close

Вход

Log in using OpenID

embedDownload
Various Applications of Full Duplex Techniques
in Wireless Communication Networks
Hyungsik Ju
Department of Electrical and Computer Engineering
National University of Singapore
2014. 04. 04
Introduction (1)
• Two Streams of Research on Full-Duplex
Hardware Implementation
– Self-interference cancellation
– Antenna structure
– Hardware demos
Communication Performance
– Theoretic approach
– Ideal hardware
– Perfect SIC or Simple SI Model
– Resource use in FD communication
– Various applications
– More practical SI & hardware model
– Advance protocol, new use of FD
Introduction (2)
– Simultaneous DL & UL support
– Advanced MAC protocol
– Advanced ARQ
U1
DL Users
UL Users
Improved PHY
layer security
Eavesdropper
– Advanced spectrum sensing
– Cognitive FDR
Cognitive radio network
Wireless powered network
Energy / information
full-duplex
Full-duplex multi-hop/twoway transmission
Ad-hoc network
Application of Full Duplex Technique 1
Full Duplex Communication in
Cellular Networks
Simultaneous DL and UL Support by FD BS
Self
interference
• HD BS : orthogonal DL & UL
• FD BS : simultaneous DL & UL
Effective Use
of Resource
Increased
Interference
Cochannel
interference
UL Users
DL Users
Transmission in slot 1
Transmission in slot 1
2
Transmission in slot 2
Interference Management
 Scheduling
 Precoding [Nguyen2013]
System Model
Received Signals
System Model [Nguyen2013]
DL :
DL
UL :
SI
CCI
UL
Assumptions
– Single-cell FD MU-MIMO
– No CCI, no scheduling considered
– SI considered to be background noise
– BD-ZF precoder & MMSE-SIC receiver at BS
Coupled
Joint design of inputs
for DL & UL
Design Objectives
 Spectral efficiency (SE)
: network sum rate for both DL & UL
 Energy efficiency (EE)
Total NW sum-rate
:
NW total energy consumption
Spectral Efficiency Maximization
Achievable Rates
–
– BD-ZF :
 No MUI
: linear precoder
–MMSE-SIC receiver at BS
Spectral Efficiency Maximization
Efficient
use of time
+
Spatial
multiplexing
gain
Loss of
spatial
degrees-offreedom
Energy Efficiency Maximization
• Linear Power Model [Arnold2010], [Xu2011]
Power Consumption in DL
Power Consumption in UL
,
,
Energy Efficiency – FD BS
Energy Efficiency – HD BS
Energy Efficiency with FD BS < Energy Efficiency with HD BS
due to sub-linear increase of data rate by FD
Simulation Results (1) – Spectral Efficiency
• Maximum Tx Power at BS
• Effect of Self-Interference
 Exploits spatial multiplexing gain
 Max. 25% gain with small SI in case of SM
 SIC performance is a performance limiting factor
Simulation Results (2) – Energy Efficiency
• Maximum Tx Power at BS
Dominated by
circuit power
consumption
• Effect of Self-Interference
Dominated by
Tx power
consumption
Similar EE in the low Tx power regime,
but EE of SE-optimal design decreases
as max. Tx power increases
 EE of HD > EE of FD
 Simultaneous DL & DL support
decreases EE
Application of Full Duplex Technique 2
MAC Layer Protocol with Full Duplex
Full Duplex Applied to MAC Protocol
HD MAC
– Simultaneous Tx & Rx Prohibited
– Less Spatial Reuse
– Hidden & Exposed Node Problems
Less Spectral
Efficient
Full Duplex
at (each) Node
FD MAC
–
–
–
–
Simultaneous Tx & Rx allowed
More Spatial Reuse
Resolved Hidden Node Problems
More Chance for Advanced Scheduling
More Spectral
Efficient
MAC Protocol Based on FD (1)
• Dual-Link by FD Transmission [Singh2011]
Symmetric Dual-Link
Asymmetric Dual-Link
– A starts packet transmission
– B decodes A’s header, and transmits packet or ‘busy’ tone while receiving
– A does not detect any transmission from B,
transmission to B is not successful
MAC Protocol Based on FD (2)
• Primary Sender
– Initiate (dual) link
– Transmit packet to primary receiver
• Primary Receiver = Secondary Sender
– Receive packet from primary sender while transmit packet to secondary receiver
– Only the primary receiver is allowed to initiate secondary transmission
• Secondary Receiver
– Receive packet from secondary sender
Primary receiver
= Secondary sender
Primary sender
Secondary receiver
MAC Protocol Based on FD (3)
• Signal Transmission in Dual Link [Singh2011]
A transmits ‘busy’ tone until
the end of B’s transmission
to prevent neighborhood interferer
Beginning transmission
= successful decoding
B transmits packets to C
while receiving packets
from A
Pros. & Cons.
• Pros: Hidden Node Problem Resolution
– Rx B sends a signal while receiving
 Prevents neighborhood nodes starting
transmission
– Perfect RTS/CTS with no Overhead
• Cons
 Secondary link may not be well protected as primary link
 Overly restrictive
advanced scheduling required
Simulation Results (1/2)
• Probability of Successful Transmission
A-B-A Route
Isolated
link (A,B)
A-B-C Route
Isolated
link (A,B or C)
– Increased packet loss probability in each individual link
– Improved combined throughput
Simulation Result (2/2)
• Throughput of One Example Network Topology
– Full-duplex nature : throughput improvement up to 30-50%
– Why not 100% improvement?
 Increased packet loss probability
 Full exploitation of FD may not be possible depending on traffic pattern
Application of Full Duplex Technique 3
Improved ARQ Based on Full-Duplex
Limit of FD : High Node Density Networks
Capacity Gain
by Full-duplex Tx
Pair Interference
by Simultaneous Tx
Vs.
Challenges
In High Node Density

Full-duplex < Half-duplex
Method : ARQ Protocol
BFD in Two-way Ad-hoc Networks with Retransmissions
Network Performance
(Transmission Capacity)
P2P Environment
with High Node Density
Effect of ARQ Protocol
BFD is More Efficient Way of Two-way Communication
Background – Transmission Capacity
• Transmission Capacity in Ad-Hoc Networks [Weber2005]
Channel Capacity
Transmission Capacity
 Amount of reliable transmitted
information over a channel
 Maximum number of successful
transmission per unit area
Interference
Desired
 Definition
S

C  log 2 1   [bits/s/Hz]
 N
Signal-to-noise ratio
 Definition
TC   (1- )C0 [bits/s/Hz/m 2 ]
Max. spatial density guaranteeing
C0  log 2 (1   ),   P(   ) : outage probability
Bi-Directional ARQ Protocol (1)
• Improving Reliability of BFD System (Single Feedback CH) [Kim2014]
Conventional ARQ
Protocol
DATA i
DATA i
ACK
i
a0
a0  b0
DATA i
b0
ACK i
DATA i1
a0  b0
NACK i
b0  a0
DATA i
ACK i DATAi1
b0  a0
Proposed Bi-directional ARQ (Bi-ARQ) Protocol
DATA i
a0
a0  b0
DATA i
b0
ACK i
b0  a0
DATA i1
ACK i1 DATA i 2
a0  b0
NACK i
DATA i
ACK i DATAi1
b0  a0
 Retransmit protocol : only error data packet is retransmitted
 Bi-directional ACK/NACK : reduced waiting time with full-duplex
Bi-Directional ARQ Protocol (2)
• Aggregated Interference in BDF System with Retransmission
b1
Performance Metric
a1 '
a1
b1 '
Conventional Two-way TC
Retransmitted node pair
a0
b0
b3
a3
a2
TC  P (1  Pout ) Cab  Cba 
b2
Nodes in  A
Nodes in  B
linkab , linkba
Nodes in  A '
Nodes in  B '
linka 'b ' , linkb ' a '
Modified Two-way TC
with Bi-ARQ [Definition 3.1]
TC 
P
'
(1   ) Cab
 Cba' 
1  N rt
① ②
③
 Loose : ①Normalization factor, ③SIR degradation ( eff  (1  N rt ) )
 Gain : ② Decreased outage probability (  ( Pout ) N RT 1 )
N rt : Maximum number of packet retransmissions
P : length of transmitted packet
N rt : Average number of packet retransmissions
*Assumption1 : echo-channel interference is perfectly eliminated
BFD vs. BHD Systems
• Two-Way TC of BFD System with Bi-AQR Protocol [Kim2014]
BFD : TC ( F ) 
P
(F )
(F )
(F )


(1


)
1
l
og
(1


)

1
log
(1


2
ab
2
ba ) 
(F )

1  N rt
BHD : TC ( H ) 
P
(H )
(H )
(H )

(1


)

log
(1


)

1


log
(1


)  , 0    1


2
ab
2
ba
(H )

1  N rt

Example (Given SIR threshold
)
N rt
0
BFD
BHD
0.01
4
0.00001
0.99
0.99999
0.0025
0.0000003125
0.95 0.0475 0.9975
0.9999996875
( F ) N rt 1
( Pout
)
(F )
Pout
 0.1
1 
0.9
0.09
(H )
( H ) N rt 1
Pout
 0.05
( Pout
)
1 
1
  ( Pout ) N
RT
1
, Pout  Pr(   )
Simulation Results (1)
• Two-Way TCs of BFD & BHD vs. N rt (High Node Density)
 =0.1
①
TC gain
via Bi-ARQ
BFD < BHD
 For high node density
 Severe pair interference
Using Bi-ARQ Protocol
②
①
②
BFD > BHD
 Optimal # of retransmissions
 BFD > BHD
 Benefit from outage prob.
 BFD is More Efficient Way of Two-way Communication
Simulation Results (2)
• Two-Way TCs of BFD & BHD vs.  (Strict Outage Constraint)
 =0.0001
②
① BFD < BHD
 For strict outage constraint
 High outage prob. of BFD
Using Bi-ARQ Protocol
② BFD > BHD
 Optimal # of retransmissions
 BFD > BHD
 Benefit from outage prob.
 Increased optimal node density
①
Low TC due to small
nodes density
Low TC due to
Severe interference
Application of Full Duplex Technique 4
Full Duplex Multi-Hop Networks
Full-Duplex in Multi-Hop Network [Ju2012]
Simultaneous Transmission
S
Typical Route
L0
R
R
R
D
FD used for 1-way packet relaying at a relay node
d 0  L0 / M 0
Simultaneous Transmission
S
R
R
R
D
FD used for 2-waypacket exchange in a hop
: Source
: Relay
: Destination
: Packet transmission within a given time slot
 N antennas, rank-1 transmission, 1 hop / route at a time slot
 Equidistance b/w nodes
 Neglecting AWGN, perfect SIC
 End-to-end delay : avg. # of slots for a packet from S to D
 Route throughput : avg. # of packets successfully transmitted for a given slot
1-Way Packet Relaying (1)
• 1-Way Packet Relaying : FDR vs. HDR (1)
HDR Mode
T1
Time Slot
Index
T3
T2
Transmitting Node(s)
HDR Mode
FDR Mode
slot t
Source
Source
slot t  1
Relay 1
Source
Relay 1
slot t  2
Relay 2
Relay 1
Relay 2
Relay M-1
Relay M-2
Relay M-1
T4
slot t  M  1
FDR Mode
slot t  M
Source
Relay M-1
slot t  M  1
Relay 1
Source
Each node transmits
every M slots
Each node transmits
twice every M+1 slots
T1
T2
transmits every
(M+1)/2 slots
1-Way Packet Relaying (2)
• 1-Way Packet Relaying (2)
FDR Mode
Avg. # of Slots to
Transmit a Packet
HDR Mode
M 1
2
M

Interference
Measure
Hop Success
Probability
End-to-End
Delay
Route
Throughput
Benefit from
Channel Use
Less
Interference
1-Way Packet Relaying (3)
• End-to-End Delay
• Route Throughput
250
FDR mode
HDR mode
FDR mode
HDR mode
150
Slots/Packet)
(#ofofPackets/Slot)
Throughput
Throughput (#
Slots/Packet)
Delay
of Slots/Packet)
(# of
Delay (#
200
The delay of FDR mode is smaller than
that of HDR mode as interferer density
is lower or required rate is smaller
100
50
-2
10
Full duplex BF has larger throughput
Throughput of FDR mode is larger than that of
HDR mode as interferer density is lower
or required rate is smaller.
Full duplex BF has smaller delay
0
-5
10
-3
-4
10
-3
10

Product of intensity and rate constraint:   th
FDR > HDR
 Low rate constraint / rare interferer
 Channel Use > interference
10
-4
10
-3
10

Product of intensity and rate constraint:   th
FDR < HDR
 High rate constraint / dense interferer
 Channel Use > interference
2-Way Packet Exchange (1)
• 2-Way Packet Exchange
T1
T3
T2
BBF Mode
SIR Threshold
to satisfyRth
Hop Success
Probability
End-to-End
Delay
Route
Throughput
Benefit from rate constraint
T4
UBF Mode
2-Way Packet Exchange (2) - Hop Density
• End-to-End Delay vs. Hop Density
• Route Throughput vs. Hop Density
400
350
When required rate is small, throughputs of
BBF mode and UBF mode are equivalent
BBF mode
UBF mode
-2
Throughput (# of Packets/Slot)
Delay (# of Slots/Packet)
300
Rth  5
250
As required rate becomes
larger, delay of BBF mode
150 becomes smaller that that
of UBF mode
200
Rth  1
100
Rth  1
As required rate becomes
larger, throughput of BBF
mode becomes larger that
that of UBF mode
-3
10
Rth  5
When required rate is small, delays of
BBF mode and UBF mode are equivalent
50
0
-5
10
10
BBF mode
UBF mode
-4
-4
10
-3
Intensity ( )
10
 Rth  1 : BBF = UBF
 Rth
: BBF > UBF
-2
10
10
-5
10
-4
10
-3
10
Intensity ( )
-2
10
Packet transmission is always successful
pB  Rth   pU  Rth 
Extended Protocol
• Time Slot Efficiency vs. Intra-Route Interference
FDR Mode
Intra-route interference
R1
S
R2
R4
R3
slot t
R5
slot t
 Concurrent transmission : S  R1 pair and R4  R5 pair
 Guard hop : R2  R3 hop and R3  R4 hop
BBF Mode
Intra-route interference
R1
S
R2
R4
R3
slot t
R5
slot t
Intra-route interference
 Concurrent transmission : S  R1 pair and R3  R4 pair
 Guard hop : R1  R2 hop and R2  R3 hop
Application of Full Duplex Technique 5
Full Duplex in Cognitive Radio
- TranSensing for Spectrum Sharing System
- Cognitive FDR
TranSensing – System Model (1)
MIMO Quiet Sensing
MIMO TranSensing [Heo2014]
① : Sensing & Transmission Phase
① : Sensing Phase
PU
PU
SU Rx
SU Tx
② : Transmission Phase
SU Tx
︙
SU Rx
PU
SU Tx
︙
︙
SU Rx
With Self-IC
Antenna
Isolation
Digital IC
TranSensing – System Model (2)
Quiet Sensing
Active SU
Sensing
(Mq)
Data Transmission
(M-Mq)
Sensing
(Mq)
Data Transmission
(M-Mq)
N t Antennas
TranSensing
Sensing
Antennas
Data
Antennas
Sensing
(Ms)
Data Transmission
(M-Ms)
Data Transmission
(M)
Sensing
(Ms)
Data Transmission
(M-Ms)
N s Antennas
Data Transmission
(M-Ms)
N d Antennas
– N t transmit antenna, N s sensing antennas,N d data antennas
– Sensing antennas : Ms sensing period Ms and M - Ms data transmission period
– Data antennas: M frame duration is available for data transmission
TranSensing – Avg. SU Achievable Rate
Quiet Sensing
TM ,Q


 Mq 
SNRs 1/2
H
H /2
 1 
Rt HRr H R t 
 1  Pf ( N t , M q ) E log 2 det I N r 
M 
Nt



Time Loss
False Alarm
N r  N t MIMO Capacity
Average SU capacity for CR environment by dividing time resources:
N r  N t MIMO Correlated Channel with N - Ns times.
MIMO
TranSensing
TranSensing
TM ,T


SNRs 1/2
H
H /2
 1  Pf ( N s , M s ) E log 2 det I N r 
Rt HRr H R t 
Nd


False Alarm
N r  N d MIMO Capacity
Average SU capacity for CR environment by dividing spatial resources:
N r  N d MIMO Correlated Channel with N times.
TranSensing – Proposed Algorithm
Objective:
Antenna Selection
Ns
Sensing Duration Control
Sensing
(Ms)
Nd
Correlation based
Data Transmission
(M-Ms)
TranSensing – Simulation Result
• Avg. SU Capacity vs. PU SNR
– Low PU SNR region (< -2dB)
: Correlation based antenna partitioning = full search-based antenna partitioning
– High PU SNR region (> -2 dB)
: Sensing duration control affects the average SU throughput
Cognitive Full-Duplex Relay (1)
• Cognitive FDR vs. Cognitive HDR
Cognitive HDR
Licensed
 Orthogonal transmission of SU and SR
 Interference constraint
Primary Receiver (PR)
bSP PS  I th , bRP PR  I th
Unlicensed
Secondary
Source
vs.
PR
PS
Cognitive FDR
Relay
PS : Source power
Secondary
Destination
PR : Relay power
: Secondary Link
: Interference Link
 Concurrent transmission of SU and SR
 Residual self-interference at SR
 Interference constraint
Cognitive Full-Duplex Relay (2)
• Outage Probability of Secondary User
RT=1b/s/Hz, I th=1
0
10
 Less power consumption at SS and SR
with cognitive FDR
 Residual self-interference at SR
Outage probability of the secondary user
Equal power
allocation (EPA)
Low
interference
-1
10
Reduced Tx Power
Residual SI
Optimal power allocation
(OPA)
-2
10
CogFRN using EPA ( RR=  SD=-30dB)
Employing FDR as SR is beneficial
in spectrum-sharing system
CogHRN
OPA (Analysis,  RR=  SD=-30dB)
-3
10
OPA (Simulation,  RR=  SD=-30dB)
0
5
10
15
20
1SNR [dB]1
 2
2
 RN
 DN
<
Benefit from
Concurrent Tx
25
30
Application of Full Duplex Technique 6
Full Duplex to Improve PHY Layer Secrecy
Full Duplex for PHY Layer Secrecy (1)
• PHY Layer Information Protection: 1 Transmit Ant. at Source
Half-Duplex Nodes
 Cooperative nodes relaying source
information & transmitting jamming
signal [Dong2010]
 Decrease in secrecy capacity due to
orthogonal time allocation to relays
Simultaneous Tx and Rx by FD
Full-Duplex Receiver
Intended
message
in 1st slot
Intended
message
in 1st slot
Jamming
in 1st slot
Relayed
messagesignal
+jamming
signal
nd
in 2 slot
 Transmit jamming signal while
receiving information
 No additional helper or time slot
Full Duplex for PHY Layer Secrecy (2)
• System Model [Zheng2013]
Objective
Optimize
to maximize the
secrecy capacity such that
s
n
Considerations
s : Tx signal of source,
n : jamming signal by destination,
 Destination side
– Received signal : y D  h sd s   H si n  n d
– Linear receiver : r
 Eavesdropper side
– Received signal : y E  h se s  H ed n  n e
– CSI availability at Tx side of D
 Global CSI
 CDI only
– Linear receiver at D ( r ) is
 Fixed receiver
 Optimally designed w.r.t channel
– Receiver at eavesdropper is
 MRC w/o knowing FD of D
 MMSE knowing FD of D
Secrecy Capacity Maximization (1)
• Secrecy Capacity
 Max. Tx rate at which eavesdropper cannot decode any information
 Difference S-D capacity and S-E capacity
• Global CSI & Linear Receiver at D, MRC Receiver at E (1)
Properties
Secrecy Capacity
– Rank  Q*   1, trace  Q*   Pd
(S-D Ch.)
(S-E Ch.)
Q  Pd qq H , q  1
– Non-interference limited
RS keeps increasing with PS
Secrecy Capacity Maximization (2)
• Global CSI & Linear Receiver at D, MRC Receiver at E (2)
Fix Receiver at D
Optimal Receiver at D
 Maxmizes Rx SINR
 Given by
 Independent of Q
 MRC, MMSE, ...
 Change variable as
 Change variable as
RS  q 
 Objective : functions of t
 (Quasi-) concave function of
t
RS  t 
1-Dimensional
Search over t !!!
Secrecy Capacity Maximization (3)
• Global CSI & Optimal Linear Receiver at D, MMSE Receiver at E
 E aware of FD operation of D
 Defense itself from jamming by using MMSE receiver
Secrecy Capacity
 Assuming optimal linear MMSE receiver at D,
Solved by Difference of Convex Functions Method
Secrecy Capacity Maximization (4)
• CDI on Channels to E & Linear Receiver at D
Average Secrecy Capacity
 Channel knowledge at D : perfect
, but only CDI on E
 Suboptimal MRC receiver :

, where
 Approximation : expectation on each individual variables
x
Convex Function of
x
!!!
Simulation Results
• Secrecy Rate vs. Total Tx. Power
• Ergodic Secrecy Rate with CDI Only
HD Case
 No external helper
 Secrecy rate saturate from low SNR
FD Case
vs.
 Multi-ant. at Rx help suppress SI
and generate jamming
 Secrecy rate keep increasing w/o
bound
Application of Full Duplex Technique 7
Wireless Powered Communication Networks
Full Duplex WPCN (1/2)
[Ju2014]
Self-interference (SI)
caused by energy signal
U1
hD ,1
hU ,1
hD ,2
Users
U 2 operating in HD
hU ,2
Hybrid AP
operating in FD
hD , K
hU , K
Energy transfer
Information transfer
UK
• FD H-AP : broadcasts energy in DL and receive information in UL simultaneously
• Pros. & Cons.
Pros.
– Enlarged time for energy harvesting
– More efficient use of time
– Deterministic CW for energy signal
– Makes SI cancellation (SIC) easier
Cons.
– SIC is required at the H-AP
Full Duplex WPCN (1/2)
• Protocol & DL WET : HD-WPCN vs. FD-WPCN
HD-WPCN
FD-WPCN
H-AP
DL
WET
 0T
H-AP
UL
WET
U1
U2
UK
 1T
 2T
 KT
DL
WET
 0T
UL
WET
U1
U2
UK
 1T
 2T
 KT
– Tx. Power of H-AP
– Tx. Power of H-AP
– Harvested Energy by Users
– Harvested Energy by Users
Full Duplex WPCN (2/2)
• UL WIT
– Transmit Power of
:
– Self-Interference Cancellation (SIC) at the H-AP
• RF and analog domain SIC
• Digital domain SIC based on channel estimation
• Additional quantization noise due to finite dynamic range of receive filter of H-AP
– Received Signal at the H-AP after SIC
Desired signal
– Achievable Rate of
Residual SI Quantization
Noise
: energy efficiency
: product channel gain
: SI coefficient
WSR Maximization Problem
• WSR Maximization - General Case
– Non-convex due to non-convexity
in objective function & avg. power
constraint
– Find a suboptimal solution based
on a special case with perfect SIC
• WSR Maximization – Perfect SIC
– Setting
and
– Time & power allocation
time & energy allocation
– Avg, power constraint (non-convex)
sum energy constraint (affine)
– Convex problem
Optimal Time & Energy Allocation
Imperfect SIC
Perfect SIC
Initial Point :
(
)
Update
(e.g. GP method)
Update
(Lagrangian duality)
Until
• Special Case with Infinite PPC
Equivalent for the cases with
perfect & imperfect SIC,
and even for HD-WPCN
A Special Case – Infinite PC
• Power & Time allocation – 10 Users
• Achievable Rate Region – 2 Users
Ppeak  2 Pavg
500
2
400
1.8
300
1.6
0
2
100
Achievable Rate of U (bps/Hz)
Pavg / Ppeak
200
0
0.1
0.2
0.3
0.4
 1*  2*  3*  4*  5*  6*  7*
500
 8*
0.5
0.6
 9*
0.7
0.8 *
 10
Ppeak  5 Pavg
0.9
1
400
300
Pavg / Ppeak
1.4
1.2
1
0.8
FD-WPCN,
Ppeak =inf
FD-WPCN, Ppeak
HD-WPCN, Ppeak
HD-WPCN,
Ppeak =inf
FD-WPCN,
P
Pavg
FD-WPCN, Ppeak
peak =55P
HD-WPCN,
Ppeak =55P
Pavg
HD-WPCN, Ppeak
FD-WPCN,
P

2
P
FD-WPCN, Ppeak
peak = 2P
avg
HD-WPCN,
Ppeak =22P
Pavg
HD-WPCN, Ppeak
0.6
0.4
200
0.2
100
0
0
0.1
   
 2*  4*
*
1
*
3
*
5
0.2
*
6

*
7
0.3

*
8
0.4
0.5

*
9
0.6
0.7
0.8

*
10
0.9
Optimal time & power allocation
optimally exploit the available
multi-user channel diversity
1
0
0
0.5
1
1.5
2
2.5
3
3.5
Achievable Rate of U1 (bps/Hz)
4
Employing FD is more beneficial
as peak power constrain is
more stringent
4.5
Simulation Results
• Avg. Sum-Rate vs. Avg. Tx. Power
• Avg. Sum-Rate vs. Number of Users
11
8
7
Average Sum-Rate (Mbps)
9
Avg. Sum-Rate (Mbps)
8
FD-WPCN,
FD-WPCN,
No No
SI, SI
FD-WPCN,
 80dB
FD-WPCN,
SI, 
eee=bbbbbb
FD-WPCN,
60dB
FD-WPCN,
SI, 
all-60dB
FD-WPCN,
SI, 
all-40dB
FD-WPCN,
40dB
HD-WPCN
HD-WPCN
10
7
6
5
4
3
6
5
    60dB
4
Ppeak  2 Pavg
Pavg  20dBm
3
FD-WPCN,
FD-WPCN,
No No
SI SI
FD-WPCN,
 80dB
FD-WPCN,
SI,eee=bbbbbb
FD-WPCN,

 60dB
data3
data4
FD-WPCN,   40dB
HD-WPCN
data5
2
2
    60dB
0
1
Ppeak  2 Pavg , K  10
1
0
5
10
15
Pavg (dBm)
20
25
With sufficient SIC, FD-WPCN
outperforms HD-WPCN thanks to
more efficient time & power usage
30
0
2
4
6
8
10
12
14
Number of Users (K)
16
18
Employing FD is more beneficial
as number of users increases
thanks multi-user diversity
20
Concluding Remarks
Conclusion
Full-Duplex Technology
– Simultaneous Signal Tx & Rx
– New Way to Use Tx / Rx Signals
– Increased Interference
– Implementation Issues
More Advanced Protocols to Improve Various
Future Wireless Communication Networks
Thank You
Any Question?
References
[Nguyen2013] D. Nguyen, L. –N. Tran, P. Pirinen, and M. Latva-aho, “Precoding for full duplex multiuser MIMO systems: spectral and energy efficiency
maximization,” IEEE Trans. Signal Process., vol. 61, no. 16, pp. 4038-4050, Aug. 2013.
[Arnold2010] O. Arnold, F. Richter, G. Fettweis, and O. Blume, “Power consumption modeling of different base station types in heterogeneous cellular
networks,” in Proc. 19th Future Netw. MobileSummit (ICT Summit ‘10), pp. 1-8, Florence, Italy, June 2010.
[Xu2011] J. Xu, L. Qiu, and C. Yu, “Improving energy efficiency through multimode transmission in the downlink MIMO systems,” Eurasip J. Wireless
Commun. Netw., 2011.
[Singh2011] N. Singh, S. Diggavi, A. Proutiere, B. Radunovic, H. V. Balan, and P . Key, “Efficient and fair MAC for wireless networks with self-interference
cancellation,” in Proc. Int. Symp. Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks (WiOpt), pp. 94–101, Princeton, NJ, May 2011.
[Weber2005] S. P. Weber, X. Yang, J. Andrews, and G. De Veciana, “Transmission capacity of wireless ad hoc networks with outage constraints,” IEEE Trans.
Inform. Theory, vol. 51, no. 12, pp. 4091–4102, 2005.
[Kim2014] D. Kim, S. Park, H. Ju, and D. Hong, “Transmission capacity of full-duplex based two-way ad-hoc networks with ARQ protocol,” To be published
in IEEE Trans. Veh. Technol., Accepted
[Ju2012] H. Ju, S. Lim, D. Kim, H. V. Poor, and D. Hong, “Full duplexity in beamforming-based multi-hop relay networks,” IEEE J. Sel. Areas in Commun.,
vol. 30, no. 8, pp. 1554-1565, Sep. 2012.
[Heo2014] J. Heo, H. Ju, S. Park, E. Kim, and D. Hong, “Simultaneous sensing and transmission in cognitive radio,” To be published in IEEE Trans. Wireless
Commun., Accepted
[Kim2012] H. Kim, S. Lim, H. Wang, and D. Hong, “Optimal power allocation and outage analysis for cognitive full duplex relay systems,” IEEE Trans.
Wireless Commun., vol. 11, no. 10, pp. 4962-4974, Oct. 2012
[Dong2010] L. Dong, Z. Han, A. P. Petropulu, and H. V. Poor, “Improving physical layer security via cooperating relay,” IEEE Trans. Signal Process., vol. 61,
no. 20, pp. 4962-4974, Oct. 2013
[Zheng2013] G. Zheng, I Krikidis, J. Li, A. P. Petropulu, and B. Otterson, “Improving physical layer secrecy using full-duplex jamming receiver,” IEEE Trans.
Signal Process., vol. 61, no. 20, pp. 4962-4974, Oct. 2013
[Ju2014] H. Ju and R. Zhang, “Optimal Resource Allocation in Full-Duplex Wireless Powered Communication Networks,” submitted for publication.
Available online: arXiv:1403.2580
1/--pages
Пожаловаться на содержимое документа