Download - Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e
![Page 1: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/1.jpg)
Staircase CodesError-correction for High-Speed Fiber-Optic Channels
Frank R. KschischangDept. of Electrical & Computer Engineering
University of Toronto
Talk at Delft University of Technology, Delft,The NetherlandsApril 23rd, 2013
1
![Page 2: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/2.jpg)
Acknowledgements
Joint work with:
Benjamin P. SmithUniversity of TorontoAndrew Hunt and John LodgeCommunications Research Centre, OttawaArash FarhoodCortina Systems Inc., Sunnyvale/Ottawa
Thank you to Drs. Nader Alagha and Fernando Kuipers for theinvitation!
2
![Page 3: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/3.jpg)
Fiber-Optic Communication Systems
Physics: Enabling Technologies
1 Low-loss optical fiber (∼54 THz bandwidth)
2 Optical amplifiers
3 Laser transmitters and Mach-Zehnder modulators
3
![Page 4: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/4.jpg)
Fiber-Optic Communication Systems: Challenges
Reliability
Pe < 10−15
Speed
100 Gb/s per-channel data rates
Non-Linearity
The fiber-optic channel is non-linear in the input power
4
![Page 5: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/5.jpg)
Fiber-Optic Communication Systems: Challenges
Reliability
Pe < 10−15
Speed
100 Gb/s per-channel data rates
Non-Linearity
The fiber-optic channel is non-linear in the input power
4
![Page 6: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/6.jpg)
Fiber-Optic Communication Systems: Challenges
Reliability
Pe < 10−15
Speed
100 Gb/s per-channel data rates
Non-Linearity
The fiber-optic channel is non-linear in the input power
4
![Page 7: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/7.jpg)
Outline of Talk
Part IBinary Error-Correcting Codes for High-SpeedCommunications
Syndrome-based interative decoding
Performance optimization for product-like codes
Staircase codes
FPGA-based simulation results
Analytical error-floor predictions
Part IISpectrally-Efficient Fiber-Optic Communications
Memoryless capacity estimates, with and without digitalbackpropagation
Pragmatic coded modulation via staircase codes
5
![Page 8: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/8.jpg)
Part IBinary Error-Correcting Codes for High-SpeedCommunications
most reliable least reliable
Π Π Π
6
![Page 9: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/9.jpg)
Outline
This part is about . . .
FEC for the binary symmetric channel (BSC)
After optical (and/or) electrical compensation, suitable asforward-error-correction (FEC) for 100Gb/s PD-QPSKsystems without soft information
Outline
1 Existing solutions (G.975, G.975.1)
2 Implementation considerations
3 Staircase codes
4 Generalizations and conclusions
7
![Page 10: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/10.jpg)
Outline
This part is about . . .
FEC for the binary symmetric channel (BSC)
After optical (and/or) electrical compensation, suitable asforward-error-correction (FEC) for 100Gb/s PD-QPSKsystems without soft information
Outline
1 Existing solutions (G.975, G.975.1)
2 Implementation considerations
3 Staircase codes
4 Generalizations and conclusions
7
![Page 11: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/11.jpg)
Existing Solutions
8
![Page 12: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/12.jpg)
Performance Measure
Net Coding Gain
Given a particular BER, we can obtain the corresponding Q via
Q =√
2erfc−1 (2 · BER) .
The coding gain (CG) of an error-correcting code is defined as
CG (in dB) = 20 log10(Qout/Qin),
and the Net Coding Gain (NCG) is
NCG (in dB) = CG + 10 log10(R),
where R is the rate of the code.
9
![Page 13: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/13.jpg)
ITU-T G.975 (1996)
Reed-Solomon (255,239) Code
Coding Symbols arebytes
Depth-16 interleavingcorrects (some) bursts upto 1024 bits
Coding Gain = 6.1 dB
Net Coding Gain = 5.8dB
10-15
10-12
10-9
10-6
10-3
0 5 10 15 20
P[e
]20 log10(Q) (dB)
BS
C L
imit
(C=
239/
255)
2-P
SK
Lim
it (C
=23
9/25
5)
Sha
nnon
Lim
it (C
=23
9/25
5)
6.1 dB
RS
(255,239)
uncoded 2-PS
K
Constraint: Rate and framing structure fixed for future generations
10
![Page 14: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/14.jpg)
ITU-T G.975.1 (2004)
Concatenated Codes
Product-like codes with algebraic component codes
C C
Π
Information Symbols
kana
kb
nb
Column Parity
Row Parity
Parity on Parity
11
![Page 15: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/15.jpg)
ITU-T G.975.1 (2004)
Code NCG NotesI.2 8.88 dB @ 10−15 Outer RS, Inner CSOCI.3 8.99 dB @ 10−15 Outer BCH (t = 3), Inner BCH (t = 10)I.4 8.67 dB @ 10−15 Outer RS, Inner BCH (t = 8)I.5 8.5 dB @ 10−15 Outer RS, Inner Product (t = 1)I.6 8.02 dB @ 10−15 LDPCI.7 8.09 dB @ 10−15 Outer BCH (t = 4), Inner BCH (t = 11)I.8 8.00 dB @ 10−15 RS(2720,2550)I.9 8.63 dB @ 7 · 10−14 ∼Product BCH (t = 3), Erasure Dec?
12
![Page 16: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/16.jpg)
Objectives
Increased Net Coding Gain
NCG ⇒ Shannon Limit of BSC (9.97 dB at 10−15)
Error Floor
Error floor 10−15
Lower error floor ⇒ ‘Insurance’ in presence of correlated errors
Block Length
n ≈ 2 · 106 or less
Low Implementation Complexity
Dataflow considerations at 100Gb/s
13
![Page 17: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/17.jpg)
Implementation Considerations
14
![Page 18: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/18.jpg)
Hardware Considerations: Product vs. LDPC
Product Code
C C
ΠAlgebraic component codes
Syndrome-based decoding
LDPC Code
Π SPC component codes
Belief propagation decoding
15
![Page 19: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/19.jpg)
Syndromes
The codewords of a linear (n, k) block code C are the solutions ofa homogeneous system of simultaneous linear equations, i.e.,
v ∈ C ⇔ vH> = 0.
Let v ∈ C be sent and let r be received. Define the error pattern eas e = r − v so that
r = v + e.
Then the syndrome corresponding to r is
s = rH> = (v + e)H> = vH>︸︷︷︸0
+eH> = eH>.
An error-correcting decoder must infer the most likely error patternfrom the syndrome.
16
![Page 20: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/20.jpg)
Syndrome-based Decoding of Product Codes
Decoding an (n, k) component codeword
n received symbols ⇒ n − k symbol syndrome
R = 239/255, n ≈ 1000, n − k ≈ 32
For high-rate codes, syndromes provide compression
∼ 3 decodings/component
≤ 96 bits/decoding
∼ 21000 components/symbol
⇒ 0.768 bits/symbol
AlgebraicDecoderSyndrome
Error Locators UpdateSyndromes
Total Dataflow
At 100Gb/s, 76.8 Gb/s internal dataflow
17
![Page 21: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/21.jpg)
LDPC Belief Propagation Decoding
∼ 15 iterations
2 messages/iteration·edge
∼ 5 bits/message
∼ 3 edges/symbol
⇒ 450 bits/symbol
Π
Total Dataflow
At 100Gb/s, 45Tb/s internal dataflow!
18
![Page 22: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/22.jpg)
Dataflow Comparison
76.8 Gb/s 45 Tb/s
2–3 orders of magnitude (huge implementation challenge for softmessage-passing LDPC decoders).
19
![Page 23: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/23.jpg)
Our Solution
20
![Page 24: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/24.jpg)
Coding with Algebraic Component Codes
C1 C2 C3 Cl
Π
Graph Optimization
Degrees of Freedom
Mixture of component codes (e.g., Hamming, BCH)
Multi-edge-type structures
21
![Page 25: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/25.jpg)
Staircase Codes: Construction
Consider a sequence of m-by-m matrices Bi
B−1 B0 B1 B2
m
m
and a linear, systematic, (n = 2m, k = 2m − r) component code C
2m− r r
ParityInformation
Encoding Rule
∀i ≥ 0, all rows of[BTi−1Bi
]are codewords in C
22
![Page 26: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/26.jpg)
Staircase Codes: Construction
B−1
BT0 B1
BT2 B3
23
![Page 27: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/27.jpg)
Staircase Codes: Construction
B−1
BT0 B1
BT2 B3
23
![Page 28: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/28.jpg)
Staircase Codes: Construction
B−1
BT0 B1
BT2 B3
23
![Page 29: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/29.jpg)
Staircase Codes: Construction
B−1
BT0 B1
BT2 B3
23
![Page 30: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/30.jpg)
Staircase Codes: Construction
B−1
BT0 B1
BT2 B3
23
![Page 31: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/31.jpg)
Staircase Codes: Construction
B−1
BT0
BT2 B3
B1
23
![Page 32: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/32.jpg)
Staircase Codes: Construction
B−1
BT0
BT2 B3
B1
23
![Page 33: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/33.jpg)
Staircase Codes: Properties
Hybridization of recursive convolution coding and blockcoding
Recurrent Codes of Wyner-Ash (1963)
Bi
AlgebraicEncoder
Delay
Bi−1
Infoi
Rate : R = 1− r/m
Variable-latency (sliding-window) decoder
Bi Bi+1 Bi+2 Bi+3Bi−1 Bi+4
24
![Page 34: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/34.jpg)
Staircase Codes: Properties
BTi Bi+1
BTi+2 Bi+3
The multiplicity of (minimal) stalls of size (t + 1)× (t + 1) is
K =
(m
t + 1
)·t+1∑j=1
(m
j
)(m
t + 1− j
),
and the corresponding contribution to the error floor, fortransmission over a binary symmetric channel with crossoverprobability p, is
BERfloor = K · (t + 1)2
m2· p(t+1)2 .
25
![Page 35: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/35.jpg)
General Stall Patterns
BTi Bi+1
BTi+2 Bi+3
(5,5)-stall
K L Contribution
4 4 3.55× 10−21
4 5 7.81× 10−28
5 5 2.54× 10−22
5 6 2.21× 10−28
6 6 1.40× 10−23
6 7 1.49× 10−29
7 7 8.53× 10−25
7 8 1.83× 10−32
Contribution of (K , L)-stalls,p = 4.8× 10−3.
26
![Page 36: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/36.jpg)
Multi-Edge-Type Representation
ΠB
mm
ΠB ΠB
most reliable least reliable
Bi−1 Bi Bi+1 Bi+2 Bi+3
C C
Π
27
![Page 37: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/37.jpg)
Movie Time
Bi Bi+1 Bi+2 Bi+3Bi−1 Bi+4
28
![Page 38: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/38.jpg)
Braided Block Codes: Construction
P
P
I
Felstrom, Truhachev, Lentmaier, Zigangirov, “Braided BlockCodes,” IEEE Trans. on Info. Theory, 2009.
29
![Page 39: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/39.jpg)
Braided Block Codes: Construction
I P
P
P
I
P
Felstrom, Truhachev, Lentmaier, Zigangirov, “Braided BlockCodes,” IEEE Trans. on Info. Theory, 2009.
29
![Page 40: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/40.jpg)
Braided Block Codes: Construction
I P
P
P
I
P I
P
P
Felstrom, Truhachev, Lentmaier, Zigangirov, “Braided BlockCodes,” IEEE Trans. on Info. Theory, 2009.
29
![Page 41: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/41.jpg)
Braided Block Codes: Performance
Our experiments show:
For high-rate (239/255), braided block codes yieldapproximately the same performance as the correspondingproduct code
“Multi-edge gain” diminished, since fewer bits (previous parityonly) serve as “clean”
But:
Spatially-coupled generalized LDPC ensembles with iterativehard-decision decoding can approach capacity at high rates:Y.-Y. Jian, H. Pfister, K. Narayanan, “Approaching Capacityat High Rates with Iterative Hard-decision Decoding,” Proc.IEEE Int. Symp. Inform. Theory, July, 2012.
30
![Page 42: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/42.jpg)
FPGA-based Simulation Results
Code Parameters
m = 510, r = 32, triple-error-correcting BCH component code
10-15
10-14
10-13
10-12
10-11
10-10
10-9
10-8
10-7
10-6
10-4
10-3
10-2
BERout
BERin
20 log10(Q) (dB)7.5 8.0 8.5 9.0 9.5 10.0 10.5 11.0
BS
C L
imit (C
=2
39
/25
5)
RS(255,239)
Sta
ircase
I.3 I.4 I.5
I.9G.975.1 codes
31
![Page 43: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/43.jpg)
Part IISpectrally-Efficient Fiber-OpticCommunications
most reliable least reliable
Π Π Π
32
![Page 44: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/44.jpg)
The Kerr Effect
Kerr electro-optic effect (DC Kerreffect)
An effect discovered by John Kerr in1875
It produces a change of refractive indexin the direction parallel to theexternally applied electric field
The change of index is proportional tothe square of the magnitude of theexternal field
33
![Page 45: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/45.jpg)
The Kerr Effect in Optical Fibers
Optical Kerr Effect (or AC Kerr effect)
No externally applied electric field is necessary
The signal light itself produces the electric field that changesthe index of refraction of the material (fused silica)
The change in index in turn changes the signal field
The change in index of refraction is proportional to the squareof the field magnitude
P
f
Optical
Fiber
WDM channel Channel of interest
Changes index
of refraction
Generates
distortion
A signal in a certain frequency band can distort the signal in adifferent frequency band without spectral overlap
34
![Page 46: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/46.jpg)
Nonlinear Effects in Fibers
inter-channel
signal-noise signal-signal
WDM nonlinearities
XPM FWM
NL phase noise
XPM-induced
NL phase noise
intra-channel
signal-noise signal-signal
NL phase noise parametric
amplification
SPM-induced
NL phase noise
SPM
isolated pulse
SPMIXPM IFWM
MI
NL=nonlinear; SPM=self-phase modulation; MI=modulation instability; XPM=cross-phase modulation; IXPM=intrachannel XPM
FWM=four-wave mixing; IFWM=intrachannel FWM; WDM=wavelength division multiplexing
35
![Page 47: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/47.jpg)
System Model
coherent fiber-optic communication system
standard-single-mode fiber
ideal distributed Raman amplification
But, could also consider
systems with inline dispersion-compensating fiber
lumped amplificationN spans
A(0, t) A(L, t)
SSMFTX RXAMP
36
![Page 48: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/48.jpg)
Generalized Nonlinear Schrodinger Equation
A(z , t) is the complex baseband representation of the signal(the full field, representing co-propagating DWDM signalsTransmitter sends A(0, t)Receiver gets A(L, t), where L is the total system length
Evolution of A(z , t)
The generalized non-linear Schrodinger (GNLS) equation expressesthe evolution of A(z , t):
∂A
∂z+
jβ22
∂2A
∂t2− jγ|A|2A = n(z , t).
No loss term (since ideal distributed Raman amplification isassumed).
n(z , t) is a circularly symmetric complex Gaussian noise processwith autocorrelation
E[n(z , t)n?(z ′, t ′)
]= αhνsKT δ(z − z ′, t − t ′),
where h is Planck’s constant, νs is the optical frequency, and KT isthe phonon occupancy factor.
37
![Page 49: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/49.jpg)
System Parameters
Second-order dispersion β2 -21.668 ps2/kmLoss α 4.605× 10−5 m−1
Nonlinear coefficient γ 1.27 W−1km−1
Center carrier frequency νs 193.41 THzPhonon occupancy factor KT 1.13
38
![Page 50: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/50.jpg)
Solving the GNLS Equation
Throughout propagation over an optical fiber, stochastic effects(noise), linear effects (dispersion) and nonlinear effects (Kerrnonlinearity) interact.
Noise Nonlinearity
Dispersion
Even in the absence of noise, solving the GNLS equation requiresnumerical techniques.
39
![Page 51: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/51.jpg)
Split-Step Fourier Method
divide fiber length into short segments
consider each segment as the concatenation of (separable)nonlinear and linear transforms
for distributed amplification, an additive noise is added afterthe linear step.
A(z0, t) −→ A(z0 + h, t) step size h
40
![Page 52: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/52.jpg)
Nonlinear Step
In the absence of linear effects, the GNLS equation has the form
∂A
∂z= jγ|A|2A,
with solution
A(z0 + h, t) = A(z0, t) exp(jγ|A(z0, t)|2h).
41
![Page 53: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/53.jpg)
Linear Step
We now use the previous solution as the input to the the linearstep, i.e., let
A(z0, t) = A(z0, t) exp(jγ|A(z0, t)|2h)
be the input to the linear step. The linear form of the GNLSequation is
∂A
∂z= −α
2A− jβ2
2
∂2A
∂t2,
which can be efficiently solved in the frequency domain. Defining
A(z , t) =1
2π
∫ ∞−∞
A(z , ω) exp(jωt)dω,
it can be shown that
A(z0 + h, ω) = A(z0, ω) exp
((jβ22ω2 − α
2
)h
).
42
![Page 54: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/54.jpg)
Split-Step Propagator
Putting this together, we have
A(z0 + h, t) = F−1FA(z0, t)
exp
((jβ22ω2 − α
2
)h
),
where F is the Fourier transform operator, and where
A(z0, t) = A(z0, t) exp(jγ|A(z0, t)|2h)
43
![Page 55: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/55.jpg)
Digital Backpropagation
Digital backpropagation = split-step Fourier method, using anegative step-size h, performed at the receiver
Full compensation (involving multiple WDM channels)generally impossible, even in absence of noise (due towavelength routing)
Noise is neglected (cf. zero-forcing equalizer)
Single-channel backpropagation typically performed, afterextraction of desired channel using a filter
44
![Page 56: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/56.jpg)
Capacity Estimation
See: R.-J. Essiambre, G. Kramer, P. J. Winzer, G. J. Foschini, B.Goebel, “Capacity limits of optical fiber networks,” J. Lightw.Technol., vol. 28, pp. 662–701, Sept./Oct. 2010.
System Model
DSP1
2Ts− 1
2Ts
ChannelTX
A(L, t)t = kTs φk,0
LPF
A(0, t)
45
![Page 57: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/57.jpg)
Transmitted Signal
Channel l signal:
Xl(t) =∞∑
k=−∞
φk,l√Ts
sinc
(t − kTs
Ts
),
where sinc(θ) = sinπθπθ .
φk,l are elements of a discrete-amplitude continuous-phase inputconstellation M, i.e, for N rings, θ ∈ [0, 2π), and r ≥ 0,
M = m · r exp (jθ) |m ∈ 1, 2, . . . ,N .
Each ring is assumed equiprobable, and for a given ring, the phasedistribution is uniform.
46
![Page 58: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/58.jpg)
Multi-channel systems
In the general case of a multi-channel system having 2B + 1channels with a channel spacing 1/Ts Hz, the input to the fiberhas the form
A(z = 0, t) =∞∑
k=−∞
B∑l=−B
φk,l√Ts
sinc
(t − kTs
Ts
)e j2πlt/Ts .
47
![Page 59: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/59.jpg)
Towards a Probability Model
Back-rotation:
φk,l = φk,l exp (−j(ΦXPM + ∠φk,l)) ,
where ΦXPM is a constant (input-independent) phase rotationcontributed by cross-phase modulation (XPM).
−8 −6 −4 −2 0 2 4 6 8
x 10−3
−8
−6
−4
−2
0
2
4
6
8x 10
−3
Imag
(m
W1/
2 )
Real (mW1/2)
−8 −6 −4 −2 0 2 4 6 8
x 10−3
−8
−6
−4
−2
0
2
4
6
8x 10
−3
Imag
(m
W1/
2 )
Real (mW1/2)
48
![Page 60: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/60.jpg)
Gaussian fitting
For each i and a fixed l (the channel of interest), we calculate themean µi and covariance matrix Ωi (of the real and imaginarycomponents) of those φk,l corresponding to the i-th ring, andmodel the distribution of those φk,l by N (µi ,Ωi ).
From the rotational invariance of the channel, the channel ismodeled as
f (y |x = r · i exp (jφ)) ∼ N (µi exp (jφ) ,Ωi ),
where the (constant) phase rotation due to ΦXPM is ignored, sinceit can be canceled in the receiver.
49
![Page 61: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/61.jpg)
Capacity Estimation
The mutual information of the (assumed) memoryless channel is
I (X ;Y ) =
∫ ∫f (x , y) log2
f (y |x)
f (y)dx dy ,
where f (x) represents the input distribution on M withequiprobable rings and a uniform phase distribution, which providesan estimate of the capacity of an optically-routed fiber-opticcommunication system.
50
![Page 62: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/62.jpg)
Signaling Parameters
Baud rate 1/Ts 100 GHzChannel bandwidth W 101 GHzNumber of rings N 64Number of channels 2B + 1 = 5
51
![Page 63: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/63.jpg)
Achievable Rates from Memoryless CapacityEstimate
10 15 20 25 30 35 400
1
2
3
4
5
6
7
8
9
10
SNR (dB)
Spectral Efficiency (bits/s/Hz)
L=2000 km, Eq. only L=2000 km, BPL=1000 km, Eq. onlyL=1000 km, BPL=500 km, Eq. onlyL=500 km, BPShannon Limit (AWGN)
(BP adds 0.55 to 0.75 bits/s/Hz relative to EQ) 52
![Page 64: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/64.jpg)
Pragmatic Coded Modulation via Staircase Codes
Approach: BICM + shaping, modulation 2K+2-QAM,hard-decisions
K bits/sym
Modulator
Channel
Dem
odulator
Viterbi
Decoder
Binary FEC
Decoder
Binary FEC
Encoder
HTU
2 bits/sym
X Y
K · Rbits/sym
1bit/sym
[H−1U ]T
1bit/sym
K · Rbits/sym
Syndrome-former matrix
HTU = [1 + D + D2, 1 + D2]T .
53
![Page 65: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/65.jpg)
Achievable Rates
Pin IPFiber System K pavg (dBm) (bits/s/Hz)
L = 500 km, EQ 8 1.61× 10−2 −6 8.05L = 500 km, BP 8 3.52× 10−3 −4 8.73L = 1000 km, EQ 6 3.88× 10−3 −6 6.78L = 1000 km, BP 8 2.22× 10−2 −4 7.77L = 2000 km, EQ 6 2.52× 10−2 −6 5.98L = 2000 km, BP 6 5.16× 10−3 −4 6.72
(These achieve within 0.4 to 0.6 bits/s/Hz of estimated channelcapacity.)
54
![Page 66: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/66.jpg)
Staircase code design
First, design a collection of staircase codes of various (appropriate)rates:
10−3 10−2 10−110−14
10−12
10−10
10−8
10−6
10−4
10−2
BERin
BERout
R=77/95R=239/255R=3/4R=11/15R=146/157
55
![Page 67: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/67.jpg)
Coded Modulation Performance
Then, simulate their performance on the actual channel:
10 15 20 25 30 35 400
1
2
3
4
5
6
7
8
9
10
SNR (dB)
Spectral Efficiency (bits/s/Hz)
L=2000 km, Eq. only L=2000 km, BPL=1000 km, Eq. onlyL=1000 km, BPL=500 km, Eq. onlyL=500 km, BPShannon Limit (AWGN)
Performance to within 0.62 bits of estimated capacity is achieved!(Much of the gap is due to quantization, i.e., hard-decisions).
56
![Page 68: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/68.jpg)
Code Parameters
Spec. Eff.Fiber System m t R (bits/s/Hz)
L = 500 km, EQ 190 4 77/95 7.48L = 500 km, BP 255 3 239/255 8.50L = 1000 km, EQ 255 3 239/255 6.62L = 1000 km, BP 144 4 3/4 7.00L = 2000 km, EQ 120 4 11/15 5.40L = 2000 km, BP 628 4 146/157 6.58
57
![Page 69: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/69.jpg)
Conclusions
At rate 239/255, staircase codes provide best-in-class 9.41 dBNCG
For high-spectral efficiency modulation, a pragmatic codingapproach using staircase codes yields performance with 0.6bits/s/Hz (at L = 2000 km) with practical decodingcomplexity
58
![Page 70: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/70.jpg)
Future Directions
Can these capacity estimates be improved?
Can soft-decisions be incorporated while retaining lowcomplexity?
Are there (well-principled) alternatives to digitalbackpropagation?
Thank you for your attention!
59
![Page 71: Staircase Codes - IEEE Web Hosting...Fiber-Optic Communication Systems: Challenges Reliability P e](https://reader035.vdocuments.net/reader035/viewer/2022071419/61176bf3c9c8cd257f121f27/html5/thumbnails/71.jpg)
Future Directions
Can these capacity estimates be improved?
Can soft-decisions be incorporated while retaining lowcomplexity?
Are there (well-principled) alternatives to digitalbackpropagation?
Thank you for your attention!
59