Fixed - mean BER 0.01%
Fixed - mean BER 1%
AQAM - mean BER 0.01%
AQAM - mean BER 1%
TU Channel
Figure 2: Channel capacity upper bound of HSDPA-style AQAM and xed modulation
schemes over the COST 207 TU Rayleigh Fading channel for BER=1% and BER=0.01%.
Averting the WWW: MIMOs, Cooperative
Communications and Transmit Preprocessing
Capacity of MIMOs
C min(M; N)
The classic Shannon-Hartley law suggests that the achievable channel capacity
increases logarithmically with the transmit power.
By contrast, the MIMO capacity increases linearly with the number of transmit
antennas, provided that the number of receive antennas is equal to the number of
transmit antennas. With the further proviso that the total transmit power is increased
proportionately to the number of transmit antennas, a linear capacity increase is
achieved upon increasing the transmit power, which justies the spectacular success
of MIMOs...
L. Hanzo, O. Alamri, M. El-Hajjar, N. Wu: Near-Capacity Multi-Functional MIMO
Systems; John Wiley and IEEE PRESS, 2009
Capacity of MIMO Systems
Discrete-Input Continuous-Output Memoryless Channel (DCMC)
J The Full-multiplexing-gain sys-
tem has a higher asymptotic ca-
pacity as a benet of its multi-
plexing gain.
J The gap between the capacity
curves of the Rayleigh fading
channel and the AWGN channel
is narrower for the full-diversity
system as a benet of its diver-
sity gain.
ray-capacity-mimo-16qam-3.gle
-10 0 10 20 30
SNR (dB)
0
1
2
3
4
5
6
7
8
9
10
11
C
(
b
i
t
/
s
y
m
b
o
l
)
.......
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.....................
..............
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
................
......
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..............................
.............
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.........................
N
t
= N
r
= 2
.
Multiplexing
Rayleigh
AWGN
16-QAM, D=4
16-QAM, D=2
16-QAM, D=4
16-QAM, D=2
Diversity
The capacity of D = 2 and 4 dimensional 16QAM-based
MIMO DCMC uncorrelated Rayleigh-fading channel and
AWGN channel.
Evolution of the Four MIMOs in Wireless
Communications
Type I: Beamforming
Type II - SDMA: Spatial Division Multiple Access
Type III - SDM: Spatial Division Multiplexing
Type IV - Space-Time Coding: STTC, STBC and STS
Type I MIMOs: Beamforming
Base Station
Mobile Stations
Multiple Antenna Aided Spatial Division Multiple Access
OFDM
Type II MIMOs: SDMA
0 16 32 48 64 80 96 112 128
Symbol Index
0.0
0.2
0.4
0.6
0.8
1.0
A
m
p
l
i
t
u
d
e
(a) CIR 1: user 1, antenna 1
0 16 32 48 64 80 96 112 128
Symbol Index
0.0
0.2
0.4
0.6
0.8
1.0
A
m
p
l
i
t
u
d
e
(b) CIR 2: user 1, antenna 2
0 16 32 48 64 80 96 112 128
Symbol Index
0.0
0.2
0.4
0.6
0.8
1.0
A
m
p
l
i
t
u
d
e
(c) CIR 3: user 2, antenna 1
0 16 32 48 64 80 96 112 128
Symbol Index
0.0
0.2
0.4
0.6
0.8
1.0
A
m
p
l
i
t
u
d
e
(d) CIR 4: user 2, antenna 2
Figure 3: Evolution from CDMA to SDMA: Four different channel impulse responses
(CIR) recorded at the two receiver antennas for the two users supported.
0 16 32 48 64 80 96 112 128
Subcarriers
0.0
0.5
1.0
1.5
2.0
2.5
M
a
g
n
i
t
u
d
e
(a) CTF 1: user 1, antenna 1
0 16 32 48 64 80 96 112 128
Subcarriers
0.0
0.5
1.0
1.5
2.0
2.5
M
a
g
n
i
t
u
d
e
(b) CTF 2: user 1, antenna 2
0 16 32 48 64 80 96 112 128
Subcarriers
0.0
0.5
1.0
1.5
2.0
2.5
M
a
g
n
i
t
u
d
e
(c) CTF 3: user 2, antenna 1
0 16 32 48 64 80 96 112 128
Subcarriers
0.0
0.5
1.0
1.5
2.0
2.5
M
a
g
n
i
t
u
d
e
(d) CTF 4: user 2, antenna 2
Figure 4: Channel transfer functions (CTF) for the CIRs seen in Figure 3 (a) CTF 1,
(b) CTF 2, (c) CTF 3, and (d) CTF 4.
Type III MIMOs: SDM ML Sphere-Detector for Rank-Decient Scenarios
10
-6
10
-5
10
-4
10
-3
10
-2
10
-1
10
0
0 5 10 15 20
B
E
R
E
b
/N
0
[dB]
sdm-cmp-16qam : 27-Jun-2005
SOPHIE
MMSE
m
t
=6,n
r
=
16QAM 4
64QAM 6
The proposed SOPHIE SDM detection method exhibits near-optimum performance and relatively low
complexity in high-throughput scenarious such as 6x6 64QAM as well as rank-decient 6x4 16QAM
MIMO-OFDM. The throughput is 36 and 24 bits/symbol, respectively.
Channel Variation in Space-Time Coded OFDM:
Type IV MIMO
1 Tx 1 Rx
0
1
2
8
2
5
6
3
8
4
5
1
2
S
u
b
c
a
rrie
r
in
d
e
x
0
25
50
75
100
125
150
Transm
ission
fram
e
(Tim
e)
4
5
6
7
8
9
10
11
12
13
14
I
n
s
t
a
n
t
e
n
o
u
s
S
N
R
(
d
B
)
2 Tx 1 Rx
0
1
2
8
2
5
6
3
8
4
5
1
2
S
u
b
c
a
rrie
r
in
d
e
x
0
25
50
75
100
125
150
Transm
ission
fram
e
(Tim
e)
4
5
6
7
8
9
10
11
12
13
14
I
n
s
t
a
n
t
a
n
e
o
u
s
S
N
R
(
d
B
)
2 Tx 2 Rx
0
1
2
8
2
5
6
3
8
4
5
1
2
S
u
b
c
a
rrie
r
in
d
e
x
0
25
50
75
100
125
150
Transm
ission
fram
e
(Tim
e)
4
5
6
7
8
9
10
11
12
13
14
I
n
s
t
a
n
t
a
n
e
o
u
s
S
N
R
(
d
B
)
2 Tx 6 Rx
0
1
2
8
2
5
6
3
8
4
5
1
2
S
u
b
c
a
rrie
r
in
d
e
x
0
25
50
75
100
125
150
Transm
ission
fram
e
(Tim
e)
4
5
6
7
8
9
10
11
12
13
14
I
n
s
t
a
n
t
a
n
e
o
u
s
S
N
R
(
d
B
)
Near-Instantaneously Adaptive Cooperative Uplink
Schemes Based on Space-Time Block Codes and
V-BLAST
Mohammed El-Hajjar
, Salam Zummo
, and Lajos Hanzo
School of ECS, University of Southampton, UK.

Email: meh05r,lh@ecs.soton.ac.uk
Outline
J Overview of the State-of-the-art
J Cooperation Strategies
J Adaptive System 1
J Adaptive System 2
J Conclusions and Future Work
Key References
J L. Hanzo, C. Wong, and M. Yee, Adaptive wireless transceivers: turbo-coded, turbo-equalized and
space-time coded TDMA, CDMA, and OFDM systems. Chichester, UK: John Wiley and Sons, 2002.
J A. Sendonaris, E. Erkip, and B. Aazhang, User cooperation diversity. Part I. System description, IEEE
Transactions on Communications, 2003.
J G. Scutari and S. Barbarossa, Distributed space-time coding for regenerative relay networks, IEEE
Transactions on Wireless Communications, 2005.
J M. Dohler, E. Lefranc, and H. Aghvami, Space-time block codes for virtual antenna arrays, IEEE
International Symposium on Personal, Indoor and Mobile Radio Communications, 2002.
J E. C. Van Der Meulen, Three-terminal communication channels, Adv. Appl. Prob. 3, 1971.
J T. Cover, A. El Gamal, Capacity Theorems for Relay Channel, IEEE Transactions on Information Theory,
1979.
J A. Nosratinia, T. Hunter, A. Hedayat, Cooperative Communication in Wireless Networks, IEEE
Communications Magazine, 2004.
J N. Ahmed, M. A. Khojastepour, B. Aazhang, Outage Minimization and Optimal Power Control for the Fading
Relay Channel, IEEE Information Theory Workshop, 2004.
J J.N. Laneman, D. Tse, G. Wornell, Cooperative Diversity in Wireless Networks: Efcient Protocols and
Outage Behaviour, IEEE Transactions on Information Theory, 2004.
Multiple Input Multiple Output Systems
J Benets?
Increase of capacity by a factor of minM,N
J How?
Mitigating multipath fading
J Challenges
Channel estimation, synchronization, power control,
resource-allocation/relay-selection, multiplexing loss.
Cooperation in Wireless Networks
J Transmit diversity requires more than one antennas at the transmitter.
J However, many wireless devices are limited by size, cost and hardware complexity.
J By using cooperative communications, several transmitters or receivers may share
their antennas to form a virtual or distributed multiple-antenna system.
J Distributed nodes re-transmit the signal either in a regenerative or non-regenerative
fashion.
Cooperation Strategies
J Virtual MIMO
H Transmitting nodes have to dedicate resources to ex-
change their data. What is the optimum split of:
primary versus secondary signalling
When also considering the detrimental effects of
extra interference.
H Use space-time code for transmission.
J Relaying
1. Decode-and-Forward
2. Compress-and-Forward
3. Amplify-and-Forward
Nodes can use orthogonal or non-orthogonal
channels.
Relay Operation
J Full Duplex
J Half Duplex (simultaneous transmission and reception is unrealsitic)
Virtual MIMO
J Tx
1
and Tx
2
exchange data for cooperation (virtual STC or SDM).
J Tx
1
and Tx
2
broadcast data.
J Tx
1
and Tx
2
cooperation as well as Rx
1
and Rx
2
cooperation form a virtual MIMO
system.
J Use antennas for diversity
J Use antennas for multiplexing
System 1
User 1
User 2
User 3
User 4
System 1 Exploit users or MSsantennas for diversity.
J MS 1 broadcasts its data in time slot 1.
J Four users transmit in a G4-like STBC conguration.
J total number of time slots for transmission is 12.
J Can save one time slot if the BS can receive the users data while the cooperating
users are exchanging their data.
J In order to increase the bandwidth efciency of the system, Adaptive Quadrature
Amplitude Modulation (AQAM) can be combined together with the proposed
cooperative scheme.
J The adaptation is carried out according to the following table, while maintaining a
target BER of 10
3
.
No. of users per cluster 4
No. of Rx Antennas 1
Mode 1 G4 STBC, BPSK
BE= 0.364 bits/sec/Hz
Mode 2 G4 STBC, QPSK
BE=0.727 bits/sec/Hz
Mode 3 G4 STBC, 16-QAM
BE=1.454 bits/sec/Hz
Mode 4 G4 STBC, 64-QAM
BE=2.18 bits/sec/Hz
System 1
AQAM aided 1Tx, 1Rx
non-coop. benchmarker System
System 1
J Diversity order of 4.
J Transmit four users data in 11
time slots.
Benchmarker Scheme
J Diversity order of 1.
J Transmit four MSs data in 4 time slots.
Thus, the proposed Scheme has a higher diversity than a non-cooperative system and
hence reaches the target BER at SNR of 7.5 dB as compared to SNR of 23 dB for the
non-cooperative system.
This is achieved at the expense of a throughput reduction in the proposed scheme as seen
in the gures.
System 2
User 1
User 2
User 3
User 4
System 2 Exploit the users or MSs antennas for Multiplexing.
J Four users transmit in a four-layer V-BLAST-like transmission conguration.
J Adaptation is carried out according to the following table; (BE: Bandwidth Efciency)
No. of users per cluster 4
No. of Rx Antennas 4
Mode 1 one user
BE=2 bits/sec/Hz
Mode 2 Two users
BE=4 bits/sec/Hz
Mode 3 Three users
BE=6 bits/sec/Hz
Mode 4 Four users
BE=8 bits/sec/Hz
System 2
Non-Coop. benchmarker System
Conclusions
J We proposed two adaptive systems, which amalgamate the advantages of cooperative diversity, distributed
STBC as well as V-BLAST, while near-instantaneously adapting the system conguration for the sake of
achieving the highest possible throughput, while maintaining a given target BER of 10
3
.
J System 1 benets from a higher diversity gain with the aid of STBC while varying the BE by adapting the
modulation scheme employed. System 1 is capable of maintaining a BE between 0.364 bits/sec/Hz and
2.18 bits/sec/Hz.
J System 2 benets from the higher multiplexing gain of V-BLAST and thus has an effective BE varying
between 2 bits/sec/Hz and 8 bits/sec/Hz.
Future Work
J Mathematical performance analysis of the two proposed systems.
J Design of an optimised adaptive scheme, where adaptation will be based on the reliable channel quality
metric of the estimated BER value of the received frames, rather than on the less reliable channel-SINR
metric.
A Detailed Design Example
MIMO
Encoder
MIMO
Decoder
M Transmit antennas N Receive antennas
Questions concerning the design of MIMO systems:
(Q1) How can we achieve the maximum diversity gain, while maintaining a high
throughput?
(Q2) How can we contrive practical near-capacity schemes for arbitrary MIMO
congurations for operating right across a wide SNR region, rather than at a single
SNR value?
Orthogonal Approach for Diversity-Oriented
STBCs
Alamoutis scheme:
G
2
=
_
_
s
1
s
2
s
2
s
1
_
_
.
G
2
is an orthogonal/unitary matrix.
Simple single-symbol decoding.
Orthogonality can be relaxed to create Quasi-Orthogonal STBCs (QO-STBCs).
Can we view G
2
differently?
Layered Approach
The Alamouti scheme:
G
2
=
_
_
s
1
s
2
s
s

1
_
_
.
Two layers, each layer encoded using repetition coding.
Orthogonality enables the separation of the layers.
How many layers can a STBC codeword accommodate?
How to design better encoding method within each layer?
Linear Dispersion Codes
+
+
A
Q
s
1
s
Q
]
T
A
1
T temporal dimensions
= [
dimensions
spatial
M
S =
Q
q=1
A
q
s
q
=
K
LDC(MNTQ), with arbitrary modulation schemes.
Q non-separable layers.
Optimization of .
A single Dispersion Character Matrix (DCM) .
The Burden of Channel Estimation
Complexity and power consumption.
Channel estimation errors degrade the attainable performance.
Hence, two questions are asked.
1. How to dispense with the pilot-based channel estimation?
2. 3. At the same time, maintain the general framework of LDCs?
Unied Framework of DLDCs
SpaceTime
Coding
Detector
ML
Transform
Cayley
Delay
SpaceTime
Mapper
Differential Encoder
Mapper
PAM
SpaceTime
Demapper Decoding
PAM
Q symbols
S
n1
M
X
n
K
n
= [s
1
n
, , s
Q
n
]
T
. .
X
n
S
n
N
Y
n
K
n
X
n
Differential encoder: S
n
= S
n1
X
n
imposes the unitary constraint.
Linear structure:

X
n
=

Q
q=1
A
q
s
q
.
The Cayley transform: unique mapping between

X
n
and X
n
.
Cooperative STBCs
Penetration into inside room
Base Station
Coverage extesnsion at cell edge
Shadow of buildings
Relays
Relays
Relays
Relays
Underground
In cellular networks.
In wireless sensor networks, ad-hoc networks.
Schematic of the CLDCs
N receive antennas
+ T =
TwinLayer CLDC(MNTQ)
Source Node
R1
RM
. .
Y
. .
K = [s1, , sQ
| {z }
Q
]
T
Base Station having
Broadcast Interval (T
1
)
M number of Relays
Cooperation Interval (T
2
)
Z1
ZM
S
2
(
2
) S
1
(
1
)
Two phases of transmission.
Parameter(MNTQ), T = T
1
+T
2
.
How to design S
1
and S
2
?
Dispersion Matrices of CLDCs
+ +
T2
[
+ +
T2
[ r
2
1
Z
M
r
1
M
r
2
M
R
M
=
The M-th Relay
The 1st Relay
_
_
_
_
_
_
_
z
1
1
z
1
2
.
.
.
.
.
.
z
1
T2
_
_
_
_
_
_
_
. .
=
_
_
_
_
_
_
_
z
M
1
z
M
2
.
.
.
.
.
.
z
M
T2
_
_
_
_
_
_
_
. .
=
Z
1
r
1
1
R
1
= r
T1
1
]
T
r
T1
M
]
T
B
1
B
M
S
2
=
_
_
Z
T
1
.
.
.
Z
T
M
_
_
Linking LDCs, DLDCs and CLDCs
Colocated MIMO Systems Cooperative MIMO Systems
LDC
STBC
DLDC
Unitary constraint
CLDC
DSTBC Relay Protocol
Twinlayer structure

1
,
2
Existing Knowledge
MIMO channels capacity.
Serial Concatenated Codes (SCCs), which are capable of achieving an innitesimally
low BER.
SCCs can be analyzed and designed using EXIT charts.
Irregular design principle, using Irregular Convolutional Codes (IRCCs).
Improving the exibility using irregular inner codes.
Schematic of the Irregular Scheme
MMSE Decoder Precoder Decoder
MMSE Decoder Precoder Decoder
Conv. Decoder
Conv. Decoder
Precoder
Precoder
Conv. Encoder
Conv. Encoder LDC Encoder
LDC Encoder
Q1
2
Q
2
Q1
2
Q
2
Q
1
1
Partitioner
Irregular
Partitioner
Irregular
Partitioner
Irregular
Partitioner
Irregular
Partitioner
Irregular
N
1
IRCC Encoder
Q
1
IRCC Decoder
P
in
M
Q
2
Q
2
Mapper
P
in
P
out
Y
u2
c2 u3
c2 u3
ST
P
out
c1
u1
IR-PLDC Encoder
IR-PLDC Decoder
S
IA
IE
IA
IE
Operational SNR Region
15 10 5 0 5 10 15
0
0.5
1
1.5
2
2.5
3
3.5
4
SNR (dB)
C
(
b
i
t
s
/
s
y
m
/
H
z
)
MIMO, CCMC capacity
IRCCcoded IRPLDC
P
in
and P
out
= 6.
The recorded throughput values were achieved by changing
out
and
in
.
Operational SNR region can be extended.
Small-Scale Fading with Perfect CSI
0 5 10 15 20 25 30 35 40
0
0.2
0.4
0.6
0.8
1
1.2
SNR or
RB
(dB)
E
f
f
e
c
t
i
v
e

t
h
r
o
u
g
h
p
u
t

(
b
i
t
s
/
s
y
m
/
H
z
)
LDC(323Q)
DLDC(323Q)
CLDC(323Q)
Throughput comparison for the LDCs, the DLDCs and the CLDCs recorded at
BER=10
4
.
Small-Scale Fading with Imperfect CSI
0 5 10 15 20 25 30 35 40
0
0.2
0.4
0.6
0.8
1
1.2
SNR (dB)
E
f
f
e
c
t
i
v
e

t
h
r
o
u
g
h
p
u
t

(
b
i
t
s
/
s
y
m
/
H
z
)
LDC(323Q), = 10dB
LDC(323Q), = 9dB
DLDC(323Q)
BER=10
4
and the imperfect CSI governed by (dB).
Large-Scale Shadowing with Perfect CSI
0 5 10 15 20 25 30 35 40
0
0.2
0.4
0.6
0.8
1
1.2
SNR or
RB
(dB)
E
f
f
e
c
t
i
v
e

t
h
r
o
u
g
h
p
u
t

(
b
i
t
s
/
s
y
m
/
H
z
)
LDC(323Q), shadowing with =6dB
DLDC(323Q), shadowing with =6dB
CLDC(323Q), no shadow fading
BER=10
4
.
Implementational
complexity
Channel
characteristics
Effective
throughput
bandwidth
System
Bit error
rate
Coding gain
scheme
Modulation
Coding/
Coding rate
Coding/interleaving
delay
Figure 5: Factors affecting the design of MIMO-aided wireless communications
schemes c _John Wiley, Hanzo, Liew, Yeap: Turbo coding, turbo equalization and space-
time coding, Wiley & IEEE Press 2002
IEEE Vehicular Technology Conference 2009 Fall
Anchorage
Multiple-Symbol Differential Sphere Detection for the
Amplify-and-Forward Cooperative Uplink
Li Wang and Lajos Hanzo
School of Electronics and Computer Science ,
University of Southampton, UK.
Email: lh@ecs.soton.ac.uk
Outline
J Motivation & Contributions
J The principle of Multiple-Symbol Differential Sphere De-
tection (MSDSD)
J MSDSD Designed for the Amplify-and-Forward Coopera-
tive Uplink
J Results and Discussions
J Conclusions
Motivation
J Background
H It is widely recognized that the Differential Amplify-and-Forward (DAF) transmission scheme
is capable of providing a superior performance compared to classic direct transmissions
employing differential detection in slow-fading channels.
H In reality the channels connecting the multiple nodes of a cooperative system typically
become time-selective due to the relative mobility of the cooperating terminals.
J Problem To Be Solved:
H The performance gain achieved by the DAF-aided cooperative system may erode as the
environment becomes more time-selective.
H How to mitigate the performance degradation induced by the relative mobility of the
cooperating terminals?
Contributions
J Based on the principle of the single-path Maximum-Likelihood Multiple-Symbol
Differential Detection (ML-MSDD) contrived for conventional direct transmission
systems, a multi-path ML-MSDD scheme was specically designed for mitigating the
error oor encountered by the DAF-aided user-cooperation assisted system in
time-selective channels.
J Sphere Detection (SD) is employed to reduce the complexity imposed by the
ML-MSDD, resulting in Mutliple-Symbol Differential Sphere Detection (MSDSD), which
is invoked for our DAF-aided cooperative system.
Classication of the Differential Detection
Differential Detection
MultipleSymbol
Conventional
DecisionFeedback
MSDSD
Optimum
MLMSDD
Suboptimum
MSDD MSDD
Effects of Doppler Frequency on Conventional Differential Detection (CDD)
0 5 10 15 20 25 30 35 40
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
t
[
]
|

f
d
=0.03
f
d
=0.01
f
d
=0.001
(a) Magnitude of temporal correlation
function of Rayleigh fading channels
0 5 10 15 20 25 30 35 40
10
4
10
3
10
2
10
1
10
0
SNR (dB)
B
E
R

f
d
=0.03
f
d
=0.01
f
d
=0.001
DQPSK
(b) BER of CDD
The Rayleigh channels temporal autocorrelation function E h
k
h
k+
= 2
2
J
0
(2B
d
T).
3dB loss in comparison to the coherent detection aided systems in a slow-fading scenario.
Time-variant channel (fast fading) signicantly degrades the CDDs performance, since the observation window size is only
N
wind
= 2.
Principle of the ML-MSDD for Direct Transmission
J Basic Ideas:
H In simple terms, the MSDD is a de-
tector that makes a decision about
a block of (N
wind
1) consecutive
PSK symbols based on N
wind
re-
ceived samples.
H This enables the detector to exploit
the correlation between the phase
distortions experienced by the con-
secutive transmitted PSK symbols.
J ML Metric:
s
ML
=arg max
sC
N
P(r[s)
=arg min
sC
N
exp(Trr
H
1
r)
(
N
wind
det)
=arg min
sC
N
Trr
H
1
r,
where the vectors s and r contain
the consecutively transmitted and re-
ceived symbols, respectively, and =
E rr
H
[s.
0 5 10 15 20 25 30 35 40
10
4
10
3
10
2
10
1
10
0
P/N
0
(dB)
B
E
R

f
d
=0.03
f
d
=0.01
f
d
=0.001
N
wind
=2
DQPSK
N
wind
=9
Performance Improved by the ML-MSDD.
Single-Symbol System Model for the DAF-Aided Cooperative System
source
destination
relay
T
R
2
T
S
T
R
1
T
D
h
srU1
h
r1d
h
r2d
h
rU1d
h
sd
h
sr1
h
sr2
T
R
U1
TDMA-Based User-Cooperation Aided System.
Frame length: L
f
.
Power Constraint:
U1
u=1
P
r
u
= 1 P
s
, where
P
s
and P
r
u
represent the power transmitted by
the source node and the uth relay node, re-
spectively.
The gain factor employed by the uth relay
node f
AM
r
u
=
_
P
r
u
P
s
2
sr
u
+N
0
Y
n
= S
n
H
n
+W
n
, (1)
where S
n
, H
n
and W
n
, respectively, are given as follows:
S
n
= diags
sd
[n], , s
sd
[n]
. .
U elements
= diage
j2m/M
, , e
j2m/M
. .
U elements
,
H
n
=
_
_
P
S
h
sd
[n],
_
P
S
f
AM
r
1
h
sr
1
[n]h
r
1
d
[n+1 L
f
],
,
_
P
S
f
AM
r
U1
h
sr
U1
[n]h
r
U1
d
[n+(U 1) L
f
]
_
T
, (2)
W
n
=
_
w
sd
[n], f
AM
r
1
w
sr
1
[n]h
r
1
d
[n+1 L
f
]
+w
r
1
d
[n+1 L
f
], , f
AM
r
U1
w
sr
U1
[n]h
r
U1
d
[n+(U 1) L
f
] +w
r
U1
d
[n+(U 1) L
f
]
_
T
, (3)
Multiple-Symbol System Model for the
DAF-Aided Cooperative System
Then, based on the single-symbol model of Eq. (1), we can construct the equivalent
multiple-symbol system model as:
Y = S
d
H+W, (4)
where the block matrix Y of the received signal contains N
wind
user-cooperation based
received symbols corresponding to N
wind
consecutively transmitted differentially encoded
symbols s
sd
[n], (n = 0, 1, , N
wind
1) of the source node.
Explicitly: Y = [Y
T
n
Y
T
n+1
Y
T
n+N
wind
1
]
T
, and the channels block matrix H as well as
the AWGN block matrix W are dened likewise as H = [H
T
n
H
T
n+1
H
T
n+N
wind
1
]
T
, and
W = [W
T
n
W
T
n+1
W
T
n+N
wind
1
]
T
, respectively.
Moreover, the diagonal block matrix of the transmitted signal is constructed as
S
d
= diagS
n
, S
n+1
, , S
n+N
wind
1
.
ML Metric for the MSDSD Designed for the DAF-Aided Cooperative System
J ML-MSDD Metric: Based on the multiple-symbol system
model of Eq. (4), the decision metric of the ML-MSDD
can be expressed as:
S
ML
= arg min
SC
N
wind
TrY
H
1
Y, (5)
where the conditional autocorrelation matrix is:
= S
d
(EHH
H
+E WW
H
)S
d
H
= S
d
CS
d
H
(6)
J Transformation to MSDSD Metric: Then, by further
constructing a (UN
wind
U
2
N
wind
)-element upper-
triangular block matrix U as:
U (FT
d
(Y))
, (7)
where the (UN
wind
UN
wind
)-element upper-triangular
matrix F, which satises F
H
F = C
1
with the aid of
Cholesky factorization, we nally arrive at:
S
ML
= arg min
sSC
N
wind
[[Us[[
2
R, (8)
J Tree Search within a Conned Hyper-Sphere: Since we
have an upper-triangular block matrix U, a tree search can
be carried out from the (n = N
wind
1)th element of s down-
wards to the (n = 1)th element, making sure that the accu-
mulated Partial Euclidean Distance (PED) does not exceeds
the introduced search radius R.
J Search Radius Update: The search radius R is updated by
calculating the Euclidean distance between the newly ob-
tained signal point s and the origin. Then a new search is
carried out with the updated search radius.
MSDSD Tree Search Example (DBPSK, N
wind
= 5)
1 (0.17)
2 (6.45) 3 (0.20)
4 (3.4)
5 (4.2) 6 (4.8) 8 (1.8)
7 (1.2)
9 (3.3) 14 (0.23) 15 (1.0)
13 (0.22) 16 (1.9)
12 (0.21) 11 (2.1)
10 (0.18)
0 (0)
n = 1
n= 4
n = 2
"1"
"0"
n = 3
Initially, R = 5
Figure 6: Illustration of the MSDSD algorithm with the aid of the classic tree searching:
The gure in ( ) indicates the PED of a specic node for the trial point in the modulated
constellation; while the number outside represents the order in which the points are visited.
BER Performance
0 5 10 15 20 25 30 35 40
10
4
10
3
10
2
10
1
10
0
P/N
0
(dB)
B
E
R

f
d
=0.03
f
d
=0.01
f
d
=0.001
DQPSK
N
wind
=11
N
wind
=2
TwoUser
Cooperation
NonCooperative
DQPSK
NonCooperative
QPSK
Complexity Reduction Achieved by the MSDSD in Two-User Cooperation Scenarioblack
J A potentially excessive-complexity search is
likely to be encountered when ML-MSDD is
employed, depending on the size of the con-
stellation M
c
and the observation window
size N
wind
, which prevents the application of
the full-search-based ML-MSDD detectors in
most practical cases.
J For example, in the scenario of using
DQPSK (M
c
=4) and an observation window
size of N
wind
= 11, the number of PED eval-
uations per symbol block is:
M
(N
wind
1)
c
N
wind
=
4
10
11
.
0 5 10 15 20 25 30 35 40
4
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
SNR (dB)
#

o
f

P
E
D

E
v
a
l
u
a
t
i
o
n
s

p
e
r

S
y
m
b
o
l

MSDSD (N
wind
=11, f
d
=0.03)
MSDSD (N
wind
=11, f
d
=0.01)
MLMSDSD (N
wind
=11)
MLMSDSD (N
wind
=2)
DQPSK
Conclusions
J A MSDSD scheme was proposed for mitigating the error oor encountered by the DAF-aided
user-cooperation aided system in time-selective channels, leading to a signicant performance gain over the
system using the CDD at a relatively low complexity.
J For example, given a target BER of 10
3
, a performance gain of about 10 dB can be attained by the
proposed MSDSD employing N
cand
= 11 for a DQPSK modulated cooperative two-user system in a
relatively fast-fading channel having a normalized Doppler frequency of 0.03.
IEEE Vehicular Technology Conference 2009 Fall
Anchorage, Alaska, USA
Distributed Self-Concatenated Codes for Low-Complexity
Power-Efcient Cooperative Communication
M. F. U. Butt, R. A. Riaz, S. X. Ng and L. Hanzo
School of Electronics and Computer Science ,
Outline
Introduction and Motivation
SECCC-ID
Binary SECCC-ID
Non Binary SECCC-ID
System Model
DSECCC-ID Encoder
DSECCC-ID Decoder
Design and Analysis Using EXIT Charts
Results and Discussions
Conclusions and Future Research
Key References
[1] A. Sendonaris, E. Erkip and B. Aazhang, User cooperation diversity part I: System description,
IEEE Transactions on Communications, vol. 51(11), pp. 19271938, 2003.
[2] N. Laneman, D. N. C. Tse and G. W. Wornell, Cooperative diversity in wireless networks: efcient
protocols and outage behavior, IEEE Trans. on Information Theory, vol. 50, no. 12, pp. 30623080,
2004.
[3] M. F. U. Butt, S. X. Ng and L. Hanzo, EXIT Chart Aided Design of Near-Capacity
Self-Concatenated Trellis Coded Modulation Using Iterative Decoding, in IEEE Vehicular Technology
Conference, Marina Bay, Singapore, pp. 734-738, May 2008.
[4] M. F. U. Butt, R. A. Riaz, S. X. Ng and L. Hanzo, Near-Capacity Iteratively Decoded Binary
Self-Concatenated Code Design Using EXIT Charts,, Proceedings of the Global Communications
Conference, Nov/Dec 2008.
[5] S. ten Brink, Convergence behavior of iteratively decoded parallel concatenated codes, IEEE
Transactions on Communications, vol. 49, pp. 17271737, Oct. 2001.
[6] B. Zhao and M. C. Valenti, Distributed turbo coded diversity for relay channel, IEE Electronics
Letters, vol. 39, pp. 786787, May 2003.
Introduction and Motivation
Cooperative Diversity Scheme [1,2]
small mobile unit: correlation of signals in MIMO.
User cooperation: independent fading.
Decode-And-Forward: error propagation.
Amplify-And-Forward: noise enhancement.
Self-Concatenated Convolutional Codes with Iterative Decoding (SECCC-ID)
Low complexity scheme invoking a single encoder and decoder [3,4].
EXIT chart analysis gives insight into decoding convergence behaviour [5].
Distributed Turbo Coding Scheme [6]
Power efcient.
Assumming a perfect source-relay link.
Non-binary SECCC-ID
An SECCC-ID scheme was designed using Trellis Coded Modulation (TCM) as con-
stituent codes [3].
The proposed design was symbol-based, therefore it had the inherent problem of ex-
hibiting a mismatch between the EXIT curve and the bit-by-bit decoding trajectory.
The studies published in the literature of SECCC-ID scheme do not aim for nding
good codes in terms of decoding convergence.
An EXIT chart based analysis of the iterative decoder provides an insight into its de-
coding convergence behaviour and hence it is helpful for nding the best constituent
codes for SECCC-ID without time-consuming bit-by-bit simulation of the actual sys-
tem.
The SNR value, where the turbo-cliff in the BER curve of a concatenated code appears
can be successfully predicted with the aid of EXIT charts.
However, the EXIT chart computation assumes the employment of a sufciently long
interleaver, so that the LLRs may be rendered Gaussian distributed.
Non Binary SECCC-ID Analysis Using EXIT charts
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
I
A
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
I
E
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Parity Gen Poly [77 2 10]
8
Estimated E
b
/N
0
= 3.02dB
.
EXIT curve
Trajectory of snapshot # 6
Trajectory of snapshot # 19
Fig 1. EXIT chart and two snapshot decoding trajectories
for half-rate 4PSK-assisted non binary SECCC-ID using
a block length of 10
4
symbols at E
b
/N
0
= 3.02 dB.
0 2 4 6 8 10 12 14 16 18 20
E
b
/N
0
(dB)
10
-5
10
-4
10
-3
10
-2
10
-1
1
B
E
R
.
.
.
.
.
.
.
.
.
.
.
.
0 iteration
4 iteration
.
10 iteration
20 iteration
Fig 2. Simulations results for half-rate non binary
SECCC-ID Code when communicating over uncorrelated
Rayleigh fading channels.
Binary SECCC-ID
The bit-based design eliminates the mismatch inherited by the symbol-based design
[4].
RSC encoder using the generator polynomials G=[13 15]
8
with a rate of R
1
= 1/2 and
memory = 3. Then there is an interleaver followed by a rate R
2
=
3
4
puncturer. The
overall code rate becomes R =
R
1
2R
2
=
1/2
2(3/4)
=
1
3
.
+
+
SISO
SECCC Decoder
SECCC Encoder
Mapper
Depuncturer
MAP Demapper
h n
1
1
L
e
(b
2
)
L
e
(b
1
)
L
o
(b
2
)
L
o
(b
1
)
L
a
(b
2
)
L
a
(b
1
)
L
a
(b
2
)
L
a
(b
1
)
b
1
R
y x
b
2
b
1
b
1
P
/
R
2
R
1 S
Puncturer
2
RSC
1
2
1
Encoder
Decoder
Fig 3. The block diagram of binary SECCC-ID System.
Self-concatenated Decoder
The received signal is passed through a soft demapper to calculate the conditional
probability density function (PDF) of receiving y, when x
(m)
was transmitted:
P(y[x
(m)
) =
1
N
0
exp
_
_
_
y hx
(m)
2
N
0
_
_
_
, (9)
where x
(m)
is the hypothetically transmitted QPSK symbol for m 0, 1, 2, 3.
These probabilities are passed to a soft-value depuncturer, which converts them to
bit-based Log-Likelihood Ratios (LLRs) and inserts zero LLRs at the punctured
bit positions. These are then deinterleaved and fed to the Soft-Input Soft-Output
(SISO) Maximum A Posteriori Probability (MAP) decoder.
The decoder calculates the extrinsic LLR of the information bits, namely L
e
(b
1
)
and L
e
(b
2
). They are appropriately interleaved to yield the a priori LLRs of the
information bits, namely L
a
(b
1
) and L
a
(b
2
), as shown in Fig. 3.
Self-concatenated decoding proceeds, for a xed number of iterations.
Distributed Binary Self-Concatenated Coding scheme using Iterative Decoding
(DSECCC-ID) for cooperative communications is analyzed with the aid of EXIT
charts.
The realistic condition of having an imperfect source-relay communication link is
considered.
Destination
Source
Relay
R
S
h
sr
D
S
rd
S
sr
h
sd
x
s
= x
sd,k
S
sd
h
rd
x
r
= x
rd,(k+N
s
)
Fig 4. Schematic of a two-hop relay-aided system.

DSECCC-ID Encoder
The source node transmits SECCC symbols to both the relay and the destination nodes
during the rst transmission period.
The relay performs SECCC-ID decoding. It then re-encodes the information bits using a
Recursive Systematic Convolutional (RSC) code during the second transmission period
The resultant symbols transmitted from the source and relay nodes can be viewed as the
coded symbols of a three-component parallel-concatenated SECCC-ID encoder.
DSECCC-ID Encoder
RSC
Encoder
Puncturer
Mapper
Relay Transeiver
SECCC Encoder
RSC
Encoder
Puncturer
Mapper
SECCCID Decoder
x
s
b
2
b
1
1
/
S
P
c
R
1

2
R
2
b
1
y
(T
1
)
sd
n
(T
1
)
sd
h
(T
1
)
sd
x
r c
0
p
0
p
1
1
r
n
(T
2
)
rd
R
3
R
4
y
(T
2
)
rd
h
(T
2
)
rd
R = R
1
/(2R
2
)
r
h
(T
1
)
sr
n
(T
1
)
sr
y
(T
1
)
sr
b
0
b
1
Fig 5. DSECCC-ID Encoder.
DSECCC-ID Decoder
There are two inputs to the RSC MAP decoder block, in Fig. 6.
The rst is the extrinsic information of bit b
1
provided by the SECCC-ID decoder,
which is obtained from the addition of L
e
(b
1
) and the deinterleaved version of L
e
(b
2
).
The resultant L
e
1
(b
1
) stream is interleaved by
r
to generate L
a
(b
0
).
The second input of the RSC MAP decoder is the interleaved and depunctured version
of the soft information provided by the QPSK demapper denoted as P(y
rd
[x
rd
).
DSECCC-ID Decoder
+
+
Decoder
(1)
Decoder
(2)
MAP
RSC
+
RSC Decoder
+
+
SISO
Demapper
SECCCID Decoder
+
+
Demapper
Depuncturer
+
MAP
1
1
L
e
(b
2
)
L
a
(b
1
)
L
a
(d)
P
S
/
L
e
(d)
S
P
/
L
e
(b
1
)
r
R
1
4
L
a
(b
2
)
b
1
L
o
(d)
R
1
2

1
2
L
o
(b
0
) L
e
(b
0
)
1
r
L
e
1
(b
1
)
1
1
L
a
(b
0
)
P(y
rd
[x
rd
)
L
a
(c) P(y
sd
[x
sd
)
y
sd
y
rd
L
a
2
(b
1
)
L
a
1
(b
1
)
L
a
1
(b
2
)
L
a
2
(b
2
)
1
Fig 6. DSECCC-ID Decoder.
DSECCC-ID Decoder
The RSC decoder then provides the improved extrinsic LLR of the data bit b
0
namely L
e
(b
0
) as its output, which is deinterleaved by
1
r
to yield L
a
2
(b
1
). The
LLR L
a
2
(b
1
) can be further interleaved using
1
to generate L
a
2
(b
2
).
These a priori LLRs output by the RSC can be added to the SECCC-ID
decoders a priori LLRs of b
1
and b
2
, thus completing the iteration between the
RSC and SECCC-ID decoders.
+
+
Decoder
(1)
Decoder
(2)
MAP
RSC
+
RSC Decoder
+
+
SISO
Demapper
SECCCID Decoder
+
+
Demapper
Depuncturer
+
MAP
1
1
L
e
(b
2
)
L
a
(b
1
)
L
a
(d)
P
S
/
L
e
(d)
S
P
/
L
e
(b
1
)
r
R
1
4
L
a
(b
2
)
b
1
L
o
(d)
R
1
2

1
2
L
o
(b
0
) L
e
(b
0
)
1
r
L
e
1
(b
1
)
1
1
L
a
(b
0
)
P(y
rd
[x
rd
)
L
a
(c) P(y
sd
[x
sd
)
y
sd
y
rd
L
a
2
(b
1
)
L
a
1
(b
1
)
L
a
1
(b
2
)
L
a
2
(b
2
)
1
Fig 6. DSECCC-ID Decoder.
Binary SECCC-ID Analysis Using EXIT charts
R
1
=1/2 and R
2
=3/4, QPSK-assisted
SECCC-ID, =3, =0.67 bit/s/Hz
achieves decoding convergence at
received SNR (SNR
r
) = -0.15 dB
for transmission over an uncorre-
lated Rayleigh fading channel.
Using SNR
r
at the relay determine
the equivalent SNR at the source
node (SNR
e
):
SNR
r
= SNR
e
+10log
10
(G
sr
) [dB] (10)
For G
sr
= 4, SNR
e
=3.5 dB.
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
I
A
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
I
E
R
1
=1/2, R
2
= 3/4, =3
SNR
r
= -0.15 dB
EXIT Curve
trajectory snapshot 1
trajectory snapshot 6
Fig 7. EXIT chart and snap-shot decoding trajectories
of binary SECCC-ID System.
DSECCC-ID Analysis Using EXIT charts
Self-concatenated iterations at the
source-destination link uses I
sd
= 2
and source-relay link uses I
sr
= 8.
The number of iterations between
the SECCC-ID and RSC at the des-
tination node is I
sd,rd
= 12.
At the DSECCC-ID decoder we
have I
DSCC
=I
sd
I
sd,rd
=24 decod-
ing iterations.
This makes the total decoding iter-
ations in the overall system equal
to I
sr
+I
DSCC
= 32 as compared to
a non-cooperative SECCC-ID em-
ploying I = 40 iterations..
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
I
A
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
I
E
SECCC-ID EXIT Curve, SD link
RSC EXIT Curve, RD link, G
sr
= 4
Trajectory, SNR
e
= -3.5dB
Fig 8. EXIT charts and a snap-shot decoding trajectory
of DSECCC-ID scheme for SNR
e
=3.5 dB both at the
source as well as at the relay nodes.
Our prediction is veried by computing the corresponding Monte-Carlo simulation-
related decoding trajectory for the DSECCC-ID scheme for a frame length of 120, 000
bits.
Thus the DSECCC-ID outperforms the SECCC-ID scheme by about 3.35 dB in SNR
terms.
-5 -4 -3 -2 -1 0 1 2
SNR
e
(dB)
10
-8
10
-7
10
-6
10
-5
10
-4
10
-3
10
-2
10
-1
1
B
E
R
SECCC-ID, R
1
=1/2,R
2
=3/4, = 3
DSECCC-ID, perfect relay
DSECCC-ID, realistic relay
Fig 9. BER versus SNR
e
performance of SECCC-ID and DSECCC-ID schemes.
Conclusions
A power and bandwidth efcient DSECCC-ID scheme has been proposed for cooperative
communications based on three-component concatenated design.
To exploit the low complexity offered by SECCC-ID codes we explored their application in
distributed cooperative communications.
Once the received SNR at the relay node exceeds the error-free decoding threshold, the
SECCC-ID decoder employed at the relay node becomes capable of reliably decoding the
source signals.
The EXIT chart of the three-component DSECCC-ID decoder seen in Fig. 8 reveals that a
benecial combination of the equivalent SNR of the source and the relay nodes results in a low
BER at the destination, despite considering a potentially error-prone reception at the relay.
Thus we reduced the total power required by the overall system as compared to an SECCC-ID
system. The overall complexity of this distributed cooperative communication system is less
than a non-cooperative SECCC-ID system.
Future Work
Near-capacity non-coherently detected DSECCC-ID cooperative schemes.
Performance for different relay location scenarios and power optimisation in
DSECCC-ID.
Performance Analysis of Hard Versus Soft Relaying in DSECCC-ID.
IEEE ICC2012
Ottawa, Canada
First-Hop-Quality-Aware Dynamic Resource Allocation
for Amplify-and-Forward Opportunistic Relayed SC-FDMA
Jiayi Zhang, Lie-LIang Yang and Lajos Hanzo
Communications Research Group
School of Electronics and Computer Science
Key References
C. Han, J. Zhang, L. Hanzo, et al., Green Radio: Radio Techniques to Enable Energy
Efcient Wireless Networks, in IEEE Commun. Mag. Special Issue: Green Commun.,
vol. 49, no. 6, pp. 46-54, Jun. 2011.
J. Zhang, L.-L. Yang and L. Hanzo, Energy-Efcient Channel-Dependent Cooperative
Relaying for the Multi-User SC-FDMA Uplink, in IEEE Trans. Veh. Technol., vol. 60,
no. 3, pp. 992-1004, Mar. 2011.
W. Dang, M. Tao, et al., Subcarrier-pair based resource allocation for cooperative
multi-relay OFDM systems, in IEEE ToW, vol. 9, no. 5, pp. 1640-1649, May 2010.
Y. Jing, et al., Single and multiple relay selection schemes and their achievable
diversity orders, in IEEE ToW, vol. 8, no. 3, pp. 1414-1423, Mar 2009.
K. B. Letaief, et al., Dynamic multiuser resource allocation and adaptation for wireless
systems, in IEEE Wireless Commun. Mag., vol. 13, no. 4, pp. 38-47, Aug. 2006.
A. Bletsas, et al., A simple Cooperative diversity method based on network path
selection, in IEEE J. Sel. Areas Commun., vol. 24, no. 3, pp. 659-672, Mar. 2006.
A Cooperative Communication Framework
Why Cooperation?
Aims: improving . . .
Ideal: benecial multiple-antenna technique MIMO
a
Reality: diversity gain is limited by the size of MT
Solution: virtual MIMO relay-assisted user cooperation, BS cooperation
a
multiple-input multiple-output
Why SC-FDMA? - A PAPR-aware Transmitter
Single-carrier vs. multi-carrier modulation?
a
a PAPR
b
issue
OFDM
c
IDFT
Fdomain Tdomain
BICM
S/P
shaping
Pulse
Insert
CP
P/S
xxx
f
[i] bbb
o
s(t)
(t)
xxx
t
[i]
TTT
H
N
SC-FDE
d
Tdomain
shaping
Pulse
Insert
CP
BICM
s(t)
(t)
xxx
t
[i] bbb
o
a
cyclic prex (CP) insertion, forward error correction (FEC) enabled by bit-interleaved coded modulation (BICM)
b
peak-to-average power ratio
c
orthogonal frequency-division multiplexing
d
frequency-domain equalisation
Why SC-FDMA? - Complexity aware receiver
design
Frequency-selective fading multi-path CIR
a
Conventional time-domain (TD) multi-tap equalisation
Sampling
Matched
filter
Equaliser BICM
ID
Tdomain
r(t)
(t)
WWW
H
b
o
zzz
t
[i] yyy
t
[i] y(t)
high complexity
Frequency-domain (FD) single-tap equalisation
Tdomain
CP
S/P
Fdomain Tdomain
P/S
IDFT Equaliser DFT Remove
Sampling
Matched
filter
BICM
ID
y(t)
WWW
H
TTT
H
N
zzz
t
[i] zzz
f
[i] yyy
f
[i]
TTT
N
yyy
t
[i]
r(t)
b
o
(t)
low complexity
a
channel impulse response
Dynamic Resource Allocation for Opportunistic
Relaying
Rationale: limited performance due to rst-hop quality
Purpose: exploring extra selection diversity
Aspects: rst-hop quality awareness
Objective: evaluate the advantages of joint design of
dynamic relay selection (DRS)
dynamic subband allocation (DSA)
Scope and Assumptions - Opportunistic
Relay-aided Multi-user SC-FDMA Uplink
Direct link is unavailable
Relays geographically localised in a cluster at midway
Cooperating relays are capable of exchanging CQI
a
BSs receiver employs multiple antennas
MTs BS ORs
firsthop secondhop
0
1
K 1
0
1
2
1
g
RD
j,m
0
2
N
r
1
J 1
g
SR
k,j
d
SR
= 0.5 d
RD
= 0.5
a
channel quality information
Source/Relay Model - Multi-user Version
Tdomain Tdomain
mapper
Subband
DFT
S/P
BICM
IDFT
P/S
Insert
CP
shaping
Pulse
Fdomain
source transmitter subbandbased AF relay transmitter
TTT
()
k
xxx
t
k
[i]
TTT
H
U
x x x
f
k
[i] xxx
f
k
[i]
(t)
x x x
t
k
[i]
s
k
(t)
TTT
N
bbb
o
k
K-user uplink, N subbands per user, U subbands in system
N-point DFT-spreading FFF
N
, U-point IDFT-modulation FFF
H
U
bandwidth expansion factor M full-load K = M
Channel coded version: using bit-interleaved coded modulation (BICM)
Source/Relay Model - Multi-user Version
IDFT DFT
Fdomain Tdomain Tdomain
S/P P/S
Subband
mapper
TTT
H
U
x
f
k,0
[i]
x
f
k,1
[i]
x
f
k,0
[i]
x
t
k,(N1)
[i]
x
t
k,1
[i]
x
t
k,0
[i]
x
t
k,0
[i]
x
t
k,1
[i]
x
t
k,(U1)
[i]
x
f
k,1
[i]
x
f
k,(U1)
[i]
TTT
N
TTT
k
x
f
k,(N1)
[i]
U = MN
frame
x x x
t
k
frame
xxx
t
k
SC-FDMA Signalling - Symbol-to-subband
Mapping
Localised FDMA
block duration T
v
, symbol duration T
s
, chip duration T
c
user-specic symbol-to-subband mapping PPP
S
k
for source or PPP
R
k
for relay
cyclic prex (CP) insertion
TD signal waveform
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
-1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
A
m
p
l
i
t
u
d
e
(
r
e
a
l
p
a
r
t
)
-1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
TD chip index: u
SC-FDMA BPSK TD signal waveform (N=4, M=4, U=16, k=0)
SC-FDE
TFD LFDMA
E.g. user k = 0, for N = 4,
M = 4, U = 16
FD subband mapping
3 0 0 0 0 1 1 1 1 2 2 2 2 3 3 3
subbands
subbands
DFTspread OFDM
signal
users
k k k k
N = 4
m = 0 m = 1 m = 2 m = 3
U = NM = 16
k-th
Relay/BS Receiver Model - Multi-user Version of
SC-FDE Receiver
Fdomain Tdomain Tdomain
S/P
Remove
CP
Sampling
Matched
filter
ID
BICM
P/S
DFT
Amplifier
/equaliser
Subband
demapper
IDFT
BS receiver subbandbased AF relay receiver
TTT
U
y(t) y y y
t
[i]
r(t)
(t)
yyy
f
k
[i] y y y
f
[i] zzz
t
k
[i]
b
o
k
zzz
f
k
[i]
TTT
H
N (TTT
()
k
)
T
f
k
or WWW
H
k
subband demapping PPP
S
k
at the relay or PPP
R
k
at the BS
relay: subband-based amplify-and-forward (AF)
f
k
BS: single-tap FDEWWW
H
k
diverse criteria
a
residual inter-symbol interference (ISI)
a
matched-ltering (MF), zero-forcing (ZF) or minimum mean-square-error (MMSE)
Relay/BS Receiver Model - Multi-user Version
Tdomain Fdomain Tdomain
IDFT DFT
S/P
demapper
Subband
P/S
Equaliser
www
H
k,0
TTT
N
www
H
k,(N1)
www
H
k,1
z
t
k,0
[i]
z
t
k,1
[i]
z
t
k,(N1)
[i]
y
f
0
[i]
y
f
(U1)
[i]
z
f
k,0
[i] y
f
k,1
[i]
z
f
k,1
[i]
z
f
k,(N1)
[i] y
f
k,(N1)
[i]
TTT
T
k
U = MN
frame
y
t
0
[i]
y
f
1
[i] y
t
1
[i]
y
t
(U1)
[i]
zzz
t
k
y
f
k,0
[i]
y y y
t
TTT
H
U
frame
Dynamic Resource Allocation for Opportunistic
Relaying
Relay selection
random relay selection (RRS)
dynamic relay selection (DRS)
Subband allocation
static subband allocation (SSA)
dynamic subband allocation (DSA)
Dynamic resource allocation (DRA) for OR
conventional DRS combined with DSA (DRS-DSA)
rst-hop-quality-aware joint DRA (FHQA JDRA)
JDRA-1: dominated by the second-hop quality
JDRA-2: dominated by the rst-hop quality
Dynamic Resource Allocation - Comparisons
Applied scenarios
Mode Channel dependence Relay strategy Controller
DRS-SSA R-D, low Distributed non-coop. BS
DRS-DSA R-D, high Localised non-coop. BS
JDRA-1 S-R & R-D, high Localised coop. RCH
a
JDRA-2 S-R & R-D, high Localised coop. RCH
CQI exchange
Mode Strategy Context
JDRA-1 MWR
b
S-R CQI order, and R-D CQI
JDRA-2 MWR S-R CQI and R-D CQI
Dynamic Resource Allocation - Channel and
Frame-structure
Example of dynamic subband allocation for two users
Head
MWR
Cooperative Cluster of Relays
Multipleaccess phase Broadcast phase
J
Dynamic Resource Allocation - Channel and
Frame-structure
Frame structure resource vector (RV) and resource block (RB)
RVs RBs 1 frame
FD
s
u
b
b
a
n
d
s
1 time slot
TD
Pilot RV Data RV RB
(N
rb
= N
v
/N
rv
)
= N
rb
= N
v
N
N
v
N
rv
DRA - Conventional Schemes
TD
TD
TD
TD
secondhop firsthop
FD
FD
FD
FD
FD
FD
FD
FD
FD
FD
j = 1
j = 2
j = 4
j = 0
j = 3
k = 2
k = 1
m = 0 m = 1 m = 2 m = 3 m = 0 m = 1 m = 2 m = 3
g
RD
j,m
g
SR
k,j
k = 2
k = 0 k = 1 k = 0
DRS-SSA
1. 2nd-hop DRS
2. 2nd-hop SSA
3. user assignment
TD
TD
TD
TD
secondhop firsthop
FD
FD
FD
FD
FD
FD
FD
FD
FD
FD
TD
TD
j = 1
j = 2
j = 4
j = 0
j = 3
k = 2
k = 1
m = 0 m = 1 m = 2 m = 3 m = 0 m = 1 m = 2 m = 3
g
RD
j,m
g
SR
k,j
k = 2
k = 1
k = 0
k = 0
DRS-DSA
1. 2nd-hop DRS
2. 2nd-hop DSA
3. user assignment
(randomly)
DRA - FHQA JDRA Schemes
RD channel gains at each subband
groups as RBs
Order the selected RBs having the
highest channel gains at all subband
groups of the selected relays
Firsthop
awareness ?
groups at each relay
SR channel gains at all subband
Order the users having the highest Assign the best
RBs to users
randomly
DSADRS
Yes
No
RD DRS
RD DSA
Joint DRA1
Select the relays having the highest
Assign the best RBs at the
corresponding relays to the
ordered users
Select the relays having the highest
SR channel gains of each user
SR DRS
Order the users having the highest
SR channel gains via their best
relay
Order the RBs having the highest
channel gains at all subband groups
of each selected relay
Assign the selected RBs at the
corresponding relays to the ordered
users
awareness
Firsthop
RD DSA
Joint DRA2
Parameters
S-D fading no S-D link
S-R, R-D fading frequency-selective Rayleigh
Power delay spread uniform
Shadowing absent (
2
= 0 dB)
Path-loss G
SR
= G
RD
= 0.5
4
Modulation 4-QAM (set-partitioning)
FEC coding Non or RSC (2,1,3), R
c
= 1/2
Channel model uncorrelated
Relaying protocol amplify-and-forward (AF)
Relay selection static or dynamic
Subband mapping localised
Subband allocation static or dynamic
Number of subbands per user N = 12
Bandwidth expansion factor M = 6
Total number of subbands U = 72
Number of source users K = 6
Number of paths L = 4
Transmit power P
S
k
= P
R
k
= 0.5
Number of BS receiver antennas N
r
= 1, 8
BER vs. E
b
/N
0
Uncoded OR SC-FDMA using MMSE FD-LE
10
-5
2
5
10
-4
2
5
10
-3
2
5
10
-2
2
5
10
-1
2
5
1
2
-10 -5 0 5 10 15 20 25
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
10
-5
2
5
10
-4
2
5
10
-3
2
5
10
-2
2
5
10
-1
2
5
1
2
B
E
R
-10 -5 0 5 10 15 20 25
E
b
/N
0
(dB)
SC-FDMA AF OR (N=12, M=K=J=6, L=4)
Non-FEC, 4-QAM (Gray), MMSE FD-LE Rayleigh:
DT, N
r
=1
DT, N
r
=8
OR, N
r
=1
OR, N
r
=8
.
RRS-SSA
DRS-SSA
DRS-DSA
JDRA-1
JDRA-2
AWGN DT
BER vs. E
b
/N
0
Uncoded OR SC-FDMA using MMSE-SIC FD-DFE
10
-5
2
5
10
-4
2
5
10
-3
2
5
10
-2
2
5
10
-1
2
5
1
2
-10 -5 0 5 10 15 20 25
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
10
-5
2
5
10
-4
2
5
10
-3
2
5
10
-2
2
5
10
-1
2
5
1
2
B
E
R
-10 -5 0 5 10 15 20 25
E
b
/N
0
(dB)
Non-FEC, 4-QAM (Gray), MS-MMSE FD-DFE (SIC) Rayleigh:
DT, N
r
=1
DT, N
r
=8
OR, N
r
=1
OR, N
r
=8
.
RRS-SSA
DRS-DSA
JDRA-1
JDRA-2
AWGN DT
BER vs. E
b
/N
0
BICM OR SC-FDMA using MMSE FD-LE
10
-5
2
5
10
-4
2
5
10
-3
2
5
10
-2
2
5
10
-1
2
5
1
2
-10 -5 0 5 10 15 20 25
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
10
-5
2
5
10
-4
2
5
10
-3
2
5
10
-2
2
5
10
-1
2
5
1
2
B
E
R
-10 -5 0 5 10 15 20 25
E
b
/N
0
(dB)
CONV(2,1,3), 4-QAM (SP), R
c
=0.5, MMSE FD-LE iter(0)
Rayleigh:
DT, N
r
=1
DT, N
r
=8
OR, N
r
=1
OR, N
r
=8
.
RRS-SSA
DRS-DSA
JDRA-1
JDRA-2
AWGN DT
BER vs. the number of users K
10
-5
2
5
10
-4
2
5
10
-3
2
5
10
-2
2
5
10
-1
2
0 2 4 6 8 10 12
10
-5
2
5
10
-4
2
5
10
-3
2
5
10
-2
2
5
10
-1
2
B
E
R
0 2 4 6 8 10 12
Number of Users: K
SC-FDMA AF OR (N
r
=1, N=M=12, J=6, L=4)
Non-FEC, 4-QAM (Gray), MS-MMSE FD-DFE (SIC)
E
b
/N
0
=3dB
.
RRS-SSA
DRS-SSA
DRS-DSA
JDRA-1
JDRA-2
BER vs. the number of relays J
10
-5
2
5
10
-4
2
5
10
-3
2
5
10
-2
2
5
10
-1
2
0 5 10 15 20 25 30 35
.
. .
. .
.
10
-5
2
5
10
-4
2
5
10
-3
2
5
10
-2
2
5
10
-1
2
B
E
R
0 5 10 15 20 25 30 35
Number of Relays: J
SC-FDMA AF OR (N
r
=1, N=12, M=K=6, L=4)
Non-FEC, 4-QAM (Gray), MS-MMSE FD-FDE (SIC)
E
b
/N
0
=12dB
.
RRS-SSA
DRS-SSA
DRS-DSA
JDRA-1
JDRA-2
DRA-aided OR-based SC-FDMA Uplink vs. DT
Benchmark
Transmitted power reduction
b
per bit may be achieved.
The schemes are listed in descending order of complexity.
b
(dB) Non-FEC BICM
N
r
1 8 1 8
JDRA-2 AF 11.5 1 9.5 2
JDRA-1 AF 11.5 11 9.5 1
DRS-DSA AF 8.5 14 7.5 3.5
RRS-SSA AF 6.5 14 5 3.5
Conclusions
We have: assumed that there are a number of inactive MTs acting as potential relays,
which are geographically co-located in a cell;
focused on the optimum exploitation of the relay selection and subband allocation for
the two-hop transmissions;
proposed two types of rst-hop-quality-aware joint dynamic resource allocation
schemes designed for the SC-FDMA uplink.
As a result, the reliability and power-efciency of our proposed systems was signicantly
improved.
Future Work
In order to mitigate the inter-cell interference (ICI) and reduce the energy per bit in the
multi-user multi-cell scenario, a trade-off has to be struck between the energy-efciency
and spectrum-efciency at a system level.
Relay-assisted multi-cell resource allocation
Link-quality-aware relay-assisted network optimisation
IEEE Wireless Communications and Networking Conference 2012
Paris, France
Effects of Practical Impairments on Cooperative
Distributed Antennas Combined with Fractional
Frequency Reuse
Jie Zhang, Rong Zhang, Xinyi Xu, Guangjun Li, Lajos Hanzo
Communications Research Group
School of Electronics and Computer Science
Email: jz4, rz, lh@ecs.soton.ac.uk
Outline
J System model
J Assumptions
J Precoding
J Practical impairments
J Synchronisation errors
J Results and discussions
J Conclusions
Key References
J M. Sadek and A. Tarighat and A. H. Sayed A leakage-based precoding scheme for downlink
multi-user MIMO channels, IEEE Transactions on Wireless Communications, vol.55, no. 5, pp.
17111721, May 2007.
J M. Speth and S. A. Fechtel and G. Fock and H. Meyr Optimum receiver design for wireless
broad-band systems using OFDM - Part I, IEEE Transactions on Communications, vol. 47, no.
11, pp. 16681677, Nov. 1999.
J X. Xu and R. Zhang and S. Ghafoor and L. Hanzo, Imperfect Digital-Fibre-Optic-Link-Based
Cooperative Distributed Antennas With Fractional Frequency Reuse in Multicell Multiuser
Networks, IEEE Transactions on Vehicular Technology, vol. 60, no. 9, pp. 44394449, Nov.
2011.
Motivation
J High Frequency Reuse (FR)
Factor:
low spectral efciency but high
SINR at the cell-edge
J Received signal and SINR
y
j
= h
0, j
x
j
+
iB
h
i, j
x
i
+n
j
j
=
[h
0, j
[
2
[n
j
[
2
+
iB
[h
i, j
[
2
B =B
i
, i 8, 10, , 18
FR transmission
J Unity Frequency Reuse (UFR)
high spectral efciency but low
SINR at the cell-edge
J Received signal and SINR
y
j
= h
0, j
x
j
+

iB
o
h
i, j
x
i
+n
j
j
=
[h
0, j
[
2
[n
j
[
2
+
iB
o
[h
i, j
[
2
B
o
=B
i
, i 1, 2, , 18
UFR transmission
J Fractional Frequency Reuse
(FFR)
high SINR at the cell-edge but
reduced spectral efciency
J Received signal and SINR at
cell-edge is the same as that of
FR transmission
J Received signal and SINR at
cell-centre is the same as that of
UFR transmission
FFR transmission
J Distributed Antenna System (DAS)
aided FFR Transmission
spatial diversity gain, mitigates the
pathloss and large-scale fading
high SINR and high spectral ef-
ciency
J Received signal and SINR at the
cell-edge
y
j
= h
j, j
x
j
+
N
a
i=1,i,=j
h
i, j
x
i
+n
j
=
[h
j, j
[
2
[n
j
[
2
+
N
a
i=1,i,=j
[h
i, j
[
2
n
j
= n
j
+
iB
h
i, j
x
i
denotes the
AWGN and the CCI from tire-two
cells.
FFR-DAS transmission
NonCoMP-aided FFR-DAS transmission
Non-Cooperative DAS-aided FFR DL transmission?
J FFR-DAS transmission is capa-
ble of providing
N
a
hot spots around the N
a
DAs
Path loss gain
Spatial diversity gain
Throughput of DAS-aided FFR transmission without CoMP
Throughput of FFR-DAS transmission without CoMP
Why cooperative?
J Reason
Strong CCI for users roaming near the angle halfway between the adjacent DAs for the
Non-CoMP aided FFR-DAS.
J Benets of CoMP
Increasing system throughput
Extending cellular coverage
Supports multiple users
J Requirement
full Channel State Information (CSI) at all BSs
CoMP-aided FFR-DAS DL transmission
J Congure
6 DAs act as a distributed BS
Supports N
u
= 6 users simulta-
neously
J Received signal and corre-
sponding SINR
y
j
= h
j
t
j
x
j
+
N
u
i=1,i,=j
h
j
t
i
x
i
+n
j
=
[h
j
t
j
[
2
[n
j
[
2
+
N
u
i=1,i,=j
[h
j
t
i
[
2
t
j
- precoding vector for MS j
employed at transmitter.
CoMP-aided FFR-DAS
Assumptions
J MS located at the angle halfway between DAs.
J The DL signal destined for the MSs is perfectly received from the BS at the DAs.
J Complex-valued additive white Gaussian noise (AWGN) C N (0, N
0
) at the MS
J Single receive antenna at each MS.
J Single DL transmit antennas at both the BS and the DAs.
Signal-to-Interference-Leakage-plus-Noise-Ratio (SILNR) DL
Precoding
J Aims for
Maximising the SILNR for the intended MS.
Minimising the leakage interference imposed on other MSs
J SILNR expresion
i, j
=
[h
i, j
t
i, j
[
2
N
u
k=1,k,=j
[h
i,k
t
i, j
[
2
+N
0
/P
i, j
h
i, j
channel vector between the DA i and the MS j
t
i, j
Precoding vector at DA i for transmitting the jth MSs signal.
P
i, j
Transmission power at DA i for MS j.
Signal-to-Interference-Leakage-plus-Noise-Ratio (SILNR) DL
Precoding
J Proportional power allocation strategy
P
i, j
=
[h
i, j
[
2
N
u
i,k=1
[h
i,k
[
2
P
i
P
i, j
power allocated for the jth MS at DA i
P
i
available power at DA i
J Solution
t
i, j
= max. eigv[(N
0
/P
i, j
+
N
u
k=1,k,=j
h
H
i,k
h
i,k
)
1
h
H
i, j
h
i, j
]
t
i, j
is multiplied by the allocated power P
i, j
.
CSI Imperfection: Estimation errors
J Channel at MS j with CSI estimation errors
h
j
=

h
j
+e
j
e
j
- Additive Gaussian noise at the jth MS having variance of
2
e
.

h
j
- The estimated channel at the jth MS with variance of 1
2
e
.
J The model is also suitable for other CSI errors, such as the CSI feedback delay.
CSI feedback model
CSI Imperfection: Quantisation
J Quantisation channel model
h
j
=|
h
j
|
h
j
|
h
j
| - Channel Quality Information (CQI).

h
j
- Channel Direction Information (CDI).
J Assumption
C =c
1
, c
2
, , c
2
b - Codebook having 2
b
entries is pre-designed and it is available to both
the MS and the BS.
CQI |
h
j
| and quantisation index
j
are perfectly fed back to the BS.
J Random Vector Quantisation (RVQ)
Codebook element c
i
is a randomly generated zero-mean, unity-norm Gaussian vector.
The quantisation index
j
= max
i=1,2, ,2
b cos, =(
h
j
, c
i
).
Synchronisation errors
Time offset: The Inter-Symbol-Interference (ISI) imposed by the time offset is treated
as additive noise n
with a variance of
m
[h
j
(
m
)[
2
[2
m
N
(
m
N
)
2
]
Frequency offset =
f
1/T
u
: Similar to the time-offset errors, the FD cross talk (ICI) is
also treated as additive noise n
having a variance of

2
3

2
Both the time offset as well as the frequency offset errors will degrade the received
SINR at the MS.
Worst-Case-Angle Results & Discussions
J Achievable SINR as a
function of Tx SNR with
imperfect CSI
Highest SINR is
achieved by FFR
transmission at high
Tx SNR.
SINR of CoMP trans-
mission under perfect
CSI outperforms the
UFR and FFR bench-
markers.
CoMP transmission
degrades at low
quantisation quality.
J SINR of CoMP-FFR un-
der both imperfect CSI as
well as a synchronisation
error of 0.2 x CP=0.2 x
72
Achievable SINR
becomes worse than
that of NonCoMP
transmission at low
SNR, when synchro-
nisation errors and
CDI quantisation
errors coexists.
J SINR loss at different
time offsets (CP=72)
Highest SINR loss for
FFR transmission
Similar SINR loss for
idealised and practi-
cal CoMP transmis-
sion.
Small SINR loss for
Non-CoMP and UFR
J SINR loss for different fre-
quency offsets
Similar trends to
those of time offsets
But more sensitive to
frequency errors than
to time offsets.
Conclusions
J CoMP-aided FFR-DAS was studied for imperfect CSI and synchronisation errors,
when the MS is located at half the angle between the DAs
SILNR precoding technique was employed.
Practical CSI errors as well as synchronisation errors were considered.
Higher SINR for CoMP transmission for both imperfect CSI as well as for synchronisation
errors.
More sensitive to frequency offsets than that to time offsets.
Future Research
Near-capacity multi-functional MIMO arrays supporting diverse non-coherent detection
aided asynchronously detected wireless tranceivers.
Cross-layer designed assisted green wireless systems
Thank You

Base Station Cooperation

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Base Station Cooperation

Încărcat de

Drepturi de autor:

Formate disponibile

School of ECS, Univ. of Southampton, UK. http://www-mobile.ecs.soton.ac.

uk [1] - [10] 1/ 137 [

, and Lajos Hanzo

School of ECS, University of Southampton, UK.

Fig 4. Schematic of a two-hop relay-aided system.

S-ar putea să vă placă și