faster-than-nyquist transmission for wireless and optical
TRANSCRIPT
Lutz LampeUniversity of British Columbia, Canada
Faster-than-Nyquist transmission for wireless and optical fibre communication
(FTN) OR (Faster than Nyquist) OR (time-frequency packing) OR (super Nyquist)
0
0.5
1
1.5
2
2.5
3
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
2011
2013
2015
2017
What is faster than Nyquist?What is faster than Nyquist?
More bits per second per Hertz, i.e., higher spectral efficiency
Why fast (-er than Nyquist)?©
Sca
ling
Opt
ical F
iber
Net
wor
ks: C
halle
nges
and
Sol
utio
nsPe
ter J
. Winz
er,
Opt
ics a
nd P
hoto
nics N
ews V
ol. 2
6, Iss
ue 3
, pp.
28-3
5 (2
015)
http
s://d
oi.o
rg/1
0.136
4/O
PN.26
.3.00
0028
Why fast (-er than Nyquist)?
Cisco Visual Networking Index: Forecast and Methodology, 2016–2021
0
50000
100000
150000
200000
250000
300000
2016 2017 2018 2019 2020 2021
Global IP traffic in PB per month
© Scaling Optical Fiber Networks: Challenges and SolutionsPeter J. Winzer, Optics and Photonics News Vol. 26, Issue 3, pp. 28-35 (2015)https://doi.org/10.1364/OPN.26.3.000028
CAGR 24%
Why fast (-er than Nyquist)?
VNI Mobile Forecast Highlights, 2016-2021 IEEE Signal Processing Magazine, 2014
Why fast (-er than Nyquist)?
© Ericsson, “Microwave towards 2020,” White paper, September 2015
. © Huawei, “From today to tomorrow,” White paper, February 2016.
Faster-than-Nyquist transmission for wireless and optical fibre communication
I. Concept of FTN
II. Pragmatic approaches
III. Two case studies: some numerical results
Nyquist Signalling
s(t) =Pna[n]h(t� nT )
0 1 2 3 4 5 6-2
-1.5
-1
-0.5
0
0.5
1
1.5
t/T �!
+1 +1-1 -1 -1 +1 +1
AWGN, matched filterr(t) =
Pna[n]g(t� nT ) + w(t)
r[n] = a[n] + w[n]
Sampling Nyquist pulse
Achievable rate*:
*complex baseband signal
CN =
1T log2
⇣1 +
PN0/T
⌘
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 10
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
fT �!
|H(f)|2
/T�!
Observation: rate isindependent of pulse shape
Nyquist Signalling
Achievable rate*:
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 10
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
fT �!
C =
1R�1
log2
⇣1 +
PN0
|H(f)|2⌘df
=
W/2R
�W/2
log2
⇣1 +
PN0
|H(f)|2⌘df
W� 1T
> 1T log2
⇣1 +
PN0/T
⌘= CN
*complex baseband signal
|H(f)|2
/T�!
Not Nyquist Signalling
s(t) =Pna[n]h(t� n⌧T ) 0 < ⌧ 1
Rate increases by a factor ⌧
P = �2a
⌧TPower is scaled:
Faster-than-Nyquist Signalling
0 5 10 15 20 25 30 35 400
2
4
6
8
10
12
14
16
18
bit/T
s�!
SNR = P · T/N0 �!
CN
C ⌧ = {0.7, 0.8, 0.9, 1}
CFTN
CFTN =
1/(2⌧T )R
�1/(2⌧T )
log2
1 +
PN0
1Pk=�1
��H�f +
k⌧T
���2!df
⌧=1/WT= C
Observation: capacity increases with decreasing 𝜏
⌧=1/WT> CN
Faster-than-Nyquist Capacity
CFTN =
1/(2⌧T )R
�1/(2⌧T )
log2
1 +
PN0
1Pk=�1
��H�f +
k⌧T
���2!df
⌧=1/WT= C
⌧ = {0.7, 0.8, 0.9, 1}
bit/T
s�!
0 5 10 15 20 25 300
2
4
6
8
10
12
14
16
18
SNR = Eb/N0 �!
Observation: capacity increases with decreasing 𝜏
CN
C
⌧=1/WT> CN
Faster-than-Nyquist Capacity
⌧ = {0.7, 0.8, 0.9, 1}
CFTN
-0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
-1.5
-1
-0.5
0
0.5
1
1.5
-0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
-1.5
-1
-0.5
0
0.5
1
1.5
t/T �!
+1 +1-1 -1 -1
Observation: loss in eye opening …
Nyquist
FTN
t/T �!
Faster-than-Nyquist Distance
t/T �!
… but minimum distance retained
-0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
-1.5
-1
-0.5
0
0.5
1
1.5
-0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
-1.5
-1
-0.5
0
0.5
1
1.5
t/T �!
+1 +1-1 -1 -1
Faster-than-Nyquist Distance
1
2
3
4
5
6
7
8
SNR = Eb/N0 �!
bit/(s·H
z)�!
�⌧
2 dB
*absolute numbers need to be verified
2PAM/4QAM
4PAM/16QAM
8PAM/64QAMSince dmin does not change, spectral efficiency increases for fixed Eb/N0 *
beyond Mazo limit, trade-off between spectral and power efficiency
Faster-than-Nyquist Mazo Limit
Faster-than-Nyquist transmission for wireless and optical fibre communication
I. Concept of FTN
II. Pragmatic approaches
III. Two case studies: some numerical results
Faster-than-Nyquist ISI
-5 -4 -3 -2 -1 0 1 2 3 4 5
-0.2
0
0.2
0.4
0.6
0.8
1
Intersymbol-interference (ISI) channel
n-k ⟶
g(n-
k) ⟶
r[n] =X
k
a[k]g(n� k) + w[n]
Faster-than-Nyquist Equalization Method Pro Con
MLSE performance optimal complexity polynomial in constellation size and exponential in L
RSSEM-algorithm
close to optimal performance with reduced complexity relatively high complexity
DFE complexity independent of constellation size and about
linear in L
performance loss
Linear
FDE reduce complexity further through filtering via FFT possibly use of cyclic prefix
Relaxed combinatorial optimization
complexity independent of constellation size and
polynomial in L
block based and complexity polynomial in block size,
requires random trials
Faster-than-Nyquist Coding
FEC encoder
FTNtransmission
FTNequalization
FEC decoder
Turbo equalization
Channelsoft output
Faster-than-Nyquist Prequalization LDPC
Enc
oder
Inte
rleav
er
QA
M
Map
per RRC
2𝜏𝑇
Opt
ical
Fro
nt-e
nd
DA
C
SSMF
Coh
eren
t Rx.
2x2
MIM
O B
utte
rfly
PM
D E
q.
Dem
appe
r
𝑣′ LDPC
Dec
oder
𝑎
Car
rier R
ecov
ery
FTN
Pr
e-eq
ualiz
er
𝑟
QAMMapper
FTNPre-equalizer
DACRRC Pulse shape
ℎ(𝑡)
QA
M
Map
per
FTN
Pr
e-eq
ualiz
er
DA
C AD
CA
DC
WM
F+
CD
Com
p.
Dem
appe
r De-
inte
rleav
er
𝑣′
𝑎 𝑟
DataIn
DataOutTransmitter Receiver
WF𝐹(𝑧)
WM
F+
CD
Com
p.
SoftDemapper
Rx Matched Filterℎ∗(−𝑡)
AWGN𝜏𝑇-Sampling
Transmitter
Receiver
𝑎 𝑠𝑟
𝑣′
RRC2𝜏𝑇
Interleaver
De-interleaver
FECEncoder
FECDecoder
Data In
Data Out
Tomlison-Harashima Precoding
Linear Precoding
© M. Jana, A. Mehdra, L. Lampe, J. Mitra "Pre-equalized Faster-than-Nyquist Transmission," IEEE Transactions on Communications, vol. 65, no. 10, pp. 4406-4418, October 2017.
Faster-than-Nyquist Equalization Method Pro Con
MLSE performance optimal complexity polynomial in constellation size and exponential in L
RSSEM-algorithm
close to optimal performance with reduced complexity relatively high complexity
DFE complexity independent of constellation size and about
linear in L
performance loss Linear
FDE reduce complexity further through filtering via FFT possibly use of cyclic prefix
Relaxed combinatorial optimization
complexity independent of constellation size and
polynomial in L
block based and complexity polynomial in block size,
requires random trials
Precoding low complexity, high performance
application to limited FTN acceleration range
Faster-than-Nyquist Complications
SynchronizationNo excess bandwidth for timing synchronization
Varying signal amplitude complicates carrier-phase estimation
Channel estimation
Effective constellation changes due to FTN signalling is a challenge for pilot-based and blind channel
estimation
Peak-to-average power ratio Generally increases
Faster-than-Nyquist transmission for wireless and optical fibre communication
I. Concept of FTN
II. Pragmatic approaches
III. Two case studies: some numerical results
Optical Fibre CommunicationLD
PC E
ncod
er
Inte
rleav
er
QA
M
Map
per RRC
2"#
Opt
ical
Fro
nt-e
nd
DA
C
SSMF
Coh
eren
t Rx.
2x2
MIM
O B
utte
rfly
PMD
Eq.
LDPC
Dec
oder
Car
rier
Rec
over
y
QA
M
Map
per
FTN
Eq
ualiz
er
DA
C AD
CA
DC
WM
F+
CD
Com
p.
De-
inte
rleav
er
DataOutTransmitter Receiver
WM
F+
CD
Com
p.RRC2"#
FTN
Eq
ualiz
er
LDPC
Enc
oder
Inte
rleav
er
QA
M
Map
per RRC2"#
Opt
ical
Fro
nt-e
nd
DA
CSSMF
Coh
eren
t Rx.
2x2
MIM
O B
utte
rfly
PMD
Eq.
Dem
appe
r
LDPC
Dec
oder
Car
rier
Rec
over
y
FTN
Pr
e-eq
ualiz
er
QA
M
Map
per
FTN
Pr
e-eq
ualiz
er
DA
C AD
CA
DC
WM
F+
CD
Com
p.
Dem
appe
r De-
inte
rleav
er
DataIn
DataOutTransmitter Receiver
WM
F+
CD
Com
p.RRC2"#
© M. Jana, A. Mehdra, L. Lampe, J. Mitra "Pre-equalized Faster-than-Nyquist Transmission," IEEE Transactions on Communications, vol. 65, no. 10, pp. 4406-4418, October 2017.
BER
�!
OSNR �!
QPSK, roll-off 0.3, optical comms setting, fixed baud rate𝜏=0.85
8 8.5 9 9.5 10 10.5 11 11.5 12 12.5 1310-6
10-5
10-4
10-3
10-2
10-1
Nyquist (τ = 1)THP FTNTHP FTN demapperMAP, 10 iterations
BER
�!
OSNR �!
QPSK, roll-off 0.3, optical comms setting, fixed baud rate𝜏=0.80
8 9 10 11 12 13 14 1510-6
10-5
10-4
10-3
10-2
10-1
Nyquist (τ = 1)THP FTNTHP FTN demapperMAP, 10 iterationsLinear pre-filtering
Microwave Communication
Ana
log
Fron
t-end
&U
p-co
nver
sion
H-Antenna
V-Antenna
InterleaverFECEncoder
QAMMapper
DACRRC Pulse shape
!(#)%&
InterleaverFECEncoder
QAMMapper
DACRRC Pulse shape
!(#)%'
Sync
hron
ized
R
ecei
ver
Har
d D
ecod
ing
2-D
Ada
ptiv
e D
FE(I
SI &
XPI
)+
Phas
e no
ise
trac
king
Rx Matched Filter!∗(−%)
'(-Sampling
Rx Matched Filter!∗(−%)
'(-Sampling
QAMDemapper
De-interleaver+
FEC Decoder
QAMDemapper
De-interleaver+
FEC Decoder
Rx DSP
)*
)+
M. Jana, L. Lampe, J. Mitra "Dual-polarized Faster-than-Nyquist Transmission Using Higher-order Modulation Schemes,” IEEE SPAWC 2018.
24 26 28 30 32 34 36 38 4010
−6
10−5
10−4
10−3
10−2
10−1
100
SNR in dB
BE
R
Nyq (τ=1), w/o PN
Nyq (τ=1), IPNT
Nyq (τ=1), CPNT
FTN, w/o PN
FTN, IPNT
FTN, CPNT
CPNT vs IPNT 3.2 dB
3.3 dB
1024−QAM Nyquist
256−QAM FTN
256−QAM Nyquist
CPNT vs IPNT 0.55 dB
Dual-polarized FTN transmission over microwave channel with phase noise, roll-off 0.4, 𝜏=0.80
M. Jana, L. Lampe, J. Mitra "Dual-polarized Faster-than-Nyquist Transmission Using Higher-order Modulation Schemes,” IEEE SPAWC 2018.
Dual-polarized FTN transmission over microwave channel with phase noise, spectral efficiency (SE)
22 24 26 28 30 32 334
6
8
10
12
SNR in dB
SE (b
its/s
/Hz/
pol.)
Shannon bound, w/o PN and XPINyquist (τ = 1)DFE−FTNLPE−FTN
1024Qβ = 0.4
1024Qβ = 0.3
512Qβ = 0.25256Q
β = 0.4,τ = 0.8
β = 0
β = 0.3
β = 0.4
256Q β = 0.3, τ = 0.8
β = 0.25
256Q β = 0.25, τ = 0.89
256Qβ = 0.25
256Qβ = 0.4
M. Jana, L. Lampe, J. Mitra "Dual-polarized Faster-than-Nyquist Transmission Using Higher-order Modulation Schemes,” IEEE SPAWC 2018.
Faster-than-Nyquist Extension2D (time-frequency packing)
𝜏T
𝜍𝛥f
a good number of works …
Faster-Than-Nyquist Signal Design for Multiuser Multicell Indoor Visible Light Communications, IEEE Photonics Journal, 2016
Faster-Than-Nyquist Precoded CAP Modulation Visible Light Communication System Based on Nonlinear Weighted Look-Up Table Predistortion, IEEE Photonics Journal, 2018
Polar coding for faster-than-Nyquist signaling, IEEE/CIC International Conference on Communications in China (ICCC), 2017
Application domain
Makes use of excess bandwidth to transmit dataImproves rate in non-(modulation-interference) limited systemsAlternative/complement to larger constellation sizes
Requires equalization, changes to synchronization, channel estimation etcPre-filtering is akin to pulse-shaping and partial-response transmission
Faster-than-Nyquist Transmission