You are on page 1of 29

MLSSE Equaliz

Un
Dep
Dep
zers for Fr
A
nder the able
partment of
partment of
Motilal N
requency D
Transm
A Term Pap
Bha
Sha
Vih
e guidance o
f Electronic
MNN
f Electronic
Nehru Natio
Deem
Allaha
Discrimin
mission Sys
per Report Su
By
arat Tandon
ardul Verma
hag Mishra

of Lecturer
cs & Commu
IT Allahaba
to the
cs & Commu
onal Institu
med Universi
abad 2110
ation Rece
stem
ubmitted
Mr. Asim M
unication E
ad

unication E
ute of techno
ity
004
eiver of M
Mukherjee
Engineering
Engineering
ology
MSK Opticcal
i
Department of Electronics and Communication Engineering
Motilal Nehru National Institute of Technology
Allahabad


CERTIFICATE


This is to certify that the term paper entitled MLSE Equalizers for Frequency Discrimination
Receiver of MSK Optical Transmission System submitted by Bharat Tandon (20055036),
Shardul Verma (20055038), Vihag Mishra (20055017) for the course of Optical Communication
in Electronic and Communication Engineering at Motilal Nehru National Institute of
Technology, is a bonafide record of their own work and has been successfully completed under
my supervision.


Date:


Mr. Asim Mukherjee
Lecturer
Department of Electronics and Communication Engineering
Motilal Nehru National Institute of Technology
Allahabad - 211004

ii
Abstract

This work illustrates the maximum-likelihood sequence estimator (MLSE) equalizers based on
either Viterbi algorithm or template matching temple matching (TM) for the equalization of
impairments imposed on the minimum shift keying (MSK) modulation formats in long haul
transmission without optical dispersion compensation. The TM-MLSE equalizer is proposed as a
simplified alternative for the Viterbi-MLSE equalizer.
The performance analysis is done on the following three factors: (i) Number of states, (ii)
distance covered on Standard Single Mode Fiber (SSMF) and (iii) Speed on the link. The
performance of 16-state Viterbi-MLSE equalizers for optical frequency discrimination receiver
(OFDR)-based optical MSK transmission systems for PMD mitigation is also numerically
investigated.

iii
Acknowledgement


We are extremely grateful to Mr. ASIM MUKHERJEE for his revered guidance and
cooperation, which led to the completion of this Term Paper. Without his constant appraisal and
efforts, this task would have been a mere dream. He was always there to help us throughout this
project. He provided us with all the necessary resources and guidance during the project which
helped us to complete the project successfully.
Finally, we deem it a great pleasure to thank one and all that helped us directly or indirectly in
carrying out this term paper project work. We are also thankful to our colleagues and friends for
their support.

Date: 11
th
November, 2008

Place: Allahabad

Bharat Tandon
Shardul Verma
Vihag Mishra





iv
Contents

Sl no. Chapter Page no
Certificate i
Abstract ii
Acknowledgement iii
1 Introduction

01
2 Minimum Shift Keying
2.1 Minimum Shift Keying (MSK)
2.2 Advantages of MSK

03
04
04
3 MLSE and State Trellis implementation of Viterbi
Algorithm
3.1: Maximum Likelihood Sequence Estimator (MLSE)
3.2: State Trellis Structure
3.3: Viterbi Algorithm
3.4: Explanation of Viterbi Algorithm
3.5: A Viterbi Example
3.6: Viterbi in MLSE
06

07
07
08
09
10
11
v

4


5










6


7
Implementation of Viterbi Algorithm: Hidden Markov
Process
4.1: MATLAB code

MLSE Equalization in Optical Communications
5.1: Template Matching Algorithm
5.2: Intersymbol Interference (ISI) in Optical Systems
5.3: Construction of State Trellis Structure
5.4: MLSE Equalizers for OFDR-BASED MSK Systems
5.4.1 OFDR as Optical Front-End
5.4.2 Comparison of MLSE Equalizer based on Viterbi
Algorithm and Template Matching Algorithm
5.5 Noise Distribution
5.6 Performance Analysis
5.7 Suggestions on Improvement of Performance

Scopes and Conclusion
6.1 Conclusion
6.2 Future Proposals

REFERENCES
13

14

15
16
16
16
17
17
17

18
19
20
21
22
22

23
1






Chapter 1
Introduction









2
The paper MLSE Equalizers for Frequency Discrimination Receiver of MSK Optical
Transmission System by Le N. Binh, Member, IEEE, Member, OSA, Thanh Liem Huynh, K. K.
Pang, Life Senior Member, IEEE, and T. Sivahumaran, Student Member, IEEE describes
maximum-likelihood sequence estimator(MLSE) equalizers based on either Viterbi algorithm or
template matching temple matching (TM) for the equalization of impairments imposed on the
minimum shift keying (MSK) modulation formats in long haul transmission without optical
dispersion compensation. The TM-MLSE equalizer is proposed as a simplified alternative for the
Viterbi-MLSE equalizer. It is verified that the Viterbi-MLSE equalizer can operate optimally
when noise approaches a Gaussian distribution. Simulation results of the performances of the two
MLSE equalizers for optical frequency discrimination receiver-based optical MSK systems are
described. The transmission performance is evaluated in terms of:
1) The chromatic dispersion (CD) tolerance for both Viterbi-MLSE and TM-MLSE
equalizers;
2) Transmission distance limits of Viterbi-MLSE equalizers with various number of states;
3) The robustness to fiber polarization mode dispersion (PMD) of Viterbi-MLSE equalizers;
4) Performance improvements for Viterbi-MLSE equalizers when utilizing sampling
schemes with two and four samples per bit over the conventional single sample per bit.

The potential of Viterbi-MLSE equalizers in fiber-optic communications is reflected by the
effectiveness in combating the severe ISI caused either by fiber dispersion impairments or by
narrowband filtering of optical and electrical filters. Thus, Viterbi-MLSE equalizers have been
adopted into optical fiber communications.





3







Chapter 2
Minimum Shift Keying








4

2.1: Minimum Shift Keying (MSK)
In digital modulation, minimum-shift keying (MSK) is a type of continuous-phase frequency-
shift keying that was developed in the late 1960s. MSK is encoded with bits alternating between
quarternary components, with the Q component delayed by half the symbol period. It encodes
each bit as a half sinusoid. This results in a constant-modulus signal, which reduces problems
caused by non-linear distortion. In addition to being viewed as related to OQPSK, MSK can also
be viewed as a continuous phase frequency shift keyed (CPFSK) signal with a frequency
separation of one-half the bit rate.

Figure 1: Signal using MSK Modulation
The resulting signal is represented by the formula where a
I
(t) and a
Q
(t) encode the even and odd
information respectively with a sequence of square pulses as of duration 2T.
s(t) = a
I
(t)cos (t/2T) cos (2f
c
t) - a
Q
(t)sin (t/2T) sin (2f
c
t)
2.2: Advantages of MSK
Most of the recent studies have focused either on onoff keying (OOK)/amplitude-shift keying
(ASK) or DPSK modulation formats. Very few works have been reported on the performance of
MLSE equalizers for optical transmission systems employing minimum shift keying (MSK)
format.
In OOK and DPSK transmission systems, optical filters are used to reduce the effects of the
accumulated amplified spontaneous emission (ASE) noise. However, these filters usually have
large bandwidths (with reference to the signal bandwidth) in order to avoid filtering high-power
5
spectral sidebands of optical signals. As a result, the ASE noise is not greatly suppressed; thus
degrading the optical signal-to-noise ratio (OSNR) level for the same signal power. In addition,
the distributions of OOK or DPSK noise-corrupted received signals do not follow Gaussian
profiles. This avoids the Viterbi-MLSE equalizer from its optimal performance, and hence, the
exact noise distribution is required increasing the complexity of the MLSE receiver.
These problems, however, can be effectively mitigated by using MLSE equalizers in
combination with MSK format and narrowband optical filtering. A novel receiver scheme, the
optical frequency discrimination receiver (OFDR) scheme, has been proposed for 40 Gb/s optical
MSK systems and also reported. This receiver scheme enables a breakthrough dispersion
tolerance of up to 340 ps/nm for 1-dB power penalty. Combined with OFDR, the Viterbi-
MLSE equalizer serves as the post-detection electronic equalizer while OFDR serves as the
optical front-end. In addition, narrowly filtered noise is shaped back to a Gaussian distribution,
thus enabling the optimal performance for Viterbi algorithm as well as reducing the complexity
of the receiver.
The performance of the scheme is evaluated in terms of:
1) the CD tolerance for both Viterbi-MLSE and TM-MLSE equalizers;
2) transmission distance limits of Viterbi-MLSE equalizers with various number of states;
3) the robustness to fiber polarization mode dispersion (PMD) of Viterbi-MLSE equalizers; and
4) performance improvements for Viterbi-MLSE equalizers when utilizing sampling schemes
with two and four samples per bit over the conventional single sample per bit.







6








Chapter 3
MLSE and State Trellis
implementation of Viterbi
Algorithm




7
3.1: Maximum Likelihood Sequence Estimator (MLSE)
MLSE is a well-known technique in wireless and wire-line communications for equalization and
detection of transmitted digital signals. A MLSE receiver determines a sequence as the most
likely transmitted sequence a when the conditional probability P
r
(y|a) is maximized, where y is
the received noise-corrupted signal. If y is corrupted by a noise vector n which is modeled as an
additive Gaussian source (i.e., y=a+n), the maximization of the conditional probability is
equivalent to the minimization operation of the Euclidean distance d, given by
d `|y
k
-a
k
|
2
k

where k represents the index for the elements in the sequence. MLSE can be implemented
effectively by utilizing the Viterbi algorithm which is based on a state trellis structure.

3.2: State Trellis Structure

Figure 2: FSM modeling Block diagram
The FSM performs an one-to-one mapping from a signal symbol a to a coded signal symbol c, as
shown in Figure 2. The FSM can be a convolutional encoder or particularly in fiber-optic
communications, an optical fiber channel. A state trellis structure is defined with the FSM.
8

Figure 3: The state trellis structure
Demonstration of a state trellis structure at epoch of the FSM is given in Figure 3. For a
simplified explanation, several assumptions are made. These assumptions and according
notations are described as follows:
1) Current state B is the starting state of only two branches which connect to states E and F,
denoted as: b
BE
and b
BF
. In general, b
B*
represents all the possible branches starting from
state B in the state trellis. In Figure 3, these two branches b
BE
and b
BF
correspond to two
possible binary information symbols of 0 or 1, respectively. The notation c(B,a=0)
represents the encoded symbol at the FSM output for the branch BE(b
BE
) and corresponds
to a 0 symbol (a=0). Similarly, in the case of branch BF, b
BF
, i.e., c(B,a=1)
corresponds to a 1 symbol (a=1). Similar notations are applied for all branches in the
trellis structure.
2) There are only two branches ending at state E which are from states A and B and denoted
as: b
AE
and b
BE
. In general, b
*E
represents all possible branches ending at state E in the
trellis structure.
3) The number of states 2

in the trellis structure is determined by the number of memory


bits , also known as the constraint length.
3.3: Viterbi Algorithm
The Viterbi algorithm is a dynamic programming algorithm for finding the most likely
sequence of hidden states called the Viterbi path that results in a sequence of observed
9
events, especially in the context of Markov information sources, and more generally, hidden
Markov models. The forward algorithm is a closely related algorithm for computing the
probability of a sequence of observed events. These algorithms belong to the realm of
information theory.
The algorithm makes a number of assumptions. First, both the observed events and hidden events
must be in a sequence. This sequence often corresponds to time. Second, these two sequences
need to be aligned, and an instance of an observed event needs to correspond to exactly one
instance of a hidden event. Third, computing the most likely hidden sequence up to a certain
point t must depend only on the observed event at point t, and the most likely sequence at point t1.
These assumptions are all satisfied in a first-order hidden Markov model.
The terms "Viterbi path" and "Viterbi algorithm" are also applied to related dynamic
programming algorithms that discover the single most likely explanation for an observation. For
example, in statistical parsing a dynamic programming algorithm can be used to discover the
single most likely context-free derivation (parse) of a string, which is sometimes called the
"Viterbi parse".
The Viterbi algorithm was conceived by Andrew Viterbi in 1967 as an error-correction scheme
for noisy digital communication links, finding universal application in decoding the
convolutional codes used in both CDMA and GSM digital cellular, dial-up modems, satellite,
deep-space communications, and 802.11 wireless LANs. It is now also commonly used in speech
recognition, keyword spotting, computational linguistics, and bioinformatics. For example, in
speech-to-text (speech recognition), the acoustic signal is treated as the observed sequence of
events, and a string of text is considered to be the "hidden cause" of the acoustic signal. The
Viterbi algorithm finds the most likely string of text given the acoustic signal.
3.4: Explanation of Viterbi Algorithm
The Viterbi algorithm operates on a state machine assumption. That is, at any time the system
being modeled is in some state. There are a finite number of states, however large, that can be
listed. Each state is represented as a node. Multiple sequences of states (paths) can lead to a
given state, but one is the most likely path to that state, called the "survivor path". This is a
fundamental assumption of the algorithm because the algorithm will examine all possible paths
10
leading to a state and only keep the one most likely. This way the algorithm does not have to
keep track of all possible paths, only one per state.
A second key assumption is that a transition from a previous state to a new state is marked by an
incremental metric, usually a number. This transition is computed from the event. The third key
assumption is that the events are cumulative over a path in some sense, usually additive. So the
crux of the algorithm is to keep a number for each state. When an event occurs, the algorithm
examines moving forward to a new set of states by combining the metric of a possible previous
state with the incremental metric of the transition due to the event and chooses the best. The
incremental metric associated with an event depends on the transition possibility from the old
state to the new state. For example in data communications, it may be possible to only transmit
half the symbols from an odd numbered state and the other half from an even numbered state.
Additionally, in many cases the state transition graph is not fully connected. A simple example is
a car that has 3 states forward, stop and reverse and a transition from forward to reverse is
not allowed. It must first enter the stop state. After computing the combinations of incremental
metric and state metric, only the best survives and all other paths are discarded. There are
modifications to the basic algorithm which allow for a forward search in addition to the
backwards one described here.
Path history must be stored. In some cases, the search history is complete because the state
machine at the encoder starts in a known state and there is sufficient memory to keep all the
paths. In other cases, a programmatic solution must be found for limited resources: one example
is convolutional encoding, where the decoder must truncate the history at a depth large enough to
keep performance to an acceptable level. Although the Viterbi algorithm is very efficient and
there are modifications that reduce the computational load, the memory requirements tend to
remain constant.
3.5: A Viterbi Example
Considering two friends, Alice and Bob, who live far apart from each other and who talk
together daily over the telephone about what they did that day. Bob is only interested in three
activities: walking in the park, shopping, and cleaning his apartment. The choice of what to do is
determined exclusively by the weather on a given day. Alice has no definite information about
11
the weather where Bob lives, but she knows general trends. Based on what Bob tells her he did
each day, Alice tries to guess what the weather must have been like.
Alice believes that the weather operates as a discrete Markov chain. There are two states,
"Rainy" and "Sunny", but she cannot observe them directly, that is, they are hidden from her. On
each day, there is a certain chance that Bob will perform one of the following activities,
depending on the weather: "walk", "shop", or "clean". Since Bob tells Alice about his activities,
those are the observations. The entire system is that of a hidden Markov model (HMM).
Alice knows the general weather trends in the area, and what Bob likes to do on average. In other
words, the parameters of the HMM are known.
Alice talks to Bob three days in a row and discovers that on the first day he went for a walk, on
the second day he went shopping, and on the third day he cleaned his apartment. Alice has two
questions: What is the overall probability of this sequence of observations? And what is the most
likely sequence of rainy/sunny days that would explain these observations? The first question is
answered by the forward algorithm; the second question is answered by the Viterbi algorithm.
These two algorithms are structurally so similar (in fact, they are both instances of the same
abstract algorithm) that they can be implemented in a single function.
3.6: Viterbi in MLSE
The Viterbi algorithm is an effective method to implement the MLSE technique by sequentially
searching through the trellis structure. Operational principles of the Viterbi algorithm include
several processes that are explained as follows:
Calculation of branch and state metrics:
Branch metric:
At the n
th
sampling instance or the epoch n
th
, a branch metric is calculated for each of the
branches in the trellis structure. For example, considering the branch b
BE
, starting from the
current state B and ending at the next state E and corresponding to the 0 transmitted symbol
(a=0), the branch metric of branch BE bm
n
BL
is calculated as
bm
n
BL
= |y
n
-c
n
(B,a=0) |
2

12
where y
n
is the noise-corrupted received signal at epoch .
State metric:
As assumed previously, state E is the ending state of only two branches which start from states A
and B. Each of the branches ending at state E gives a state metric for state E. These state metrics
are computed by adding the state metric of the previous state to the calculated branch metric
obtained from above equation. The final value for the state metric of state E, sm
n
L
is achieved
with the minimization operation over all of the computed state metrics, as given in the following
equation:

Preferred path:
The preferred path is the branch which ends at state E and gives the state metric sm
n
L
for state E.
Trace-back process:
The process of calculating the state and the branch metrics continues along the state trellis
structure before terminating at epoch N
trace
that is usually referred as the trace-back length. The
value of N
trace
is normally selected to be 5 times of the constraint length . At N
trace
, the
terminating state which has the minimum state metric among 2

is selected. The previous state is
then identified by back tracing on the preferred path. The back-tracing process continues until
reaching epoch 1. Based on this final preferred path, the output symbol of either 0 or 1 is
determined. The data information is thus recovered.





13




Chapter 4
Implementation of Viterbi
Algorithm: Hidden Markov
Process





14
4.1: viterbi_hmm.m
MATLAB Code:
clc
clear
%This section shows how to set up a hidden Markov model and use it to
%generate data.
%Create the transition and emission matrices
TRANS = [.9 .1; .05 .95;]
EMIS = [1/6, 1/6, 1/6, 1/6, 1/6, 1/6;7/12, 1/12, 1/12, 1/12, 1/12, 1/12]
%Create the transition and emission guess matrices
TRANS_GUESS = [.85 .15; .1 .9]
EMIS_GUESS = [.17 .16 .17 .16 .17 .17;.6 .08 .08 .08 .08 08]

%Generate a random sequence of emissions from the model, seq, of length 20,
%using the function hmmgenerate
[seq, states] = hmmgenerate(20, TRANS, EMIS);
%The function hmmviterbi uses the Viterbi algorithm to compute the most
%likely sequence of states that the model would go through to generate the given sequence of
emissions
likelystates = hmmviterbi(seq, TRANS, EMIS);
%Compute the percentage of the time that the actual sequence states agrees
%with the sequence likelystates
sum(states==likelystates)/20
%This takes the emission and state sequences, seq and states, and returns
%estimates of the transition and emission matrices, TRANS_EST and EMIS_EST
[TRANS_EST, EMIS_EST] = hmmestimate(seq, states)
%This takes the guessed emission and state sequences, seq and states, and
%returns estimates of the transition and emission matrices, TRANS_EST and EMIS_EST

[TRANS_EST2, EMIS_EST2] = hmmtrain(seq, TRANS_GUESS, EMIS_GUESS)
%Computes the posterior state probabilities%
PSTATES = hmmdecode(seq, TRANS, EMIS)

% [PSTATES, logpseq] = hmmdecode(seq, TRANS, EMIS)
% [PSTATES, logpseq] = hmmdecode([3], TRANS, EMIS);
% exp(logpseq)
TRANS(1,:)*EMIS(:,3)






15





Chapter 5
MLSE Equalization in
Optical Communications








16
5.1: Template Matching Algorithm
There are different approaches to accomplish template matching. Some are faster performers
than others, and some find better matches.
The basic method of template matching uses a convolution mask (template), tailored to a specific
feature of the search matrix, which we want to detect. It is intuitively likely that the convolution
output will be highest at places where the matrix structure matches the mask structure, where
large image values get multiplied by large mask values.
This method is normally implemented by firstly picking out a part of the search matrix to use as
a template: We will call the search matrix S(x, y), where (x, y) represent the coordinates of each
co-ordinate in the search matrix. We will call the template T(x
t
, y
t
), where (x
t
, y
t
) represent the
coordinates of each pixel in the template. We then simply move the center (or the origin) of the
template T(x
t
, y
t
) over each (x, y) point in the search matrix and calculate the sum of products
between the coefficients in S(x, y) and T (x
t
, y
t
) over the whole area spanned by the template. As
all possible positions of the template with respect to the search matrix are considered, the
position with the highest score is the best position.
5.2: Intersymbol Interference (ISI) in Optical Systems
The optical fiber system which includes all ISI sources such as fiber CD and PMD as well as
tight optical filtering effects. These shortfalls cause waveform distortions and thus degrade
severely the system performance. When modeling an optical system as the FSM, all noise
sources and fiber nonlinearities should not be included in the system.
1) ISI Caused by Chromatic Dispersion
2) ISI Caused by Polarization Mode Dispersion
3) ISI Caused by Narrowband Filtering

5.3: Construction of State Trellis Structure
A state trellis structure, which has 2
2

states, 2 state bits/memory bits, and 2
2
+1
possible
branches, has been constructed based on the designated photonic transmission system. This
trellis structure is utilized in either Viterbi-MLSE equalizers or TM-MLSE equalizers.
17
5.4: MLSE Equalizers for OFDR-BASED MSK Systems
5.4.1: OFDR as Optical Front-End

Figure 4: Nonlinear equalizer integrated with OFDR scheme
Figure 4 shows the block diagram of MLSE equalizers incorporated with the OFDR for the
incoherent detection of MSK-modulated light waves. Two narrowband optical filters are used to
discriminate the USB and the LSB frequencies that correspond to logics 1 and 0 transmitted,
respectively. An ODL is introduced on one branch to compensate the differential delay t
d
=
|2f
D

2
L| between f
1
and f
2
, where f
D
= f
1
f
2
= R/2,
2
represents the GVD parameter of the
fiber and L is the fiber length. If the differential delay is fully compensated, optical light waves
in two paths arrive at two photodiodes simultaneously. It should be noted that these two
photodiodes are not connected. Optically filtered signals on each arm are then converted to
electrical domain by the photodiodes. These two separately detected electrical signals are then
sampled before being fed to the MLSE equalizer.
5.4.2: Comparison of MLSE Equalizer based on Viterbi Algorithm and Template Matching
Algorithm
The processing time and complexity of Viterbi-MLSE equalizers are exponentially increased
with the number of states. Thus, the implementation of Viterbi-MLSE equalizers at high speeds
such as 40 Gb/s may be beyond the reach of the processing speed of the present semiconductor
technology. However, with a small number of states, commercial chips for implementing
Viterbi-MLSE equalizers are available at 10 Gb/s operating rate.
18
Template Matching MLSE equalization technique is a single shot template matching algorithm.
The values of each element in the weighting vector for calculation of metric m in TM Algorithm,
over which, minimization is performed, is selected with the intention that when the template is
compared with the received samples, less weighting is given to the samples further away from
the middle sample.
The computational time is mainly due to the scanning process throughout the rows of the matrix.
However, the proposed TM-MLSE equalizer only requires the scanning of 2
+2
sequence patterns
instead of 2
2
+1
, leading to a considerable reduction of the processing time. Thus, the TM-MLSE
equalizer is promising for high speed electronic processing in the novel application for digital
photonic transmission systems.
5.5: Noise Distribution
Calculations of branch and state metrics in the Viterbi algorithm optimally perform when the
noise distribution complies with a normal/Gaussian distribution. If noise distribution departs
from the Gaussian distribution, the minimization process in the Viterbi algorithm gives sub-
optimum solutions. Thus, for optical OOK and DPSK systems in which large bandwidth optical
filters are used, more complex stages are required for estimating the exact noise distributions. It
is wellknown that noise distribution after the photodiodes in OOK and DPSK systems approach
the Maxwellian distribution.

Figure 5: Noise distribution following Gaussian profile due to narrowband optical filtering.

19
In optical MSK systems, the narrowband optical filter plays the main role in shaping the noise
distribution back to the Gaussian profile. Thus, the state-based Viterbi algorithm can achieve its
optimum performance and the equalizer structure is less complex than that of the OOK or DPSK
systems. In addition, a significant reduction in the computational time processing is also
accomplished.
5.6: Performance Analysis
With a small number of states (64 states), the uncompensated optical channel can equivalently
reach up to 928 km SSMF for 10 Gb/s transmission or 58 km SSMF for 40 Gb/s.

. Figure 6: Simulation setup for CD performance of Viterbi-MLSE and TM-MLSE equalizers for
40 Gb/s OFDR-based optical MSK systems
For 40 Gb/s MLSE equalizer for CD Mitigation:
1. Power level is kept low to avoid any non-linearity, controlled by EDFA1.
2. EDFA2 acts as a load and controls OSNR.
3. At 40-Gb/s bit rate, the one-sided bandwidth B is equal to 5.2 GHz with one-sided
bandwidths as 13% of the signal bandwidth for the filter.
4. For constraint length =5, trace back length=10(5)=50 and G has dimension [2
11
,11]
For 40-Gb/s Viterbi-MLSE Equalizers for PMD Mitigation:

The transmission link consists of a number of spans which are comprised of 100-km SSMF (D =
+17 ps/(nm.km), =0.2 dB/km) and 10-km DCF (D = -170 ps/(nm.km), =0.9 dB/km) . Input
power into each span (P
0
) is -3 dBm. The EDFA1 has a gain of 19 dB, hence providing input
20
power into the DCF to be -4 dBm which is lower than the nonlinear threshold of the DCF. The
transmission link consists of a number of spans which are comprised of 100-km SSMF 17 ps nm
km 0.2 dB km and 10-km DCF 170 ps nm km 0.9 dB km . Input power into each span is 3 dBm.
The EDFA1 has a gain of 19 dB, hence providing input power into the DCF to be 4 dBm which
is lower than the nonlinear threshold of the DCF. The 10 dB gain of EDFA2 guarantees the input
power into next span unchanged at -3 dBm value.
The advantages of using multiple samples per bit over the conventional single sample per bit in
the performance of Viterbi- MLSE equalizers are numerically investigated and obtained by the
Monte Carlo method. It can be observed that the increase from 2 to 4 samples per bit does not
offer any gain in the performance of the Viterbi- MLSE equalizer. Thus, two samples per bit is
preferred in order to reduce the complexity of the equalizer.
5.7: Suggestions on Improvement of Performance

One solution to improve the performance of the Viterbi- MLSE equalizer without increasing
much the complexity is to implement sampling schemes with multi samples within one bit
period. Although, a greater number of electronic samplers are required, they only need to operate
at the bit rate. Moreover, it can be observed that there is no noticeable improvement with more
than two samples per bit period. Hence, the complexity Viterbi-MLSE equalizers can be
affordable while its performance is significantly improved.








21







Chapter 6
Scopes and Conclusion












22
6.1: Conclusion
It has been shown that noise after tight optical filtering follows a Gaussian distribution. Thus, the
Viterbi algorithm can obtain its optimum performance without requiring more complex
algorithms for estimating the exact noise distribution. The combination of OFDR-based optical
MSK systems and Viterbi-MLSE equalizers, particularly with the use of multi-sample sampling
schemes, was found to be highly effective in combating the ISI effects.
The proposed equalizers significantly extend the reach of uncompensated transmission link. With
a small number of states (64 states), the uncompensated optical link can equivalently reach up to
approximately 928-km SSMF for 10-Gb/s transmission or 58-km SSMF for 40 Gb/s. The results
are comparable to the distance of 1040-km SSMF as reported for OOK/IM systems, in which up
to 8192 trellis states are required.
This work has proved the improvement in performance of Viterbi-MLSE equalizers by using the
sampling schemes with multiple samples per bit compared to the conventional single sample bit.
The novel schemes in which Viterbi-MLSE equalizers are integrated with OFDR-based optical
MSK systems significantly extend the uncompensated transmission distance as well as provide
high robustness to ISI impairment.
6.2: Future Proposals
In addition to bringing out the features of this work, the project also includes a novel algorithm
and a fully working MATLAB code for implementation of Hidden Markov Model based on
Viterbi Algorithm and finding out the sequences obtained on a predefined set of Transition and
Emission Matrices upto any number of specified states in the trellis.
Another aspect that cannot be left untouched is the implementation of Viterbi based system on a
FPGA and analysis of the design using Cadence and Synopsys Tools. The works on these titles is
still under stress and may be carried out in future. It will help enhance the speed of the complex
Viterbi Algorithm.
Also, the implementation of TM algorithm on multi samples per bit, if possible, can bring about
a revolution in the speed of the systems for MLSE based equalizers for OFDR in MSK
transmission systems.
23
REFERENCES
[1] MLSE Equalizers for Frequency Discrimination Receiver of MSK Optical Transmission
System; Le N. Binh, Member, IEEE, Member, OSA, Thanh Liem Huynh, K. K. Pang, Life
Senior Member, IEEE, and T. Sivahumaran, Student Member, IEEE
[2] Optical Fiber Communications, Second Edition; Gerd Keiser
[3] Digital and Analog Communication Systems; K.Sam Shanmugam