Se ha denunciado esta presentación.
Utilizamos tu perfil de LinkedIn y tus datos de actividad para personalizar los anuncios y mostrarte publicidad más relevante. Puedes cambiar tus preferencias de publicidad en cualquier momento.
1
POPS-OFDM:
Ping-Pong Optimized Pulse Shaping OFDM for
5G Cellular Systems and Beyond
Mohamed SIALA
Ecole Supérieure des ...
Outline
 Motivation of Research Activities on Pulse Shaping for OFDM
 Background on Multi-Carrier Systems
 5G Challenge...
Motivation of Research Activities on Pulse
Shaping for OFDM
3
5G Challenges and Requirements 1/5
4
5GNOW
5G Challenges and Requirements 2/5
5
5GNOW
5G Challenges and Requirements 3/5
6
5GNOW
5G Challenges and Requirements 4/5
7
5GNOW
5G Challenges and Requirements 5/5
 The main drivers for 5G are:
 Tactile Internet: The human tactile sense distinguishe...
4G (LTE-A) Pitfalls
 LTE is tailored to maximize performance by enforcing strict
synchronism and orthogonality 
 Machin...
Need for Non-Orthogonal Waveforms
 Non-orthogonal waveforms on the physical layer will enable:
 Asynchronous MTC traffic...
Workload of Current Mobiles
11
Outer receiver consists of channel decoder and de-interleaver
Projects on 5G
 From 2007 to 2013, the European Union set aside €700 million of
funding (FP7) for research on future netw...
5GNOW Candidate Waveforms
 European Union’s FP7 projects, 5GNOW (5th Generation Non-
Orthogonal Waveforms for Asynchronou...
GFDM: Generalized Frequency Division
Multiplexing
14
UFMC: Universal Filtered MultiCarrier
15
Spectral behavior within a single sub-band
Single PRB compared to OFDM
Background on Multi-Carrier Systems
16
17
History of OFDM
 Late 50’s: Concept of multicarrier without overlapping in frequency
 Late 60’s: Orthogonal multicarr...
18
Applications of OFDM
 Wireless applications :
 Broadcasting for digital terrestrial television (DVB- T, DVB- H)
 Dig...
Classification of Multi-Carrier Systems
19
Multi-carrier systems
Filter Lattice Symbol
Orthogonal
Hexagonal
Biorthogonal
R...
20
Pulse Shaping for OFDM without Guard Interval
 Rectangular pulse shaping filter with duration T :
1/T
t
[0, [( ) rect ...
21
OFDM without Guard Interval
t
f
nT ( 1)n T0 0f 
/mf m T
1 ( 1)/Nf N T  
… …
0 ( )n t
1, ( )N n t 
t
nT ( 1)n T...
Time-Frequency Lattice Layout: OFDM without
Guard Interval (GI)
22
Frequency
Time
1/ uF T
uT T
Area (1/ ) 1u uFT T T  ...
23
Power Spectral Density of OFDM without Guard
Interval 1/2
 Subcarrier spacing : f = 1/T
1/f T 
f
Power Spectral Density of OFDM without Guard
Interval 2/2
24
16N 
0gT 
32N 
64N  128N 
25
Analog Transmitter for OFDM
 Complex implementation: Use of a battery of N costly analog filters
ka
0na
mna
1,N na 
S...
26
Vectorial Equivalent of the OFDM Transmitter
ka
0na
mna
1,N na 
S
/
P
exp( 2 )cj f t
{} 
( )ee t ( )e t
0 ( )n t
(...
27
Analog Receiver for OFDM
 Complex implementation: Use of a battery of N costly analog filters
ˆkaP
/
S
( )t
0exp( 2 )...
28
Vectorial Equivalent of the OFDM Receiver
ˆkaP
/
S
exp( 2 )cj f t
( )r t
0
ˆ na
ˆmna
1,
ˆN na 
*
( ), ( ) ( ) ( )u t...
29
No Interference in the Gaussian Channel –
Orthogonality Contraints
 Modulated signal:
 Symbol amn transported by func...
Balian-Low Theorem for a Time-Frequency
Critical Lattice Density

30
( ) ( )exp( 2 )mn t t nT j mFt   
2 22 2
ˆ( ) (...
31
Behavior of an OFDM System without GI in the
Presence of Time Dispersive Channel
t
f
nT ( 1)n T0 0f 
/mf m T
1 ( 1)/...
32
Interference Suppression: Guard Interval Insertion with
Zero Padding (ZP)
t
f
… …
t
nT ( 1)n T
… …
( 1)n T
1, ( )N n ...
33
Interference Suppression: Guard Interval Insertion with
Zero Padding (ZP)
t
… …
No Inter-Symbol
Interference (ISI)
Pers...
34
Interference Suppression: Guard Interval Insertion with
Cyclic Prefix (CP)
t
… …
t
nT ( 1)n T
… …
( 1)n T
0 ( )n t
T...
Time-Frequency Lattice Layout: OFDM with
Cyclic Prefix (CP)
35
Time
1/ uF T
u gT T T 
Area (1/ )( ) 1 / 1u u g g uFT T ...
Power Spectrum Density of Conventional OFDM
36
16N 
/ 4g uT T
32N 
64N 
128N 
37
Interference Suppression: Guard Interval Insertion with
Cyclic Prefix (CP)
t
… …
t
nT ( 1)n T
… …
( 1)n T
T
gT uT
f
O...
38
Interference Suppression: Guard Interval Insertion with
Cyclic Prefix (CP)
t
… …
t
… …
f
Recovered
orthogonality
t
… …
...
39
OFDM/OQAM with Square Lattice
 The used shaping filter filtre is generally a root-raised cosine filter with
roll-off ...
Time-Frequency Lattice Layout: OFDM/OQAM
with Square Lattice
40
Frequency
Time
1/ uF T
/ 2uT T
Surface (1/ )( / 2) 1/ 2u...
Time-Frequency Lattice Layout: OFDM/OQAM
with Hexagonal Lattice
41
Frequency
Time
3π/4
π/2 0
π/4 3π/4
π/2 0
π/43π/4
π/2 0
...
Waveforms for OFDM/OQAM
42
Linear decreasing in the
logarithmic scale
 Exponential decrease in
time and frequency
Gaussia...
Small-Scale Propagation: Multipath Rayleigh
Fading 1/2
43
Time
F
T
Frequency
Doppler shift
Time delay
min
min Df  
max...
Small-Scale Propagation: Multipath Rayleigh
Fading 2/2
44
Time
F
T
Frequency
Doppler shift
Time delay
: Channel spread
DB
...
Doppler Spread-Delay Spread Balancing 1/3
45
Time
F
T
Frequency
Doppler shift
Time delayDB
mT
Reduction in F & Increase in...
Doppler Spread-Delay Spread Balancing 2/3
46
Time
F
T
Frequency
Doppler shift
Time delayDB
mT
ISI
ICI
ICI
ISI
Reduction in...
Doppler Spread-Delay Spread Balancing 3/3
47
Time
F
T
Frequency
Doppler shift
Time delayDB
mT
ISIISI
ICI
ICI
 Good balanc...
5G Challenges and Requirements
48
Requirements for 5G: Coordinated MultiPoint
(CoMP)
 Joint Processing (JP):
 Coordination between multiple BSs
 MSs are ...
Requirements for 5G: Coordinated MultiPoint
(CoMP) – Overlapping in Time
50
time
At the BSs
MS2MS1
timetime
TDOA  Overlap...
Requirements for 5G: Coordinated MultiPoint
(CoMP) – Overlapping in Frequency
51
MS
frequency
Carrier Frequency Offset 
O...
Requirements for 4G, 5G and DVB-T: MBMS
and SFN
52
time
At the BSs/DVB-T TV Station
Overlapping replicas  Artificial dela...
Requirements for 5G: Sporadic Traffic and Fast
Dormancy 1/4
 2, 3 and 4G systems continuously transmit reference signals ...
Requirements for 5G: Sporadic Traffic and Fast
Dormancy 2/4
 Sporadic access poses a significant challenge to mobile acce...
Requirements for 5G: Sporadic Traffic and Fast
Dormancy 3/4
  Get rid of closed-loop timing control (which costs energy ...
Requirements for 5G: Sporadic Traffic and Fast
Dormancy 4/4
56Nokia Siemens Networks, Understanding Smartphone Behavior in...
Requirements for 5G: Sporadic Traffic and Fast
Dormancy – Relaxed Frequency Synchronization
57
MS2
frequency
Reduced synch...
Requirements for 5G: Sporadic Traffic and Fast
Dormancy – Relaxed Time Synchronization
58
MS2
time
Reduced synchronization...
Requirements for 5G: Asynchronous Signaling in
the Uplink – RACH 1/2
59
MS2MS1
RACH random access
Requirements for 5G: Asynchronous Signaling in
the Uplink – RACH 2/2
60
time
No synchronization overhead  Strong overlapp...
Requirements for 5G: Spectrum Agility and
Carrier Aggregation 1/2
 TV White Spaces (TVWS) exploration can represent a new...
Requirements for 5G: Spectrum Agility and
Carrier Aggregation 2/2
62
OFDM+CP vs. ESM: Loss of efficient of traditional OFD...
Requirements for 5G: Low Latency 1/2
 4G offers latencies of multiple 10 ms between terminal and BS that
originate from r...
Requirements for 5G: Low Latency 2/2
 A 1 ms round-trip time for a typical tactile interaction requires a time
budget of ...
Requirements for 5G: Lower Latency vs Doppler
Spread-Delay Spread Balancing 1/2
65
Time
F
T
Frequency
Doppler shift
Time d...
Requirements for 5G: Lower Latency vs Doppler
Spread-Delay Spread Balancing 2/2
66
Time
F
T
Frequency
Doppler shift
Time
d...
POPS-OFDM
67
POPS-OFDM Categories
68
POPS-OFDM
Continuous DiscreteTime
Optimum
exploration space
2
( ) 2
( )
Practical
exploration spac...
33φ32φ31φ30φ
23φ22φ21φ20φ
13φ12φ11φ10φ
OFDM Time-Frequency Lattice: Transmitter
Side
Time
Frequency
Signal
00 φ φ 01φ 02φ...
30 30a φ
20 20a φ
10 10a φ
00 00a φ
1
0 0
0
Q
m m
m
a


 φ
OFDM Transmitted Signal
Time
Frequency
Signal
21 21a φ
11 11...
Propagation Channel Characteristics: Delay and
Doppler Spreads
Mobile speed

( , )S p 
p
dB : Doppler spread
Doppler spr...
30 30a φ
20 20a φ
10 10a φ
00 00a φ
1
0 0
0
Q
m m
m
a


 φ
OFDM Received Signal
Time
Frequency
Signal
1
0
: Sampled Ver...
Decision variables
: Receiver Prototype Waveform (Vector)ψ
klψ
Subcarrier Index Symbol Index
Frequency shift of  by kF Ti...
Signal-to-Interference and Noise Ratio (SINR)
S
I N
P
SINR
P P


: Average power of the Useful Term
: Average power of t...
Optimization Philosophy
Transmitter Side Receiver Side(0)
φ
(0)
(0)
( , )
( , )
Maximize
1
H
S p
H
S p
SINR
SNR



 
...
Optimization Philosophy
φ
ψ
(0)
φ
(0)
ψ
(1)
φ
(1)
ψ
(2)
φ
76SINR
Equal-SINR curves
(Contour plot of SINR)
SINR maximum
First Optimization Technique
SINR


0
ψ
( , ) ( , )
1
S p S p
SINR
 φ φ
KI ψ KS ψ
Generalized Eigenvalue Problem
Fin...
Second Optimization Technique
( , )
( , )
H
S p
H
S p
SINR



φ
φ
ψ KS ψ
ψ KIN ψ
( , )
H
S p  φ
KIN UΛU
: Unitary Mat...
Signal and Interference Kernel Computation
1/3
1
( , )
0
( )k
K
H
S p nN k p
n k
 



  
   
  
 φ
K Σ...
Signal and Interference Kernel Computation
2/3
φ H
φφ  
1
0
( )k
K
H
k p
k



 Σ φφ
Duration: DT
 DN samples
80
...
Signal and Interference Kernel Computation
3/3
1
0
( )k
K
H
nN k p
n k



 
 
 
 Σ Σ φφ
Matrix shifts accordin...
Numerical Results: Impact of Initialization and
Existence of Local Maxima
82
Local maxima
Numerical Results: Evolution of Transmit and
Receive Pulse Shapes Through the Iterations
83
Iterations: 0, 1, 2, 3, 4, 5, ...
Numerical Results: Doppler Spread-Delay Spread
Balancing
84
Best balancing
Numerical Results: Doppler Spread-Delay Spread
Balancing
85
Numerical Results: Optimized Waveforms
86
Numerical Results: Optimized Waveforms
87
Numerical Results: Performance and Gain in
SINR – Identical Pulse Shape Durations
88
Gain > 5dB
Numerical Results: Performance and Gain in
SINR – Different Pulse Shape Durations
89
Numerical Results: Ventilation of Complexity
Between Transmitter and Receiver
90
PHY delay = (D+D)/2
Numerical Results: Spectrum of One Subcarrier
91
~ 60 dB
Numerical Results: Spectrum of 64 Subcarriers
92
~ 60 dB
Numerical Results: Sensitivity to an Estimation
Error on BdTm
93
Numerical Results: Sensitivity to Synchronization
Errors in Frequency
94
Numerical Results: Sensitivity to Synchronization
Errors in Time
95
38-sample error tolerence
34-sample error tolerence
Conclusion
 We proposed a new and straightforward technique for the
systematic optimization of transmit and receive wavef...
Perspectives
 Extension to OFDM/OQAM
 Extension to multi-pulse OFDM/QAM and OFDM/OQAM
 Extension to single-carrier comm...
98
Thank You for Listening
Mohamed SIALA
Ecole Supérieure des Communications de Tunis (SUP’COM)
12th International Multi-C...
References 1/4
 M. Siala, T. Kurt, and A. Yongaçoglu, “Orthonormalization for Multi-Carrier Transmission,”
Canadian Works...
References 2/4
 A. Ben Salem, M. Siala, and H. Boujemâa, “OFDM systems with hexagonal time-frequency
lattices and well ti...
References 3/4
 R. Ayadi, I. Kammoun, and M. Siala, “Optimization of the pulse shape of OFDM systems Using
the Arrow-Hurw...
References 4/4
 R. Ayadi, I. Kammoun, and M. Siala, “Optimal OFDM Pulse Design, Analysis and
Implementation Over Doubly D...
Próxima SlideShare
Cargando en…5
×

SSD 2015 Presentation, POPS-OFDM: Ping-pong Optimized Pulse Shaping OFDM for 5G Cellular Systems and Beyond

Keynote speech, entitled "POPS-OFDM: Ping-pong Optimized Pulse Shaping OFDM for 5G Cellular Systems and Beyond," at the 12th International Conference on Systems, Signals and Devices (SSD'2015), March 2015, Mahdia, Tunisia

  • Inicia sesión para ver los comentarios

SSD 2015 Presentation, POPS-OFDM: Ping-pong Optimized Pulse Shaping OFDM for 5G Cellular Systems and Beyond

  1. 1. 1 POPS-OFDM: Ping-Pong Optimized Pulse Shaping OFDM for 5G Cellular Systems and Beyond Mohamed SIALA Ecole Supérieure des Communications de Tunis (SUP’COM) 12th International Multi-Conference on Systems, Signals & Devices (SSD’2015) March 16 - 19, 2015 - Mahdia, Tunisia
  2. 2. Outline  Motivation of Research Activities on Pulse Shaping for OFDM  Background on Multi-Carrier Systems  5G Challenges and Requirements  POPS-OFDM  Conclusion and Perspectives 2
  3. 3. Motivation of Research Activities on Pulse Shaping for OFDM 3
  4. 4. 5G Challenges and Requirements 1/5 4 5GNOW
  5. 5. 5G Challenges and Requirements 2/5 5 5GNOW
  6. 6. 5G Challenges and Requirements 3/5 6 5GNOW
  7. 7. 5G Challenges and Requirements 4/5 7 5GNOW
  8. 8. 5G Challenges and Requirements 5/5  The main drivers for 5G are:  Tactile Internet: The human tactile sense distinguishes latencies in the order of 1ms  1ms round trip time requires a time budget on PHY of maximum 100 µs.  Internet of Things (IoT): A scalability problem (>100k MTC nodes in a cell) under cost, coverage, energy (life time) and privacy constraints.  Gigabit Wireless Connectivity: Quick downloads (streaming content with data rates in the order of ~100 Mbit/s)  Download times in the order of ~ 10 Gbit/s.  Fragmented Spectrum and the Spectrum Paradox: Spectrum is scarce and expensive but underutilized. With White Spaces Communication, a 100x better localization is expected. 8
  9. 9. 4G (LTE-A) Pitfalls  LTE is tailored to maximize performance by enforcing strict synchronism and orthogonality   Machine-type communication (MTC) communications requires bulky procedures to ensure strict synchronism  Collaborative schemes (e.g. CoMP) use tremendous efforts to collect gains under strict synchronism and orthogonality  Digital Agenda/Carrier aggregation forces systems to deal with fragmented spectrum 9
  10. 10. Need for Non-Orthogonal Waveforms  Non-orthogonal waveforms on the physical layer will enable:  Asynchronous MTC traffic with drastically reduced signalling and increased life time  The provision of asynchronous coordinated multi-point (CoMP) / Heterogeneous Networking (HetNet)  Implementation of asynchronous carrier aggregation concepts with well frequency localization  A (filtered) multicarrier approach will enable:  The mix of synchronous / asynchronous and orthogonal / non- orthogonal traffic types  The aggregation of non-contiguous spectrum thanks to low out-of- band emissions of the non-orthogonal waveforms 10
  11. 11. Workload of Current Mobiles 11 Outer receiver consists of channel decoder and de-interleaver
  12. 12. Projects on 5G  From 2007 to 2013, the European Union set aside €700 million of funding (FP7) for research on future networks, half of which was reserved for wireless technologies and the development of 4G and beyond-4G technologies.  METIS, 5GNOW, iJOIN, TROPIC, Mobile Cloud Networking, COMBO, MOTO and PHYLAWS are some of the latest EU research projects that address the architecture and functionality needs of 5G networks, representing some €50 million EU investment  European Union’s FP7 projects, PHYDYAS (Duration: 30 months, Start: January 2008, End: October 2010, Total Cost: 4 093 483€), investigated Filter Bank Multi-Carrier (FBMC) and corresponding transceiver functionalities 12
  13. 13. 5GNOW Candidate Waveforms  European Union’s FP7 projects, 5GNOW (5th Generation Non- Orthogonal Waveforms for Asynchronous Signaling), (Start: September 2012, End: February 2015, Total Cost: 3 526 991 €), investigated 4 candidate waveforms:  Generalized frequency division multiplexing (GFDM)  Universal Filtered Multicarrier (UFMC): UFMC applies filtering to subsets of the complete band instead of single subcarriers (GFDM) or the complete band (Filtered OFDM)  Filter Bank MultiCarrier (FBMC)  Bi-orthogonal Frequency Division Multiplexing (BFDM) 13
  14. 14. GFDM: Generalized Frequency Division Multiplexing 14
  15. 15. UFMC: Universal Filtered MultiCarrier 15 Spectral behavior within a single sub-band Single PRB compared to OFDM
  16. 16. Background on Multi-Carrier Systems 16
  17. 17. 17 History of OFDM  Late 50’s: Concept of multicarrier without overlapping in frequency  Late 60’s: Orthogonal multicarrier [Chang66, Salzberg67]  Early 70’s: Use of the Fast Fourier Transform (FFT) [Weinstein & Ebert71] - Concept of Guard Interval (GI)  Early 80’s: Concept of Cyclic Prefix (CP) [Peled & Ruiz80]  Early 90’s: DAB standardization  Late 90’s: Standardization of ADSL, DBV-T and WIFI
  18. 18. 18 Applications of OFDM  Wireless applications :  Broadcasting for digital terrestrial television (DVB- T, DVB- H)  Digital Audio Broadcasting (DAB) and Digital Radio Mondiale (DRM)  802.11a wireless networks (WIFI5) , 802.16 (WiMAX) and HiperLAN/2  New generation radio mobile networks (LTE, LTE-A)  Wireline applications:  ADSL  PLC (Power-Line Communications)
  19. 19. Classification of Multi-Carrier Systems 19 Multi-carrier systems Filter Lattice Symbol Orthogonal Hexagonal Biorthogonal Rectangular Non-orthogonal Staggered Complex Real OFDMw/oGI OFDMw/CP OFDM/QAM OFDM/OQAM OFDMw/ZP Pulse Mono-pulse Multi-pulseUFMC OFDMOQAM: Offset Quadrature Amplitude Modulation CP: Cyclic Prefix, ZP: Zero Padding, GI/ Guard Interval TimeContinuous Discrete
  20. 20. 20 Pulse Shaping for OFDM without Guard Interval  Rectangular pulse shaping filter with duration T : 1/T t [0, [( ) rect ( ) /Tt t T  1 ˆ( ) sinc( )f fT  1/T T … … f Fourier Transform
  21. 21. 21 OFDM without Guard Interval t f nT ( 1)n T0 0f  /mf m T 1 ( 1)/Nf N T   … … 0 ( )n t 1, ( )N n t  t nT ( 1)n T … … ( 1)n T ( 2)n T OFDM symbol between nT and (n+1)T Modulated Signal ee(t) Orthogonal sinusoidal functions ( )mn t
  22. 22. Time-Frequency Lattice Layout: OFDM without Guard Interval (GI) 22 Frequency Time 1/ uF T uT T Area (1/ ) 1u uFT T T   Lattice density 1/ 1FT    Critical density 0 & π/2 0 & π/20 & π/20 & π/2 0 & π/2 0 & π/20 & π/20 & π/2 0 & π/2 0 & π/20 & π/20 & π/2 0 & π/2 0 & π/20 & π/20 & π/2 Inphase and quadrature components
  23. 23. 23 Power Spectral Density of OFDM without Guard Interval 1/2  Subcarrier spacing : f = 1/T 1/f T  f
  24. 24. Power Spectral Density of OFDM without Guard Interval 2/2 24 16N  0gT  32N  64N  128N 
  25. 25. 25 Analog Transmitter for OFDM  Complex implementation: Use of a battery of N costly analog filters ka 0na mna 1,N na  S / P ( )t ( )t ( )t 0exp( 2 )j f t exp( 2 )mj f t 1exp( 2 )Nj f t  exp( 2 )cj f t {}  ( )ee t ( )e t S/P : Serial-to-Parallel Converter 0m m f f T   0 0f  in general
  26. 26. 26 Vectorial Equivalent of the OFDM Transmitter ka 0na mna 1,N na  S / P exp( 2 )cj f t {}  ( )ee t ( )e t 0 ( )n t ( )mn t 1, ( )N n t 
  27. 27. 27 Analog Receiver for OFDM  Complex implementation: Use of a battery of N costly analog filters ˆkaP / S ( )t 0exp( 2 )j f t exp( 2 )mj f t 1exp( 2 )Nj f t  exp( 2 )cj f t ( )x t P/S : Parallel to Serial Converter 0 ˆ na ˆmna 1, ˆN na  ( )t ( )t * ( ) ( )t T t  
  28. 28. 28 Vectorial Equivalent of the OFDM Receiver ˆkaP / S exp( 2 )cj f t ( )r t 0 ˆ na ˆmna 1, ˆN na  * ( ), ( ) ( ) ( )u t v t u t v t dt   0 ( ),n t  ( ),mn t  1, ( ),N n t  
  29. 29. 29 No Interference in the Gaussian Channel – Orthogonality Contraints  Modulated signal:  Symbol amn transported by function :  Absence of Inter-Symbol Interference (ISI) and Inter-Carrier Interference (ICI) equivalent to orthogonality conditions:  Functions {mn(t)} form un orthonormal base of the space of modulated signals 1 0 ( ) ( ) N e mn mnn m e t a t      ( ) ( )exp( 2 )mn mt t nT j f t    * ( ), ( ) ( ) ( )kl mn kl mn km lnt t t t dt        
  30. 30. Balian-Low Theorem for a Time-Frequency Critical Lattice Density  30 ( ) ( )exp( 2 )mn t t nT j mFt    2 22 2 ˆ( ) ( )t t dt or f f df      Fourier transform of ( )t
  31. 31. 31 Behavior of an OFDM System without GI in the Presence of Time Dispersive Channel t f nT ( 1)n T0 0f  /mf m T 1 ( 1)/Nf N T   … … t nT ( 1)n T … … ( 1)n T ( 2)n T Noiseless received signal  ( )c  Channel Inter-Symbol Interference (ISI) Loss of orthogonality! mT0 0 ( )n t 1, ( )N n t  ( )mn t
  32. 32. 32 Interference Suppression: Guard Interval Insertion with Zero Padding (ZP) t f … … t nT ( 1)n T … … ( 1)n T 1, ( )N n t  ( )mn t 0 ( )n t T gTuT  ( )c  Channel mT0 g mT T Guard Interval Zero Padding
  33. 33. 33 Interference Suppression: Guard Interval Insertion with Zero Padding (ZP) t … … No Inter-Symbol Interference (ISI) Persistent loss of orthogonality! t f … … g mT T T
  34. 34. 34 Interference Suppression: Guard Interval Insertion with Cyclic Prefix (CP) t … … t nT ( 1)n T … … ( 1)n T 0 ( )n t T gT uT  ( )c  Channel mT0 f Cyclic Prefix 1, ( )N n t  ( )mn t nT ( 1)n T( 1)n T
  35. 35. Time-Frequency Lattice Layout: OFDM with Cyclic Prefix (CP) 35 Time 1/ uF T u gT T T  Area (1/ )( ) 1 / 1u u g g uFT T T T T T      Lattice density 1/ 1FT    Frequency uTgT Cyclic Prefix Useful part
  36. 36. Power Spectrum Density of Conventional OFDM 36 16N  / 4g uT T 32N  64N  128N 
  37. 37. 37 Interference Suppression: Guard Interval Insertion with Cyclic Prefix (CP) t … … t nT ( 1)n T … … ( 1)n T T gT uT f Overlapping restricted to the Guard Interval 0 ( )n t 1, ( )N n t  ( )mn t
  38. 38. 38 Interference Suppression: Guard Interval Insertion with Cyclic Prefix (CP) t … … t … … f Recovered orthogonality t … … ( 1)n T( 1)n T Cyclic Prefix Suppression 1, ( )N n t  ( )mn t 0 ( )n t nT
  39. 39. 39 OFDM/OQAM with Square Lattice  The used shaping filter filtre is generally a root-raised cosine filter with roll-off  (typically equal to 1)  OFDM with Offset QAM (OQAM) alternately transmit phase and quadrature t 1/f T  QAM Symbol f T / 2T : In-phase component : Quadrature component
  40. 40. Time-Frequency Lattice Layout: OFDM/OQAM with Square Lattice 40 Frequency Time 1/ uF T / 2uT T Surface (1/ )( / 2) 1/ 2u uFT T T   Lattice density 1/ 2FT    Critical density π/2 π/2 π/2 π/2 π/2 π/2 π/2 π/2 π/2 π/2 π/2 π/2 π/2 π/2 π/2 π/2 Quadrature componentInphase component
  41. 41. Time-Frequency Lattice Layout: OFDM/OQAM with Hexagonal Lattice 41 Frequency Time 3π/4 π/2 0 π/4 3π/4 π/2 0 π/43π/4 π/2 0 π/4 3π/4 π/2 0 π/4 3π/4 π/2 0 π/43π/4 π/2 0 π/4 Lattice density 1   Critical density [1] M. Siala, “Novel OFDM/OQAM system with hexagonal time-frequency lattice,” Third International Symposium on Image/Video Communications over fixed and mobile networks (ISIVC’06), Hammamet, Tunisia, September 2006. [2] M. Siala and A. Yongaçoglu, “Prototype waveform optimization for an OFDM/OQAM system with hexagonal time-frequency lattice structure,” 9th International Symposium on Signal Processing and its Applications (ISSPA’07), Sharjah, United Arab Emirates, February 2007.
  42. 42. Waveforms for OFDM/OQAM 42 Linear decreasing in the logarithmic scale  Exponential decrease in time and frequency Gaussian waveform Hexagonal lattice waveform [1] M. Siala, “Novel OFDM/OQAM system with hexagonal time-frequency lattice,” Third International Symposium on Image/Video Communications over fixed and mobile networks (ISIVC’06), Hammamet, Tunisia, September 2006.
  43. 43. Small-Scale Propagation: Multipath Rayleigh Fading 1/2 43 Time F T Frequency Doppler shift Time delay min min Df   max Df   max : Doppler spread DB mT : Delay spread DB mT Scattering function
  44. 44. Small-Scale Propagation: Multipath Rayleigh Fading 2/2 44 Time F T Frequency Doppler shift Time delay : Channel spread DB mT D mB T  (Diffuse) Scattering function ISIISI ICI ICI ISI: Inter-Symbol Interference ICI: Inter-Carrier Interference
  45. 45. Doppler Spread-Delay Spread Balancing 1/3 45 Time F T Frequency Doppler shift Time delayDB mT Reduction in F & Increase in T  Substantial increase in ICI  Global increase in ICI+ISI ISIISI ICI ICI
  46. 46. Doppler Spread-Delay Spread Balancing 2/3 46 Time F T Frequency Doppler shift Time delayDB mT ISI ICI ICI ISI Reduction in T & Increase in F  Substantial increase in ISI  Global increase in ICI+ISI
  47. 47. Doppler Spread-Delay Spread Balancing 3/3 47 Time F T Frequency Doppler shift Time delayDB mT ISIISI ICI ICI  Good balancing between T and F  Global reduction in ICI+ISI mD TB F T 
  48. 48. 5G Challenges and Requirements 48
  49. 49. Requirements for 5G: Coordinated MultiPoint (CoMP)  Joint Processing (JP):  Coordination between multiple BSs  MSs are simultaneously transmitting or receiving to or from multiple BSs  Coordinated Scheduling/Coordinated Beamforming (CS/CB):  Coordination between multiple BSs  MSs are transmitting or receiving to or from a single transmission or reception BS 49
  50. 50. Requirements for 5G: Coordinated MultiPoint (CoMP) – Overlapping in Time 50 time At the BSs MS2MS1 timetime TDOA  Overlapping in time  Artificial delay spread  Inter-Symbol Interference At MS2 At MS1 TDOA: Time Difference of Arrival Applicable even for fully time synchronous BSs
  51. 51. Requirements for 5G: Coordinated MultiPoint (CoMP) – Overlapping in Frequency 51 MS frequency Carrier Frequency Offset  Overlapping in frequency  Artificial Doppler spread  Inter-Carrier Interference (ICI) At MS From BS1 frequency frequency From BS2 From BS3 Applicable only for not fully frequency Synchronous BSs
  52. 52. Requirements for 4G, 5G and DVB-T: MBMS and SFN 52 time At the BSs/DVB-T TV Station Overlapping replicas  Artificial delay spread  Interference time At the TV Set (SFN) At the MS (MBMS) SFN: Single Frequency Network MBMS: Multimedia Broadcast Multicast Service
  53. 53. Requirements for 5G: Sporadic Traffic and Fast Dormancy 1/4  2, 3 and 4G systems continuously transmit reference signals and broadcast system information that is used by terminals as they move across cells   The more signaling the cellular standard requires the more complex and power-hungry will be the devices  With denser deployment and more network nodes (MTC), such “always-on” transmissions are not attractive from an interference and energy consumption perspective   Maximizing the devices’ sleep opportunities, through sporadic access, can minimize energy consumption, leading to long battery life 53
  54. 54. Requirements for 5G: Sporadic Traffic and Fast Dormancy 2/4  Sporadic access poses a significant challenge to mobile access networks due to fast dormancy:  Fast dormancy is used to save battery power: The mobile breaks ties to the network as soon as a data piece is delivered  When the mobile has to deliver more pieces of data it will always go through the complete synchronization procedure again  This can happen several hundred times a day, resulting in significant control signaling growth and network congestion threat   It is desirable to achieve “zero-overhead” communications by providing channel access with minimal signaling 54
  55. 55. Requirements for 5G: Sporadic Traffic and Fast Dormancy 3/4   Get rid of closed-loop timing control (which costs energy and signaling overhead, being undesirable for MTC) and use open loop timing control mechanisms: The device uses the downlink pilot signals by the BS for a rough synchronization 55
  56. 56. Requirements for 5G: Sporadic Traffic and Fast Dormancy 4/4 56Nokia Siemens Networks, Understanding Smartphone Behavior in the Network, White Paper, 2011, [Available: http://www.nokiasiemensnetworks.com/sites/default/files Comparisons of Data and Signaling Traffic
  57. 57. Requirements for 5G: Sporadic Traffic and Fast Dormancy – Relaxed Frequency Synchronization 57 MS2 frequency Reduced synchronization overhead  Relaxed frequency synchronization  Carrier Frequency Offset  Overlapping in frequency  Inter-user interference in frequency From MS1 frequency From MS2 MS1 MS3 frequency From MS3 At BS frequency Inter-user interference Unaligned carrier frequencies
  58. 58. Requirements for 5G: Sporadic Traffic and Fast Dormancy – Relaxed Time Synchronization 58 MS2 time Reduced synchronization overhead  Relaxed time synchronization  Overlapping in time  Inter-user interference in time From MS1 time From MS2 MS1 At BS time Inter-user interference
  59. 59. Requirements for 5G: Asynchronous Signaling in the Uplink – RACH 1/2 59 MS2MS1 RACH random access
  60. 60. Requirements for 5G: Asynchronous Signaling in the Uplink – RACH 2/2 60 time No synchronization overhead  Strong overlapping in time  Inter-user interference in time To/from BS time To/from MS1 To/from MS2 Inter-Burst interference time Synchronization channel RACH burst from MS2 RACH burst from MS1 Propagation delay to MS1 Propagation delay to MS2
  61. 61. Requirements for 5G: Spectrum Agility and Carrier Aggregation 1/2  TV White Spaces (TVWS) exploration can represent a new niche markets if it overcome, with spectrum agility, the rigorous implementation requirements of low out of band radiation for protection of legacy systems  The LTE-A waveform imposes generous guard bands to satisfy spectral mask requirements which either severely deteriorate spectral efficiency or even prevent band usage at all  5G will address carrier aggregation by implementing non-orthogonal waveform, with low out-of-band emissions, in order not to interfere with other legacy systems and tight spectral masks 61
  62. 62. Requirements for 5G: Spectrum Agility and Carrier Aggregation 2/2 62 OFDM+CP vs. ESM: Loss of efficient of traditional OFDM with CP to fit in an ESM (Emission Spectrum Mask) due its non-negligible side lobes
  63. 63. Requirements for 5G: Low Latency 1/2  4G offers latencies of multiple 10 ms between terminal and BS that originate from resource scheduling, frame processing, retransmission procedures, and so on.  The access latency offered by LTE is not sufficient for latency-critical applications, such as tactile internet (motivated by the tactile sense of the human body, which can distinguish latencies on the order of 1 ms accuracy), traffic safety and infrastructure protection.  To ensure support for such mission-critical MTC applications, next- generation wireless access should allow for latencies on the order of 1 ms or less. 63
  64. 64. Requirements for 5G: Low Latency 2/2  A 1 ms round-trip time for a typical tactile interaction requires a time budget of maximum 100 µs on the physical layer   Far shorter than LTE-A allow, missing the target by nearly two orders of magnitude   Clear motivation for an innovative and disruptive redesign of the PHY layer  Lower latency over the radio link can be achieved by reducing transmission-time intervals and widening the bandwidth of radio resource blocks in which a specific amount of data is transmitted 64
  65. 65. Requirements for 5G: Lower Latency vs Doppler Spread-Delay Spread Balancing 1/2 65 Time F T Frequency Doppler shift Time delayDB mT Reduced global ICI+ISI  Good balancing between T and F  Increased Latency ICI ICI ISIISI Processing Time at the Rx2 mT min Contribution of the PHY to the latency
  66. 66. Requirements for 5G: Lower Latency vs Doppler Spread-Delay Spread Balancing 2/2 66 Time F T Frequency Doppler shift Time delay DB mT Decreased Latency  Bad balancing between T and F  Increased global ICI+ISI ISIISI Processing Time at the Rx 2 mTmin Contribution of the PHY to the latency  ICI ICI
  67. 67. POPS-OFDM 67
  68. 68. POPS-OFDM Categories 68 POPS-OFDM Continuous DiscreteTime Optimum exploration space 2 ( ) 2 ( ) Practical exploration space 1 0Vect({ ( )} )N k kt   1 0Vect({ ( )} )N k kt   2 ( )I  0{ ( )}k kt   0{ ( )}k kt   : Hermite functions : Prolate Spheroidal Wave Functions (PSWF) To be explored next
  69. 69. 33φ32φ31φ30φ 23φ22φ21φ20φ 13φ12φ11φ10φ OFDM Time-Frequency Lattice: Transmitter Side Time Frequency Signal 00 φ φ 01φ 02φ 03φ Time Shift by TTime Shift by 2TTime Shift by 3T Frequency Shift by F Frequency Shift by 2F Frequency Shift by 3F Symbol Period T = Symbol Spacing Symbol Bandwidth F = Subcarrier Spacing : Transmitter Prototype Waveform (Vector)φ : (OFDM) Symbol PeriodT : Subcarrier SpacingF mnφ Subcarrier Index Symbol Index Frequency shift of  by mF Time shift of  by nT69
  70. 70. 30 30a φ 20 20a φ 10 10a φ 00 00a φ 1 0 0 0 Q m m m a    φ OFDM Transmitted Signal Time Frequency Signal 21 21a φ 11 11a φ 01 01a φ 31 31a φ 1 1 1 0 Q m m m a    φ 1 0 : Sampled Version of the Transmitted OFDM Signal Q mn mn n m a    e φ 1 2 2 0 Q m m m a    φ 32 32a φ 22 22a φ 12 12a φ 02 02a φ 1 3 3 0 Q m m m a    φ 33 33a φ 23 23a φ 13 13a φ 03 03a φ SubcarriersQ 70
  71. 71. Propagation Channel Characteristics: Delay and Doppler Spreads Mobile speed  ( , )S p  p dB : Doppler spread Doppler spread spectrum : Discrete time delay : Doppler frequency shift ( , )S p  : Channel scattering function : Discrete time delay spreadmT 71
  72. 72. 30 30a φ 20 20a φ 10 10a φ 00 00a φ 1 0 0 0 Q m m m a    φ OFDM Received Signal Time Frequency Signal 1 0 : Sampled Version of the Received OFDM Signal Q mn mn n m a    r φ n 21 21a φ 11 11a φ 01 01a φ 31 31a φ 1 1 1 0 Q m m m a    φ : Additive White Gaussian Noisen : Channel distorted version ofmn mnφ φ 72 ISIICI
  73. 73. Decision variables : Receiver Prototype Waveform (Vector)ψ klψ Subcarrier Index Symbol Index Frequency shift of  by kF Time shift of  by lT H kl kl  ψ r : Decision variable on kla ( , ) ( , ) Noise TermUseful Term Interference Term H H H kl kl kl kl mn kl mn kl m n k l a a     ψ φ ψ φ ψ n 73
  74. 74. Signal-to-Interference and Noise Ratio (SINR) S I N P SINR P P   : Average power of the Useful Term : Average power of the Interference Term : Average power of the Noise Term S I N P P P ( , ) ( , ) 1 H S p H S p SINR SNR          φ φ ψ KS ψ ψ KI I ψ : Ratio of two definite positive quadratic forms on  for a given  ( , ) ( , ) 1 H S p H S p SINR SNR              ψ ψ φ KS φ φ KI I φ : Ratio of two definite positive quadratic forms on  for a given  0 : Signal to Noise Ratio E SNR N  74
  75. 75. Optimization Philosophy Transmitter Side Receiver Side(0) φ (0) (0) ( , ) ( , ) Maximize 1 H S p H S p SINR SNR          φ φ ψ KS ψ ψ KI I ψ (0) ψ (0) (0) ( , ) ( , ) Maximize 1 H S p H S p SINR SNR              ψ ψ φ KS φ φ KI I φ (1) φ (1) (1) ( , ) ( , ) Maximize 1 H S p H S p SINR SNR          φ φ ψ KS ψ ψ KI I ψ (1) ψ (1) (1) ( , ) ( , ) Maximize 1 H S p H S p SINR SNR              ψ ψ φ KS φ φ KI I φ (2) φ 75
  76. 76. Optimization Philosophy φ ψ (0) φ (0) ψ (1) φ (1) ψ (2) φ 76SINR Equal-SINR curves (Contour plot of SINR) SINR maximum
  77. 77. First Optimization Technique SINR   0 ψ ( , ) ( , ) 1 S p S p SINR  φ φ KI ψ KS ψ Generalized Eigenvalue Problem Find the eigenvector with the smallest eigenvalue SINR   0 φ ( , ) ( , ) 1 S p S p SINR     ψ ψ KI φ KS φ 77
  78. 78. Second Optimization Technique ( , ) ( , ) H S p H S p SINR    φ φ ψ KS ψ ψ KIN ψ ( , ) H S p  φ KIN UΛU : Unitary Matrix : Diagonal Positive Matrix U Λ ( , ) H H H H S p   φ ψ KIN ψ ψ UΛU ψ u u 1/2 H u Λ U ψ H H SINR  u Φu u u 1/2 1/2 ( , ) H S p     φ Φ Λ U KS UΛ maxFind the eigenvector of with maximum eigenvalueu Φ 1/2 max 1/2 max opt    UΛ u ψ UΛ u 78
  79. 79. Signal and Interference Kernel Computation 1/3 1 ( , ) 0 ( )k K H S p nN k p n k                 φ K Σ Σ φφ Ω 0 ( ( )) if ( )mod 0 0 else D s pq QJ B T p q p q Q        1 0 ( , ) 0 ( )k K H S p k p k            φ K Σ φφ Π 0 ( ( ))pq D sJ B T p q   ( , ) 0 ( , )S p S p φ φ KS K ( , ) ( , ) 0 ( , )S p S p S p   φ φ φ KI K K Π Q Ω Dependence on channel Doppler; Computed once DN Q 79
  80. 80. Signal and Interference Kernel Computation 2/3 φ H φφ   1 0 ( )k K H k p k     Σ φφ Duration: DT  DN samples 80 Matrix shifts according to The multipath power profile
  81. 81. Signal and Interference Kernel Computation 3/3 1 0 ( )k K H nN k p n k           Σ Σ φφ Matrix shifts according to the normalized symbol duration N 81
  82. 82. Numerical Results: Impact of Initialization and Existence of Local Maxima 82 Local maxima
  83. 83. Numerical Results: Evolution of Transmit and Receive Pulse Shapes Through the Iterations 83 Iterations: 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10,…,20,…,30,…,100 φ ψ Initialization: Gaussian pulse
  84. 84. Numerical Results: Doppler Spread-Delay Spread Balancing 84 Best balancing
  85. 85. Numerical Results: Doppler Spread-Delay Spread Balancing 85
  86. 86. Numerical Results: Optimized Waveforms 86
  87. 87. Numerical Results: Optimized Waveforms 87
  88. 88. Numerical Results: Performance and Gain in SINR – Identical Pulse Shape Durations 88 Gain > 5dB
  89. 89. Numerical Results: Performance and Gain in SINR – Different Pulse Shape Durations 89
  90. 90. Numerical Results: Ventilation of Complexity Between Transmitter and Receiver 90 PHY delay = (D+D)/2
  91. 91. Numerical Results: Spectrum of One Subcarrier 91 ~ 60 dB
  92. 92. Numerical Results: Spectrum of 64 Subcarriers 92 ~ 60 dB
  93. 93. Numerical Results: Sensitivity to an Estimation Error on BdTm 93
  94. 94. Numerical Results: Sensitivity to Synchronization Errors in Frequency 94
  95. 95. Numerical Results: Sensitivity to Synchronization Errors in Time 95 38-sample error tolerence 34-sample error tolerence
  96. 96. Conclusion  We proposed a new and straightforward technique for the systematic optimization of transmit and receive waveforms for OFDM/FBMC/GFDM systems  Increased SINR  6 orders of magnitude reduction in out-of-band emissions  Robustness to synchronization errors 96
  97. 97. Perspectives  Extension to OFDM/OQAM  Extension to multi-pulse OFDM/QAM and OFDM/OQAM  Extension to single-carrier communications  Extension to underwater acoustic communications  OFDM pulse shapes optimized for partial equalization  OFDM tolerant to bursty communications with relaxed synchronization  OFDM pulse shapes optimized for carrier aggregation and reduced out-of band emissions  OFDM pulse shapes optimized for further OFDM pulse shapes optimized for a lower latency  Optimization of RADAR pulses 97
  98. 98. 98 Thank You for Listening Mohamed SIALA Ecole Supérieure des Communications de Tunis (SUP’COM) 12th International Multi-Conference on Systems, Signals & Devices (SSD’2015) March 16 - 19, 2015 - Mahdia, Tunisia
  99. 99. References 1/4  M. Siala, T. Kurt, and A. Yongaçoglu, “Orthonormalization for Multi-Carrier Transmission,” Canadian Workshop on Information Theory 2005 (CWIT’05), Montreal, Quebec, Canada, June 2005.  T. Kurt, M. Siala, and A. Yongaçoglu, “Multi-Carrier Signal Shaping Employing Hermite Functions,” European Signal Processing Conference 2005 (EUSIPCO’05), Antalya, Turkey, September 2005.  N. Debbabi, M. Siala, and H. Boujemâa, “Optimization of the OFDM Prototype Waveform for Highly Time and Frequency Dispersive Channels Through a Maximization of the SIR,” 12th IEEE International Conference on Electronics, Circuits and Systems 2005 (ICECS’05), Gammarth, Tunisia, December 2005.  A. Ben Salem, M. Siala, and H. Boujemâa, “Performance Comparison of OFDM and OFDM/OQAM Systems Operating in Highly Time and Frequency Dispersive Radio-Mobile Channels,” 12th IEEE International Conference on Electronics, Circuits and Systems 2005 (ICECS’05), Gammarth, Tunisia, December 2005.  M. Siala, T. Kurt, and A. Yongaçoglu, “A Unified Framework for the Construction of OFDM/OQAM Systems,” 12th IEEE International Conference on Electronics, Circuits and Systems 2005 (ICECS’05), Gammarth, Tunisia, December 2005. 99
  100. 100. References 2/4  A. Ben Salem, M. Siala, and H. Boujemâa, “OFDM systems with hexagonal time-frequency lattices and well time frequency localized prototype functions,” Third International Symposium on Image/Video Communications over fixed and mobile networks 2006 (ISIVC’06), Hammamet, Tunisia, September 2006.  M. Siala, “Novel OFDM/OQAM system with hexagonal time-frequency lattice,” Third International Symposium on Image/Video Communications over fixed and mobile networks (ISIVC’06), Hammamet, Tunisia, September 2006.  I. Trigui, M. Siala, and H. Boujemâa, “Optimized pulse shaping for OFDM multi-user communications over doubly dispersive channels,” 9th International Symposium on Signal Processing and its Applications (ISSPA’07), Sharjah, United Arab Emirates, February 2007.  M. Siala and A. Yongaçoglu, “Prototype waveform optimization for an OFDM/OQAM system with hexagonal time-frequency lattice structure,” 9th International Symposium on Signal Processing and its Applications (ISSPA’07), Sharjah, United Arab Emirates, February 2007.  I. Trigui, M. Siala, S. Affes and A. Stephenne, “SIR Optimized Hermite-Based Pulses for BFDM Systems in Doubly Dispersive Channels,” International Symposium on Signals, Systems and Electronics (ISSSE’07), Montreal, Quebec, Canada, July 2007. 100
  101. 101. References 3/4  R. Ayadi, I. Kammoun, and M. Siala, “Optimization of the pulse shape of OFDM systems Using the Arrow-Hurwicz Algorithm,” 4th International Symposium on Wireless Communication Systems (ISWCS’07), Trondheim, Norway, October 2007.  R. Ayadi, M. Siala, and I. Kammoun, “Transmit/receive pulse-shaping design in BFDM systems over time-frequency dispersive AWGN channel,” IEEE International Conference on Signal Processing and Communications (ICSPC’07), Dubai, United Arab Emirates, November 2007.  I. Trigui, M. Siala, S. Affes, A. Stephenne, and H. Boujemaa, “Optimum Pulse Shaping for OFDM/BFDM Systems Operating in Time Varying Multi-Path Channels,” IEEE Global Telecommunications Conference (GLOBECOM’07), Washington DC, USA, November 2007.  M. Bellili, M. Siala, and L. Ben Hadj Slama, “Pulse design for maximizing SIR in partially equalized OFDM/BFDM systems,” IEEE 19th International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC’08), Cannes, France, September 2008.  M. Bellili, L. Ben Hadj Slama, and M. Siala, “Multi-pulse/single-pulse design for maximizing SIR in partially equalized OFDM systems over highly dispersive channels,” 16th IEEE International Conference on Electronics, Circuits, and Systems, 2009 (ICECS 2009), Hammamet, Tunisia, December 2009. 101
  102. 102. References 4/4  R. Ayadi, I. Kammoun, and M. Siala, “Optimal OFDM Pulse Design, Analysis and Implementation Over Doubly Dispersive Channel,” 21st European Signal Processing Conference (EUSIPCO 2013), Marrakech, Morocco, September 9-13, 2013.  M. Siala, F. Abdelkefi and Z. Hraiech, “Novel Algorithms for Optimal Waveforms Design in Multicarrier Systems,” IEEE Wireless Communications and Networking Conference (WCNC’2014), Istanbul, Turkey, April 2014.  Z. Hraiech, M. Siala, and F. Abdelkefi, “Numerical Characterization for Optimal Designed Waveform to Multicarrier Systems in 5G,” 22nd European Signal Processing Conference 2014 (EUSIPCO 2014), Lisbon, Portugal, 1-5 September 2014.  Z. Hraiech, F. Abdelkefi, and M. Siala, “POPS-OFDM: Ping-pong Optimized Pulse Shaping- OFDM for 5G systems,” accepted at IEEE International Conference on Communications (ICC’15), London, UK, June 2015.  Z. Hraiech, F. Abdelkefi, and M. Siala, “POPS-OFDM: Ping-pong Optimized Pulse Shaping- OFDM for 5G systems,” Accepted at IEEE Vehicular Technology Conference – Spring 2015 (VTC’S15), Glasgow, Scotland, May 2015. 102

×