2. What is FEC?
Forward Error Correction is the addition/interleaving of redundancy in a data
stream allowing for correction of errors during reception of data without
retransmission.
FEC is a bandwidth efficient solution to increasing the BER. No need for
retransmission.
Based on highly complex mathematical field known as sets and finite field
theory.
Provides increased gain which can be DIRECTLY applied to the optical link
budget
FEC decreases the BER
FEC is a bandwidth efficient solution to improving the BER.
Allows correction without retransmission
Allows decreased receiver cost for equivalent performance
3. What is FEC..contd
FEC is a coding technology widely used in communication systems. Using a
classical block code as an example, the FEC encoder at the transmit end uses kilo
bits of information as a block code. During the encoding, FEC adds n-k redundant
check bits to the information bits, constructing an n-bit codeword. After the
codeword is transmitted to the receive end over a channel, the FEC decoder
detects and corrects bit errors during decoding – if the errors are within the
correction range. In this manner, FEC prevents interference from channel
transmission and improves the reliability of an optical communication system. This
shows that FEC can efficiently reduce the system BER with only a small number of
redundant overhead bits, thus extending the transmission distance and reducing
system costs.
4. What is FEC..contd
FEC ensures reliable transmission
As the basic bearer technology of optical communication systems, WDM has
seen a rapid expansion in capacity alongside a similarly large increase in
network traffic. During this expansion, every increase in the rate of a single
wavelength has brought about a significant advance in communication
technology:
The increase from 2.5G to 10G wavelengths resulted in an evolution from
direct modulation to external modulation and the use of DCMs for fiber
dispersion compensation.
The increase from 10G to 40G wavelengths marked a transition from OOK
modulation to PSK modulation.
The increase from 40G to 100G brought about the introduction of DSP-enabled
coherent communication technology, in which analog-to-digital converters
(ADCs) work at a rate higher than 56Gbit/s.
As WDM technology advanced, forward error correction (FEC) took on a key
role in ensuring reliable transmission of information and gradually became a
mainstream technology essential for optical communication systems. FEC
technology for fiber communication has also experienced several generations
of evolution
5. What is FEC? (cont.)
W/Errors
1. Xmitter
Sends Message
“The Quick Brown
Fox …”
Error Environment
Data: “The Quicx Wrown Fox …”
1. Receiver
receives
data in error
Error Environment Ist pass
Data: “The Quicx Wrown Fox …”
1. Receiver
receives data in
error
2. Requests resend
data
3. Receives error
Data Received:
“The Quicx Wrown
Fox …”
ARQ
1. Xmitter
Sends Message
“The Quick Brown
Fox …”
2. Xmitter receives
request for
retransmission
3. Resends data
1. Xmitter
Sends Message
“The Quick Brown
Fox …(FEC bits)”
Error Environment 2nd pass
Data: “The Quick Brown Fox …”
Data Received:
“The Quick Brown
Fox …”
free data
FEC
Error Environment
Data: “The Quicx Wrown Fox …(FEC bits)”
1. Receiver receives
Error’d data
2. Using FEC
bits/bytes
Corrects data
Data Received:
“The Quick Brown
Fox …”
6. ABSTRACT
The optical signals undergo quality degradation during transmission, which may
lead the receiver into misjudging “1” for “0” signal or “0” for “1” signal. The
Forward Error Correction (FEC) adds the parity bits during coding by the
transmitter so that the receiver can correct the error bits in the code stream by
calculating the parity bits.
The FEC technology is originally used in the submarine cable system for ultra-long
haul transmission. But with the development of terrestrial optical communication
system and increase of single-channel rate, the FEC will become one of the
optimal choices to lower the OSNR of equipment and networking cost as well.
According to the latest ITU-T G.709 and G.975, the FEC technology is introduced
for STM-16 services and services at higher rates currently to ensure the reliability
of data transmission and greatly extend the transmission distance. Therefore, the
FEC technology has already becomes a key technology in optical communication
development and also one of the hot issues in optical communications.
7. Generation of FEC
Along with the rapid development of optical communication technologies and continual increase in
the capacity and rate of communication networks, the transmission distance (in the case that no
electrical regenerators are used) is greatly affected by the defects in fiber transmission, such as group
velocity dispersion (GVD), polarization mode dispersion (PMD), and attenuation. When no FEC
technology is used and the signal rate is increased to 10 Gbit/s or even higher, the transmission
distance without electrical regenerators turns too short to reach a practical length.
Currently, the effective technologies available to expand capacity and extend communication
distance are as follows:
Wavelength division multiplexing (WDM)
Erbium doped fiber amplifier (EDFA)
High power laser
Dispersion compensation fiber (DCF)
Forward error correction (FEC)
Introduction of EDFA can overcome the power budget problem. However, the inherent defects of
EDFA may bring new problems, such as the decreasing of OSNR because of ASE noises. If the output
optical power is increased too greatly, some negative impacts, such as non-linear effects, may be
brought. To further extend the transmission distance and improve the receiver sensitivity of the
system as well as the system requirement on input OSNR, the FEC needs to be introduced.
8. Some FEC Terminology
FEC Frame = frame of bits/bytes which includes all information(k),
redundancy(check bits), framing bits/bytes. For SONET/SDH frames, the FEC
frame will always be smaller.
t = Error correction factor. Number of bytes/bits that can be corrected
m = symbol size (Reed Solomon Codes). Smallest correctable entity.
n = Code word size. Size of FEC frame including redundant check bits/bytes.
k = Information word size. This is the raw information
(SONET/SDH/ATM/etc).
Check bits = redundant bits appended/inserted to/ito FEC frame to allow
forward error correction. Produced by the encoder.
9. Some FEC Terminology…contd
Encoder = device responsible for determining the necessary check bits and
forming the FEC frame.
Decoder = device responsible for utilizing check bits to correct errors (within
the error correction factor) in an FEC frame, including errors in the check bits.
Burst Error = A consecutive group of symbols(m) in error is a single FEC frame.
Interleaving/Interleave factor - A method for increasing the burst error
handling capability of a particular FEC code. Interleave factor is the number of
symbols interleaved.
Error Multiplication = Increasing the error rate by attempting to correct
beyond the error correction factor
In-Band FEC = Code is contained within band and protocol constraints of the
carrier stream. E.g., SONET in band utilizes un-used overhed bytes for check
bit info. No Clock increase.
Out-of-Band FEC = Code encapsulates the entire data stream and
appends/inserts additional information to the carrier stream. Clock rate
increase.
Overhead/Efficiency - For out-of-band codes, the required additional
bits/bytes to support the error correction factor of the code. Maximum
usable today in OC-48 and OC-192 systems is about 8%.
10. Some FEC Terminology (cont)
Typical “Out-of-Band” frame structure
Frame
Byte
Data to be corrected
(portion of SONET/SDH frame)
Check
Byte 1
Check
Byte 2
Check
Byte 3
Check
Byte 4
Typical “In- Band” frame structure
Check
Byte 1
Check
Byte 2
...
SONET SONET SONET
A1
A1/A2
A2
...
Check
Byte 3
Check SONET
Byte 4
J0
Additional SONET Overhead
and SONET Payload
IF FEC Code is a (255,251) Reed Solomon code then:
m = 8 (symbol size)
n = 255 (code word size)
k = 251 (Information field size)
t = 4 errors can be corrected
Clock rate increase of 255/251 = 1.57% (For out-of-band code)
11. Some FEC Terminology (cont)
Individual streams
Interleaving
A
A
...
B
B
...
C
C
...
D
D
...
Example of Interleaving (4 to 1)
A
B
C
D
A
B
C
D
Data Stream
Why Interleave?
12. Some FEC Terminology (cont)
Individually coded FEC streams
A
A
B
C
Code type:
B
C
B
...
C
...
D
Single data stream
A
...
C
D
A
B
Example of Interleaving (4 to 1)
D
...
D
t = 1 error corrected
Increases Burst Error Correction Capability
13. Some FEC Terminology (cont)
•
SNR
•
Signal-to-noise-ratio
•
ratio of signal power to noise energy
•
higher SNR the lower the BER (typically)
Signal
Signal
Power
SNR (dB)
Noise
Frequency
14. Why Use FEC? (cont.)
Lower Launch Powers
Lower power/cost transmitters/Receivers
Reduction of the effects of fiber non-linearities
Essential for OC-192
Compensation for system loss
Fiber routing, Routers/Splitters/Combiners, Connectors
Additional Performance Monitoring
Longer Spans
Effective Error Free Channel
What does this mean for carriers (Customers)
Increased system gain
Longer spans with fewer amplifiers
Lower launch power with equivalent error performance
Higher channel density
Decreased cost
Better system performance monitoring
15. Why Use FEC? (cont.)
Coding Gain
The advantage of using FEC is that the probability of an error remaining in the
decoded data is lower than the probability of an error if an FEC algorithm, such
as Reed-Solomon, is not used. This is coding gain in essence.
Coding Gain is difference in Input SNR for a given Output BER. The Input SNR is
measured either as “Q factor” or as Eb/N0 (Section 5.1.2.2), or OSNR ().
The “Net Coding Gain” takes into effect that there was a 7% rate expansion due
to the FEC. What this means is that the data rate had to increase by 7% in order
to transmit both the data and the FEC.
16. Why Use FEC? (cont.)
The coding gain provided by the FEC can be used to:
Increase the maximum span length and/or the number of spans, resulting in an
extended reach. (Note that this assumes that other impairments like chromatic and
polarization mode dispersion are not becoming limiting factors.)
Increase the number of DWDM channels in a DWDM system which is limited by the
output power of the amplifiers by decreasing the power per channel and increasing the
number of channels. (Note that changes in non-linear effects due to the reduced per
channel power have to be taken into account.)
-Relax the component parameters (e.g launched power, eye mask, extinction ratio, noise
figures, filter isolation) for a given link and lower the component costs. - but the most
importantly the FEC is an enabler for transparent optical networks:
Transparent optical network elements like OADMs and PXCs introduce significant optical
impairments (e.g. attenuation). The number of transparent optical network elements
that can be crossed by an optical path before 3R regeneration is needed is therefore
strongly limited. With FEC a optical path can cross more transparent optical network
elements. This allows to evolve from today’s point-to-point links to transparent, meshed
optical networks with sufficient functionality.
17. Why Use FEC? (cont.)
•
Coding Gain measured via Q Factor
•
The widely used technique of measuring coding gain is the Q-factor (Quality
factor) measurement. This technique estimates the OSNR at the optical amplifier
or receiver by measuring BER vs. voltage threshold at voltage levels where BER
can be accurately determined (see Figures 4 and 5). In reality, however, Q-factor is
derived from the measurement of the eye-pattern signal. It is defined as the ratio
of peak-to-peak signal to total noise (conventionally electrical)
•
u1 = ON level average value
•
s1 = ON level noise standard deviation
•
u0 = OFF level average value
•
s0 = OFF level noise standard deviation
•
A system that requires an operating BER of 10-15 has a Q-factor measurement of 18 dB without FEC If
•
RS(255, 239) FEC is employed, the Q-factor measurement decreases to 11.8 dB, yielding 6.2 dB of
•
coding gain.
18. FEC code types..glimpse
•
Hamming codes - typically used in Dynamic Ram memory systems (PC)
•
Block codes - data stream is continually sectioned into k-bit blocks and
redundant bits are appended to the data stream
•
Cyclic Codes - Block codes where code words are simply lateral shifts of one
another
•
if c = (c1, c2, c3, …, cn)
•
then (c2, c3,c4,…cn,c1) is a codeword and
•
then (c7,c8,c9,…,cn,c1,…,c6) is also a codeword
•
Convolutional Codes/Viterbi Decoders - Codes which operate as a sliding
sequence of data bits to generate the code stream.
•
Reed-Solomon Codes - subset of BCH (Bose-Chaudhuri-Hocquengheim) block
codes.
19. What is the best code to use?
Depends on
the error environment (bursty, symetrical, bit/block errors, error distribution),
cost of implementation,
power budget (dissipation),
required gain,
bandwidth limits (Components must operate above the channel rate)
Each code has it’s own advantages:
Convolutional Codes - good bit performance with fairly simple implemenation. VLSI
implementation at high speeds is non-trivial. Do not lend themselves to parallization.
Cyclic Codes - reasonable bit performance but highly inefficient for reasonable gain.
Block Codes - Depending on the code gives good bit/burst performance. Non-trivial to
implement but do lend themselves to partitioning and parallization.
BCH/Reed-Solomon Codes - Very good bit/burstperformance with some reasonable
burst performance. Complex but well known. Lend themselves to parallization. Very
good utilization
Concatenated Codes - Best of all worlds. Complex. High degree of gain.
Turbo Codes - Complex. Very high gain.
20. What is the best code to use? (Cont.)
Depends on system requirements:
5dB + gain
Integrated subchannel to support Optical Transport Network (OTN)
Integrated syndrome reporting
1’s and 0’s in error
error’d frames
uncorrectable frames
Single chip solution
Integrated Framer with SONET/SDH PM (PM-192)
Low power
Support components capable of meeting SONET jitter requirements
PRBS generation and detection - both Raw and Payload
SONET/SDH frame generation
SONET payload pattern generation
Bit Error Insertion - Pseudo random error generation with programmable rate
G.975 compliance?
Most of Vendors use to Chose Reed-Solomon
21. FEC Codes
•
The basic characteristics of t-error correcting R-S codes are described below:
-
block length:
2m), (m=symbol size)
n
-
number of parity digits:
q
–
n - k = 2t
•
=
1
(Where
q
=
e.g. the R-S code specified in G.975 for submarine systems is a q = 256, t =
8 error correcting code:
-
block length:
n = 256 -1 = 255
-
number of parity digits:
n - k = 2t = 16
-
length of data block (k):
k = n - 2t = 239
-
FEC overhead is:
16/239 = 6.69 %
22. FEC Codes
Both binary-BCH and R-S codes can correct bursts of up to t digits in length per
code block.
For R-S codes, where each q-ary digit corresponds to log2 q binary digits, the
corresponding bit burst length is log2 q longer, provided the burst is confined to
no more than t q-ary digits.
Both binary-BCH and R-S codes can be given further burst-error correction
capability by interleaving multiple, independent code blocks.
e.g. bit interleaving 8 independent 8-error correcting binary-BCH codes permits
correcting a single 64-bit error burst or up to 8 error bursts of 8 bits each.
23. Various RS FECs
•Spectral Reed-Solomon
(255,241)
1
0.1
0.01
3
1 10
1 10
1 10
1 10
1 10
Bit Error Rate
1 10
1 10
1 10
1 10
1 10
1 10
1 10
1 10
1 10
1 10
1 10
1 10
1 10
1 10
1 10
4
Overhead : 6.25%
5
6
Gain : 6.5 dB (at 1e-22 BER)
7
8
• Spectral Reed-Solomon
(31,27)
9
10
Gain
11
12
13
Overhead : 15.5%
14
15
16
Gain : 3.5 dB (at 1e-22 BER)
17
18
• Spectral Reed-Solomon
(255,225) (Not Shown)
19
20
21
22
0
2
4
6
8
10
12
14
16
Rat io of B it Energy to Noise P ower (db)
un coded
cod ed (31,27)
cod ed (255,241)
18
20
22
Overhead: 12%
Gain: 8 dB (at 1e-22 BER)
• Gain @ 1E-15 ~= 6 dB
24. Raw BER versus Corrected BER
Raw BER
1.0E-03
1.0E-04
1.0E-05
1.0E-06
R-S 255
t=8
7.9E-06
6.6E-15
6.5E-24
6.5E-33
Corrected
b-BCH 2047
t = 11
7.7E-08
6.5E-20
6.4E-32
6.4E-44
BER
R-S 255
t=7
3.3E-05
2.7E-13
2.6E-21
2.6E-29
b-BCH 2047
t = 10
4.2E-07
3.5E-18
3.5E-29
3.5E-40
25. Illustration of FEC gain
Eye @ OSNR = 9 dB
Eye @ OSNR = 17 dB
BER with FEC ~ 1E-15
BER without FEC ~ 1E-15
equal performance
28. In Band FEC
In-band FEC as defined in the ITU-T Recommendation G.707 uses certain
overhead bytes in SDF frames to carry supervisory elements of FEC codes.
(Applicable to SDH systems)
Advantage: Avoid the dispersion limitation by not increasing the line rate, support
interface compatibility, and improve error performance by 1 dB to 2 dB.
Disadvantage: low error-correcting tolerance.
29. Out-of-band FEC
Out-of-band FEC adopts the OTN FEC scheme and is supported by the ITU-T
Recommendations G.975/709. (Widely used in WDM systems)
Advantage: Support high code redundancy, good error-correcting capability, and
high code gains of 5 dB to 6 dB, and allow convenient
insertion of FEC overheads without limitation of the SDH frame structure.
Disadvantage: The inserted overheads cause the increase of line rate. Therefore,
modification is required on relevant devices.
30. Out-Of-Band FEC
In the definition for OTN architecture in the G.709, the FEC overheads
belong to the OTN OTUk layer and the Reed Solomon RS (255, 239) FEC
codes are used as standard FEC codes. In consideration of later
expansion, other codes can also be used.
Currently, the out-of-band FEC serves as the actual FEC code standard.
The FEC types in Recommendations G.975 and G.709 are all out-of-band
FEC, and can use RS (255, 239) codes. However, the two FEC types are
different. Recommendation G.975 recommends direct FEC
encoding/decoding with RS codes for SDH signals, and Recommendation
G.709 defines the OTN architecture, in which the FEC overheads are
defined as a part of OTN OTUk layer and thus become a standard part of
the OTN architecture. In the OTN architecture, columns 3825 to 4080
contain FEC codes.
31. FEC---correction Before & After
Interleaved RS coding supports convenient encoding and decoding, and its coding
structure is compatible with binary codes. RS (255, 239) (RS-8 for short) can
increase line rate by 7.14%.
The RS (255, 239) is a type of RS (n, k) coding. The maximum number of corrected
burst errors in a single block: r = (n - k)/2. The RS (n, k) supports convenient
encoding and decoding and its coding structure is compatible with binary codes.
In RS (255, 239) (RS-8 for short), k equals 239. The 239 data bits and 16 parity bits
form a packet, and the packet code length n equals 255. With RS-8, the maximum
number of corrected burst errors r equals 8 and the line rate increases by 7.14%.
32. FEC-market trend
The WDM products now use only two out-of-band FEC methods. In terms
of coding mode, the two methods are FEC and Advance FEC, namely, RS
coding and AdvanceFEC coding. At present, many boards support setting
of AdvanceFEC and FEC. If compatibility is supported, the two methods
can be used.
33. Coding-Reed-Solomon coding, RS(255,239)
Byte
1
2
3
4
Distance : N-K+1
d= 255-239+1
d=17
Correction: (d-1)/2
c=(17-1)/2
c=8
K byte
message
vector
N byte code
vector
239
240
R check
bytes
254
255
With R=16 check bytes, the RS code
can correct up to 8 erroneous bytes
per code vector
Error correction overhead = 16/255 = 6.25 %
34. Reed-Solomon coding (without interleaving)
Distance = 15-11+1= 5
Message vector
Ctrl
Data to be transmitted
Burst of errors
Transmitted data
More than 2
lost bytes
Lost data
Received data
Correction = (5-1)/2= 2
35. Reed-Solomon coding ( interleaving)
Message
vector
Check
bytes
RS word 0
K=9
Data to be transmitted
RS word 1
RS word 2
R=6
RS word 3
RS word 4
1 DMT symbol in error:
5 lost bytes
N=q*I=15
D=31
S=5/15
Transmitted Data
I=5
Received Data
RS word 0
RS word 1
RS word 2
RS word 3
RS word 4
1 Byte error
per bloc!
Correction Check Correction Check Correction Check Correction Check Correction Check
36. FEC in SDH/OTN
The FEC types in both ITU-R G.975 and G.709 are Out-of-band FEC, and they both
can use the RS (255,239) coding, but they are different. In ITU-R G. 975, the FEC
coding/decoding is directly performed for the SDH signals and the RS coding is
adopted. ITU-R G.709, however, describes the structure of the OTN and defines
the FEC overhead at the OTUk layer of the OTN to be a standard component of
OTN, as shown in fig
37. FEC in SDH/OTN
Already SDH has a FEC defined. It uses undefined SOH bytes to transport the
FEC check information and is therefore called a in-band FEC. It allows only a
limited number of FEC check information, which limits the performance of the
FEC.
For the OTN a Reed-Solomon 16 byte-interleaved FEC scheme is defined, which
uses 4x256 bytes of check information per ODU frame. In addition enhanced
(proprietary) FEC schemes are explicitly allowed and widely used.
FEC has been proven to be effective in OSNR limited systems as well as in
dispersion limited systems.
As for non-linear effects, reducing the output power leads to OSNR limitations,
against which FEC is useful. FEC is less effective against PMD, however. G.709
defines a stronger Forward Error Correction for OTN that can result in up to 6.2
dB improvement in Signal to Noise Ratio (SNR). Another way of looking at this,
is that to transmit a signal at a certain Bit Error Rate (BER) with 6.2 dB less
power than without such an FEC
38. FEC in SDH/OTN
•
G.709 FEC implements a Reed-Solomon RS(255,239) code. A Reed-Solomon code is specified as RS(n,k)
with s-bit symbols where n is the total number of symbols per codeword, k is the number of information
symbols, and s is the size of a symbol. A codeword consists of data and parity, also known as check
symbols, added to the data. The check symbols are extra redundant bytes used to detect and correct
errors in a signal so that the original data can be recovered.
•
For G.709:
•
s = Size of the symbol = 8 bits
•
n = Symbols per codeword = 255 bytes
•
k = Information symbols per codeword = 239 bytes
This means the encoder takes k information symbols of s bits, each, and adds check
symbols to make an n-symbol codeword. There are n-k check symbols of s bits, each. A ReedSolomon decoder can correct up to t symbols that contain errors in a codeword, where 2t = n-k. The
Figure shows a typical Reed-Solomon codeword:
39. FEC-Soft Decision & Hard Decision
FEC coding can be carried out with either of two decision methods: hard decision and soft decision.
The input to a hard-decision FEC decoder consists of a single level of the binary bits 0 and 1. The low
complexity but high maturity of hard decision decoding makes it widely used in a variety of scenarios.
A hard-decision FEC decoder receives data streams consisting only of the binary digits 0 and 1. Hard-decision
decoding will normally be performed based on the algebraic code format. With this decoding mode,
statistical characteristics of channel interference in a signal are lost.
On the other hand, the input to a soft-decision FEC decoder is a multilevel quantization signal.
While offering the same coding rate as hard decision FEC, soft-decision FEC provides a higher coding gain,
albeit with a greatly increased processing complexity. Furthermore, as micro-electronic technologies
advance, soft-decision decoding that can support 100G throughput is now becoming possible. With new
developments in transport technology and the advent of the 100G era, research into, and the applications of,
soft-decision FEC are maturing and will eventually be widely used in high-speed optical communication
systems that use coherent detection
To fully utilize the information in a received waveform and improve the decision accuracy of the decoder,
sampling and quantization can be performed for the received signal . Using this sampling information, the
decoder provides higher decoding accuracy and therefore greatly improves system performance. When
working at the same rate, soft-decision FEC provides a 1.5 dB higher net coding gain than hard-decision FEC.
Soft-decision FEC ensures better performance and a bit throughput that is many times that of hard-decision
FEC. However, it can be implemented only when a high-speed ADC is used to perform sampling and
quantization. In addition, the soft-decision FEC algorithm is very complex because it must consider the
changes in noise probability distribution caused by channel performance deterioration. Fortunately, the
rapid development of integrated circuits has made commercial use of soft-decision FEC a reality.
42. Soft-Decision FEC Benefits for 100G
At 100G rates, leading optical suppliers are implementing third generation FEC
capabilities to extend performance and overall optical distances even further. These
third generation FECs are based on even more powerful encoding and decoding
algorithms, iterative coding, and something referred to as soft decision FEC (SD-FEC).
In a hard decision FEC implementation, the decoding block makes a firm decision
based on the incoming signal and provides a single bit of information (a “1” or “0”) to
the FEC decoder. A signal is received and compared to a threshold; anything above the
threshold is a “1” and anything below the threshold is a “0.”
A soft decision decoder uses additional data bits to provide a finer, more granular
indication of the incoming signal. In other words, the decoder not only determines
whether the incoming signal is a “1” or “0” based on the threshold, but also provides a
“confidence factor” in the decision. The confidence factor provides an indication of how
far the signal is above or below the threshold crossing.
The use of confidence or “probability” bits along with the stronger, more complex third
generation FEC coding algorithms enables the SD-FEC decoder to provide 1–2 dB of
additional net coding gain. In practice, a 3-bit confidence estimation normally provides
most of the theoretically achievable performance improvement. While 1–2-dB coding
gain doesn’t sound like much, it can translate into a 20% to 40% improvement in overall
achievable distances, which is a very substantial improvement at 100G.
One tradeoff with these more advanced FECs is they require ~20% overhead for the
FEC bytes, more than twice the ~7% overhead of first and second generation FECs.
The higher 20% FEC overhead translates to slightly higher optical data rates, which are
already operating at the edges of currently available technology at 100G.
43. Implementing 100G SD-FEC
While the mathematics behind SD-FEC algorithms have been known for
many years and used in the wireless industry, it is only recently that SDFEC has gained interest for use on high-speed optical signals. Numerous
technology and ASIC limitations prevented implementation of third
generation SD-FEC in optical applications. In other words, the
semiconductors weren’t fast enough and didn’t have enough processing
power or memory to support SD-FEC at 100G optical rates.
Take, for example, the high-speed analog-to-digital converters (ADCs)
used inside a 100G receiver. These devices operate at an incredible 56
giga samples per second (Gsa/sec) and just became generally available in
2011. SD-FEC requires the use of even higher-speed ADCs, operating at 63
Gsa/sec to implement the SD-FEC process-ing, along with an equally fast
and powerful SD-FEC silicon implementation. Fortunately, such
component limitations are now part of the past, meaning that SD-FEC for
100G optical signals has become a reality.
44. Summary-SD-FEC
Network
operators
are
under
tremendous pressure to expand
network capacity to order to meet the
ever-increasing demand for high-speed
data services, especially Internet video
services. As backbone speeds increase
from 10G per wavelength to 100G per
wavelength, the OSNR requirements
increase by +10 dB. Without some type
of compensation or correction, 100G
optical distances would be very limited
and uneconomical. First- and secondgeneration FEC algorithms have been
used at both 10G and 40G to lower the
BER and improve overall distances.
Soft-decision FEC is a third-generation
encoding algorithm that enables longer
distances and fewer regenerations on
100G optical networks. Up until
recently, semiconductor and component
technology did not exist to allow
implementation of SD-FEC on high
speed, 100G optical signals. As of
2011, Fujitsu is leading the optical
industry by introducing SD-FEC on the
newest generation of 100G optical
modules.
46. Interworking Problem Because of FEC
The boards with FEC and those without FEC cannot be interconnected, for example, the TWC cannot interconnect with the
LWC. The boards with different FEC modes also cannot interconnect, for example, the coding type of the SSE3LWF is the
AFEC by default, but it can also be set to FEC. Though the rates of two FEC coding modes are the same, yet they still cannot
interconnect due to the difference in coding modes. Therefore, ensure the FEC coding modes of the upstream and
downstream boards are consistent.
If the upstream and downstream CARDS are both configured with the FEC mode, they work normally, as shown in Figure.
If the upstream and downstream CARDs are configured with different coding modes xFEC/FEC, they cannot
interconnect, and the IN1 optical interfaces of them report the OTU-LOF alarm
47. Alarms Related to FEC
The mechanism of the BEFFEC_EXC alarm is that the system judges
whether the value of FEC_BEF_COR_ER exceeds the “Degrade Threshold
before FEC” every second. If the value exceeds the threshold (for example,
FEC_BEF_COR_ER = 5, Degrade Threshold before FEC = 1E-6), the
BEFFEC_EXC alarm is reported. By default, the Degrade Threshold before
FEC is 1E-6. The user can modify the threshold
48. Performance Events Related to FEC
Name
Description
FEC_AFT_COR_ER
After FEC Correct Errored Rate
FEC_BEF_COR_ER
Before FEC Correct Errored Rate
FEC_COR_0BIT_CNT
Forward Error Correction – Corrected 0 Bit Count
FEC_COR_1BIT_CNT
Forward Error Correction – Corrected 1 Bit Count
FEC_COR_BYTE_CNT
Forward Error Correction – Corrected Byte Count
FEC_UNCOR_BLOCK_CN
Forward Error Correction – Uncorrected Block Count
T
49. Performance Events Related to FEC
FEC_AFT_COR_ER
FEC_AFT_COR_ER indicates the line BER after FEC. This performance is counted every second,
reflecting the network performance in real time. FEC_AFT_COR_ER = N indicates that the order of
magnitude of BER after correction is N, and the BER is m*E-N. The value of m ranges from 0.1 to 9.9.
Normally, the BER is expressed in 1E-N. When the received signal is OTU_LOF or LOS,
FEC_AFT_COR_ER = 1 indicates that the BER after FEC is extremely large.
Note: The value of BER, N only indicates the order of magnitude of the BER instead of the specific BER.
We can only infer that the BER ranges between 1.0*E-N and 9.9*E-N. Generally the specific BER is not
required and the order of magnitude of BER is already enough for problem analysis.
For a normal WDM network, the value of FEC_AFT_COR_ER should be 0. It indicates that there are no
bit errors after FEC. If the value of FEC_AFT_COR_ER is not 0, the WDM network needs to be checked
immediately to see if there are faults.
FEC_BEF_COR_ER
FEC_BEF_COR_ER indicates the line BER before FEC. This performance is counted every second,
reflecting the network performance in real time. The meaning is similar to that of FEC-AFT_COR_ER.
When the received signal is OTU_LOF or LOS, FEC_BEF_COR_ER = 1 indicates that the BER before
FEC is extremely large.
A normal WDM network allows a certain BER before FEC. Huawei recommends that when the value of
FEC_BEF_COR_ER is not more than 6, that is, when the line BER before correction is not less than 1E6, the WDM network needs to be checked immediately to see if there are faults. When the value of
FEC_BEF_COR_ER is not more than 8 but larger than 6, that is, the line BER before FEC is not less
than 1E-8 but smaller than 1E-6, check the WDM network to see whether the line bit errors are caused
by OSNR or optical power restriction (normal condition) or by some maintenance factors such as line
fiber aging.
50. Performance Events Related to FEC
FEC_COR_0BIT_CNT
FEC_COR_0BIT_CNT indicates the number of corrected “0” bits by FEC. (That is, the
downstream end sends “0” bits. After transmission over line, “0” bits turn into “1” bits due
to bit errors. With FEC function, these “1” bits are corrected to “0” bits. The number of the
corrected “0” bits is the performance value.) This value is accumulated per second until the
end of the 15th minute. Then, the accumulation restarts. The count of corrected “0” bits in
15 minutes will be stored in the history performance counter.
As an auxiliary performance for problem location, this performance can be used to analyze
the sporadic cases of the error bits by observing the change of it. Generally you only need to
concern the BER before and after the error correction.
The process of 24-hour performance count is similar to that of the 15-minute performance
count. The only difference is that the duration is extended to 24 hours.
FEC_COR_1BIT_CNT
FEC_COR_1BIT_CNT indicates the number of corrected “1” bits by FEC. It is similar to
FEC_COR_0BIT_CNT. You can refer to the descriptions of FEC_COR_0BIT_CNT.
51. Performance Events Related to FEC
FEC_COR_BYTE_CNT
FEC_COR_BYTE_CNT indicates the number of bytes, where there are “0” bit errors or “1” bit
errors, corrected by FEC. (The specific number of bits corrected is not concerned as long as
there are bits corrected in this byte)Normally when the BER is small, there is rare possibility
that one byte has two or more bit errors. Thus, the value of FEC_COR_BYTE_CNT is very
close to the sum of FEC_COR_0BIT_CNT and FEC_COR_1BIT_CNT.
Similar to “FEC_COR_0BIT_CNT” and “FEC_COR_1BIT_CNT”, this performance is auxiliary,
but it can be used to calculate the accurate BER. For example, for the LWF, the value of
FEC_COR_BYTE_CNT is 10 000 000, then BER= Number of corrected
bits/(15(m)*60(s)*10.709G)=10000000/15*60*10.709*1E+9≈1.04E-6.
FEC_UNCOR_BLOCK_CNT
FEC_UNCOR_BLOCK_CNT indicates the number of the frames that fail to be corrected by
FEC. The occurrence of this performance event indicates that there are too many bit errors
in the network, which is beyond the FEC capabilities. When this performance event occurs,
the WDM network needs to be checked immediately to find faults.
Note: If the network once had certain failure such as broken fiber and so on during one
performance period, the occurrence of FEC_UNCOR_BLOCK_CNT is normal as long as it is
not frequent.
52. List of FEC maintenance
suggestions
Condition 1
Condition 2
Suggestion on Maintenance
There
is
no The value of FEC_BEF_COR_ER The quality of the transmission signals is not good
BEFFEC_EXC
is smaller than 1E-9.
enough. No attention needs to be paid.
alarm.
(For example, 1E-10, 1E-11)
There
is
no The value of FEC_BEF_COR_ER The transmission quality is normal and the system
BEFFEC_EXC
is smaller than or equal to 1E-8.
has much margin of correction capability. The service
alarm.
will not be affected running on this level for a long
(For example, 1E-8, 1E-9)
time. Service test is not required.
There
is
no The value of FEC_BEF_COR_ER The transmission quality is not good enough.
BEFFEC_EXC
is larger than 1E-8.
Attention of the operator is required. The system has
alarm.
little margin of correction capability. The service will
(For example, 1E-7)
not be affected running on this FEC correction level
for a long time. It is recommended to locate and
rectify the problem.
There
is
no The
value
of Immediate attention of the operator is required.
BEFFEC_EXC
FEC_UNCOR_BLOCK_CNT is 0. Though current service will not be affected, there is
alarm.
potential problem in the service in the long run.
Service test is required as soon as possible.
There
is
no The
value
of The FEC correction exceeds the designed bearing
BEFFEC_EXC
FEC_UNCOR_BLOCK_CNT is not capacity. Services may be affected at any moment or
alarm.
0.
error bits are already generated. Service test and
troubleshooting are required as soon as possible.
54. Cases:
The Signal-To-Noise Ratio is Normal, but There are FEC Counts.
Problem Description:
For a network, the signal-to-noise ratio is normal (The FEC function is enabled for each OTU board. The
signal-to-noise ratio of the system exceeds Chinese standard; namely, OSNR>20dB), but the OTU has
certain FEC counts. Customers think that the system has bit errors owing to the FEC count. However,
when OSNR exceeds Chinese standard, bit error should not exist. Customers require a rational
explanation.
Cause Analysis:
The following introduces the related rules and regulations to FEC of the WDM system in Chinese
standard:
In the system without FEC, OSNR should be greater than 25 dB.
In the system with out-of-band FEC, OSNR should be greater than 20 dB. See the following figure.
Therefore, FEC should be understood in the following way:
For the 10G system (without FEC and with FEC), when OSNR > 25 dB, no bit error exists.
For the 10G out-of-band FEC system, when OSNR > 20 dB, no bit error exists after FEC, but bit
errors may exist before FEC. The bit errors of the system before FEC are corrected by the FEC
function of the board.
The FEC (in-band FEC or out-of-band FEC) in a system is used to improve the OSNR tolerance.
Therefore, to ensure that the entire system has no bit error or error correction, OSNR should be
greater than 25 dB.
55. Troubleshooting Startup
Formula of calculating the FEC count:
Error correction count = BER before FEC x Bit count = BER before FEC × Board rate (bit/s) x time
(s)
In theory, the maximum BER that can be corrected by FEC for the OTU is 8.27E-5.
For a 10G board, if the BER before FEC is 8.27E-5, the formula of calculating the FEC count is
shown as follows:
FEC count = (8.27E–5) x (10E + 9) x (60 x 15) = 744300000
According to the formula, you can calculate the FEC count in 15 minutes or 24 hours for a
2.5G/10G service.
The routine maintenance need be performed as required. For a service at 10
Gbit/s that is required to be without bit errors, the FEC count in 15 minutes
should be less than 740 million. If the BER before FEC is required to be 1E-6,
the stable FEC count should be the following:
FEC count = BER before FEC x Bit count = BER before FEC x Board rate (bit/s) x Time (s) = (1E–6)
x (10E + 9) x (60 x 15) = 9000000. That is, the FEC count n 15 minutes should be less than nine
million stably.
56. Troubleshooting contd.
In addition, pay attention to the stability of error correction. Regard the
worst point as the object. Pay attention to the following situations:
Check the network when the BER before FEC in 15 minutes is above 1E-7. Namely, for a
board at 2.5 Gbit/s, the FEC count is over 0.25 million; for a board at 10 Gbit/s, the FEC count
is over 1 million. At this moment, the network has certain margin. However, you need to find
out the reason and remove the potential problems, to avoid further degradation or
abnormal fluctuation of the network.
If the abnormal fluctuation of FEC count exceeds three magnitudes, find out the reason and
remove the potential problems.
The OSNR is an internal feature of the system. The equipment from
different manufacturers is not involved. For a particular kind of coding, the
error correction mode is certain. Therefore you cannot find an ITU-T
standard.
58. Swizzle FEC for 40G and 100G Optical Transmission
The introduction of OTU3 at 40G and OTU4 at 100G has put a great deal of pressure on
SNR budgets. A strong FEC is the most economical way of regaining some of the link
budget.
The “Swizzle” Spiral Interleaved Turbo Forward Error Correction code is PMC-Sierra’s
third-generation FEC designed to meet the needs of 40G and 100G DWDM systems.
It offers 9.45dB of net coding gain with the standard 6.7% OTN overhead, 1.35dB better
than the second-generation FECs captured in G.975.1.
The “Swizzle” FEC code is a third-generation FEC designed by PMC-Sierra to meet the
needs of 40G and 100G DWDM systems.
The Swizzle FEC is particularly attractive for systems that, due to bandwidth-limitation
or power reasons, cannot use more than the standard 6.7% overhead. Swizzle provides
these systems with gain 1.35dB superior to existing second-generation 6.7% overhead
FECs.
This additional gain can be used to extend reach, operate over lower-quality fiber, and
correct for nonlinear impairments that constrain maximum wavelength density.
59. Swizzle Applications
The Swizzle FEC is best enabled for book-ended Intra-domain interfaces that require gains
greater than that offered by the G.709 Standard FEC. It provides 9.45dB of net coding gain at
an output bit error rate of 1E-15, which is sufficient for a large number of applications and
particularly suited to metro applications.
60. What Swizzle Can?
Swizzle does three things better than existing 2-D codes:
• It uses a better interleaving structure that improves performance and decreases
latency, as described in Section 3.2.
• It allows a more parallel implementation, which increases performance for the
same latency, as described in Section 3.4.
• It replaces the simple “last decoded codeword wins” procedure with a
maximum-likelihood decode procedure that is extremely resistant to false decode,
61. Swizzle Summary
PMC’s Swizzle Spiral Interleaved Turbo FEC performance of 9.45db NECG is 1.35db higher
than typical G.709 and G.975.1 FEC solutions. The Swizzle FEC also delivers the highest NECG
compared to any announced hard FEC solutions. This superior performance is driven by the
following key advantages:
• First, it combines a hard-decode algorithm with soft-decode concepts, allowing the
maximum extraction of information from the channel.
• Second, it uses tight interleaving and parallel decoding to allow more than twice as many
iterations over the same latency.
• Third, it uses an intelligent scheduler to allocate decode resources, focusing an order of
magnitude more processing on worst-case blocks than previous implementations.
In summary PMC’s Swizzle hard decision FEC delivers the following key benefits over typical
G.975.1 based solutions:
• Corrects 4x more errors, which helps operators scale current 10Gbps transport links to
40Gbps and 100Gbps.
• Enables 35% longer optical reach, helping reduce the number of optical regenerators
required, and thus enabling operators to cost-effectively transition to scalable OTN transport
networks
coding gain: Coding gain means the improvement of received optical sensitivity by FEC,without considering penalty by bit rate increasing.net coding gain: Net coding gain means the improvement of received optical sensitivity byFEC, with considering penalty by bit rate increasing.
u1 = ON level average value s1 = ON level noise standard deviation u0 = OFF level average value s0 = OFF level noise standard deviation A system that requires an operating BER of 10-15 has a Q-factor measurement of 18 dB without FEC If RS(255, 239) FEC is employed, the Q-factor measurement decreases to 11.8 dB, yielding 6.2 dB of coding gain.
Because of the presence of impulsive noise (infrequent high amplitude bursts of noise mostly caused by central office switching transients, dial pulses, POTS ringing, neigh boring railway stations, elevators,…that can corrupt the signal beyond recognition) on the twisted pair wire means must be implemented to make sure that the ADSL transceiver is sufficiently robust against this impulsive noise and to maintain an acceptable BER for good quality of services.
Since errors often come in bursts it is likely that the error correction possibility is insufficient.
Interleaving:In stead of transmitting our RS code words directly on the line we will create a frame of the same size made up by multiple RS words by taking only a portion of each of the original RS words. This has the advantage when a burst of errors occur on the line and the original RS words are recreated on the receiving side, the errors will be spread over multiple RS words. This could mean that we are able to correct the errors within a single RS word if the number of errors are within the RS correction boundaries.Interleaving depths:The main disadvantage of interleaving is the high delay. Constructing the blocs that will finally be transmitted over the line take time as you have to wait for a time before you can actually start transmitting. In our example here we need 15 original RS words (1 Byte of each) before we can construct the first block that is actually transmitted over the line.At the receiving side it will also cost extra time to reconstruct the original RS word. The first original RS word can not be reconstructed before we have received all the bytes of this first RS word.Interleaving can be sped up by using different depths, i.e. by taking bigger chunks of the original RS words you will be able to construct your first bloc for transmission much quicker. This has the disadvantage that errors will be spread over less RS words on the receiving side with the possibility that they can not be corrected.Vendors use 3 interleaving depths: high, medium and low.Where high uses the smallest portions to construct a frame for transmission, consequently the high interleaving depth will introduce the highest delay as this one will need more frames to construct/reconstruct before transmission/reception.
For systems requiring additional gain, the Swizzle FEC could be disabled and a module with a soft-decision FEC using up to 20% redundancy could be plugged in.