You are on page 1of 18

1560

JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 29, NO. 10, MAY 15, 2011

OFDM for Flexible High-Speed Optical Networks


William Shieh, Member, IEEE, Fellow, OSA
(Invited Tutorial)

AbstractFast advancing silicon technology underpinned by


Moores law is creating a major transformation in optical fiber
communications. The recent upsurge of interests in optical orthogonal frequency-division multiplexing (OFDM) as an efficient
modulation and multiplexing scheme is merely a manifestation of
this unmistakable trend. Since the formulation of the fundamental
concept of OFDM by Chang in 1966 and many landmark works
by others thereafter, OFDM has been triumphant in almost all the
major RF communication standards. Nevertheless, its application
to optical communications is rather nascent and its potential success in the optical domain remains an open question. This tutorial
provides a review of optical OFDM slanted towards emerging
optical fiber networks. The objective of the tutorial is two-fold:
(i) to review OFDM fundamentals from its basic mathematical
formation to its salient disadvantages and advantages, and (ii)
to reveal the unique characteristics of the fiber optical channel
and identify the challenges and opportunities in the application of
optical OFDM.
Index TermsCoherent communications, digital signal processing, high spectral efficiency, orthogonal frequency-division
multiplexing (OFDM), software defined optical transmission
(SDOT).

I. INTRODUCTION

ECENTLY optical OFDM has attracted significant attention from the optical communications community,
and has certainly shown its potential to permeate into broad
ranges of applications across every level of optical networking,
from long-haul, to metro, to access, and to home networks.
There may be some side-line debate about the superiority
between optical OFDM and single-carrier systems, but the
central significance is that the rapidly rising interests in optical OFDM ushers in a new era of software-defined optical
communication (SDOT) in which various functionalities such
as optical dispersion mitigation, channel estimation, phase
estimation, performance monitoring, bandwidth provisioning,
data rate adaptation, or even modulation format, for that matter,
can be performed via software without human intervention.
Within the next decade, the electronic digital signal processing
(DSP) based transmission system is expected to fundamentally
change the way we see and operate optical networks today.
Manuscript received November 07, 2010; revised March 15, 2011; accepted
March 16, 2011. Date of publication March 24, 2011; date of current version
May 06, 2011. This work was supported by the Australian Research Council
(ARC).
The author is with the Center for Ultra-Broadband Information Networks
and National ICT Australia, Department of Electrical and Electronic Engineering, The University of Melbourne, Melbourne VIC 3010, Australia (e-mail:
shiehw@unimelb.edu.au).
Digital Object Identifier 10.1109/JLT.2011.2132115

The on-going optical OFDM study also heralds a trend that


fundamental communication theory has begun to be seriously
pursued in the field of optical communications that has its
birth from pure physics, and traditionally placed its focus on
photonic devices. DSP has been playing a vital role in wireless
communications and is on its way to enable the ultimately
flexible software-defined radio (SDR), which was once possible only theoretically [1]. On the other hand, the application
of electronic DSP to optical transmissions was hampered by
lack of sufficient computational power. Recent advances in
microelectronics, such as analog-to-digital converter (ADC),
digital-to-analog converter (DAC), and digital signal processor
(DSP), have brought about many novel optical subsystems
and systems via electronic DSP [2], [3]. It is anticipated that
more and more DSP functionalities will be applied to optical
communications, ranging from dispersion mitigation [2], [3]
to advanced error-correction codes [4], fueled by the ever-advancing silicon technology underpinned by the Moores law.
Since the formulation of the OFDM concept by Chang in
1966 and many landmark works by others thereafter, OFDM
has been triumphant in almost all the major RF communication
standards. Nevertheless, its application to optical communication is rather nascent and its potential success in optical domain
remains an open question. This tutorial provides a review of optical OFDM technology slanted towards the emerging optical
fiber networks enabled by fast advancing electronic DSP technology. In particular, we will focus on the two unique features
of optical OFDM: (i) extreme flexibility in system and network
design, and (ii) high scalability over ultrafast transmission line
rate. The organization of the tutorial is as follows. Section 2
deals with the background and basic concepts of OFDM starting
from its historical aspects, to its formal definition, to its various
important attributes such as cyclic prefix and spectral efficiency.
Section 3 illustrates various flavors of optical OFDM and their
corresponding applications. In Section 4, we identify the unique
channel characteristics of an optical fiber link in comparison to
the wireless counterpart, and show how this distinction plays an
important role in optical OFDM system design and realization.
Fiber nonlinearity has been recognized as the fundamental barrier limiting the ultimate system capacity. We dedicate Section 5
to discuss the impact of fiber nonlinearity on optical OFDM
transmission and its mitigation strategies. In Section 6, we show
that the introduction of optical OFDM in optical communications can have profound impact on emerging flexible optical networks that provide dynamic bandwidth provision, high spectral
efficiency optical add/drop, and plug-and-play link set up. One
of the major concerns of optical OFDM is whether a real-time
transceiver is feasible with state-of-the-art silicon technology.

0733-8724/$26.00 2011 IEEE

SHIEH: OFDM FOR FLEXIBLE HIGH-SPEED OPTICAL NETWORKS

1561

In Section 7, we review the progress of real-time OFDM implementation. Section 8 gives a summary of the tutorial.

II. OFDM FUNDMENTALS


The flexibility and scalability of OFDM originate from its
unique multicarrier nature. Before we move onto the discussion
about the system or network aspect of optical OFDM, a proper
understanding of OFDM basics is essential in order to appreciate how and why OFDM techniques can be applied to achieve
flexibility and scalability for future optical networks.

A. Historical Perspective of OFDM


The fundamental concept of OFDM was first formulated by
Chang in the seminal papers in 1966 [5], [6]. The field of OFDM
had been developed as a peripheral interest in military applications for a long time. This arose from lack of broadband applications for OFDM and powerful integrated electronic circuits to support the complex computation required by OFDM.
However, the arrival of broadband digital applications and maturing of very large scale integrated (VLSI) CMOS chips in the
1990s, have brought OFDM into the spotlight. In 1995, OFDM
was adopted as the European digital audio broadcasting standard (DAB), assuring its significance as an important modulation technology and heralding a new era of success for OFDM in
a broad range of applications. Among the important standards
that incorporate OFDM modulation technology are European
Digital Video Broadcasting (DVB), wireless local area networks
(WiFi, IEEE 802.11a/g), wireless metropolitan area networks
(WiMAX, 802.16e), asymmetric digital subscriber line (ADSL,
ITU G.992.1), and fourth generation mobile communications
technology, long-term evolution (LTE).
The application of OFDM to optical communications came
surprisingly late and relatively scantly compared with the RF
counterpart, although the same acronym of OFDM has long
been used to stand for optical frequency division multiplexing
in the optical communications community [7]. The first paper
on optical OFDM in the open literature was reported by Pan et
al. in 1996 [8], and some intermittent research on optical OFDM
can be also found in [9], [10]. However, the fundamental advantage of OFDM against channel dispersion had not been touched
in optical communications until 2001 when Dixon et al. proposed use of OFDM to combat modal-dispersion in multimode
fiber (MMF) [11]. Given the fact that an MMF fiber channel resembles a wireless channel in terms of multi-path fading, it is
not a surprise that the early body of work on optical OFDM is
concentrated for MMF fiber applications [11][15]. The recent
sudden rise of interest in optical OFDM is in large part attributed
to independent proposals of optical OFDM for long-haul application from three groups based on coherent optical OFDM (COOFDM) [16] and direct detection OFDM (DD-OFDM) [17],
[18]. At the time of writing, several groups have achieved spectral efficiencies of around 3.3 b/s/Hz [19][21] with a transmission distance beyond 2000 km [21], [22].

Fig. 1. Conceptual diagram for a generic multi-carrier modulation (MCM)


system.

B. OFDM Basics
1) Mathematical Formulation of an OFDM Signal: In
OFDM systems [5], [6], any signal
can be represented as

(1)
where
is the th information symbol at the th subcarrier,
is the waveform for the th subcarrier,
is the number
of subcarriers, and
is the symbol period.
is selected from a set of orthogonal functions in the sense that
(2)
or
is a Kronecker delta function. One of the most
where
popular choices of the function set is windowed discrete tones
given by

(3)

is the frequency of th subcarrier, and,


is the pulse
shaping function. In such a scheme, OFDM becomes a special class of multi-carrier modulation (MCM), a generic implementation of which is depicted in Fig. 1. The structure of a
complex multiplier (IQ modulator/demodulator), which is commonly used in MCM systems, is also shown in the figure.
The optimum detector for each subcarrier could use a filter
that matches the subcarrier waveform, or a correlator matched
to the subcarrier as shown in Fig. 1. Therefore, the detected

1562

information symbol
by

JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 29, NO. 10, MAY 15, 2011

at the output of the correlator is given

Using orthogonality condition of (6), and the convention that


(8)
and substituting (8) into (7), we have

(4)
where
is the received time-domain signal. The classical
MCM uses non-overlapped band-limited signals, and can be implemented with a bank of large number of oscillators and filters
at both transmit and receive ends [23]. The major disadvantage
of classical MCM is that it requires excessive bandwidth. This is
because, in order to design the filters and oscillators cost-effectively, the channel spacing has to be multiple times the symbol
rate, greatly reducing the spectral efficiency. Using orthogonal
subcarriers was first proposed by Mosier and Clabaugh in 1958
[24] to achieve high-spectral efficiency transmission. The orthogonality can be verified from straightforward correlation between any two subcarriers, given by

(5)
It can be seen that if the following condition
(6)
is satisfied, then the two subcarriers are orthogonal to each other,
i.e.,
only for
, and
for
. This
signifies that these orthogonal subcarrier sets, with their frequencies spaced at multiple of the inverse of the symbol rate can
be recovered with the matched filters in (4) without inter-carrier
interference (ICI), in spite of strong signal spectral overlapping.
2) Discrete Fourier Transform (DFT) Implementation of
OFDM: A fundamental challenge with OFDM is that a large
number of subcarriers are needed so that the transmission
channel appears to each subcarrier as a flat channel, in order
to recover the subcarriers with minimum signal processing
complexity. This leads to an extremely complex architecture
involving many oscillators and filters at both transmit and
receive end. Weinsten and Ebert first revealed that OFDM
modulation/demodulation can be implemented by using inverse
discrete Fourier transform (IDFT)/discrete Fourier transform
(DFT) [25]. Lets temporarily omit the index in (1) to focus
our attention on one OFDM symbol, and assume that we sample
at every interval of
, and the th sample of
from the expression (1) becomes
(7)

(9)
where stands for Fourier transform and
similar fashion, at the receive end, we arrive at

. In a

(10)
where

is the received signal sampled at every interval of


. From (9) and (10), it follows that the discrete value of
the transmitted OFDM signal
is merely a simple -point
IDFT of the information symbol , and received information
symbol
is a simple -point DFT of the sampled signal
at the receiver. It is worth noting that there are two critical
devices we have assumed for the DFT/IDFT implementation
which are (i) digital-to-analogue converter (DAC), needed to
convert the discrete value of
to the continuous analogue
value of
, and (ii) an analogue-to-digital converter (ADC),
needed to convert the continuous received signal
to the
discrete sample
. There are two fundamental advantages of
DFT/IDFT implementation of OFDM. First, because of existence of efficient IFFT/FFT algorithm, the number of complex
multiplications for IFFT in (9) and FFT (10) is reduced from
to
, which scales almost linearly with
the number of subcarrier,
[26]. Second, a large number
of orthogonal subcarriers can be generated and demodulated
without resorting to much more complex RF oscillators and
filters. This leads to a relatively simple architecture for OFDM
implementation when large number of subcarriers are required.
There has been research activity in using optical FFT circuits
for OFDM implementation [27], [28]. In this tutorial, we
focus our attention on electronic DSP based OFDM which has
advantages of high degree of adaptability and integrability.
3) Cyclic Prefix for OFDM: One of the enabling techniques
for OFDM is the insertion of cyclic prefix [29]. Let us first consider two consecutive OFDM symbols that undergo a dispersive
channel. For simplicity, each OFDM symbol includes only two
subcarriers with the fast delay and slow delay spread at , represented by fast subcarrier and slow subcarrier, respectively.
Fig. 2(a) shows that inside each OFDM symbol, the two subcarriers, fast subcarrier and slow subcarrier are aligned upon the
transmission. Fig. 2(b) shows the same OFDM signals upon reception where the slow subcarrier is delayed by against the
fast subcarrier. We select a DFT window containing a complete OFDM symbol for the fast subcarrier. It is apparent that
due to the channel dispersion, the slow subcarrier has crossed
the symbol boundary leading to the interference between neighboring OFDM symbols, formally, the so-called inter-symbol-interference (ISI). Furthermore, because the OFDM waveform in
the DFT window for slow subcarrier is incomplete, the critical

SHIEH: OFDM FOR FLEXIBLE HIGH-SPEED OPTICAL NETWORKS

1563

for symbol decision. Now we arrive at the important condition


for ISI-free OFDM transmission, given by
(11)
It can be seen that to recover the OFDM information symbol
properly, there are two critical procedures that need to be carried out, (i) selection of an appropriate DFT window, called DFT
window synchronization, and (ii) estimation of the phase shift
for each subcarrier, called channel estimation or subcarrier recovery. Both signal processing procedures are actively-pursued
research topics, and their references can be found in both books
[30], [31] and journal papers.
4) Spectral Efficiency for Optical OFDM: In direct-detection
optical OFDM (DDO-OFDM) systems, the optical spectrum is
usually not a linear replica of the RF spectrum, and therefore
the optical spectral efficiency is dependent on the detailed implementation. We will turn our attention to the optical spectral
efficiency for coherent optical OFDM (CO-OFDM) systems. In
CO-OFDM systems,
subcarriers are transmitted in every
OFDM symbol period of . Thus, the total symbol rate for
CO-OFDM systems is given by
(12)
Fig. 3(a) shows the spectrum of wavelength-division-multiplexed (WDM) channels each with CO-OFDM modulation,
and Fig. 3(b) shows the zoomed-in optical spectrum for each
wavelength channel. We use the bandwidth of the first null to
denote the boundary of each wavelength channel. The OFDM
bandwidth,
is thus given by
Fig. 2. OFDM signals (a) without cyclic prefix at the transmitter, (b) without
cyclic prefix at the receiver, (c) with cyclic prefix at the transmitter, and (d) with
cyclic prefix at the receiver.

orthogonality condition for the subcarriers (5) is lost, resulting


in an inter-carrier-interference (ICI) penalty.
Cyclic prefix was proposed to resolve the channel dispersion
induced ISI and ICI [29]. Fig. 2(c) shows insertion of a waveform, or so-called cyclic prefix by cyclic extension of the OFDM
waveform into a time interval,
. The time interval
is
called guard interval as it is used as a guard against ISI. As
shown in Fig. 2(c), the waveform in the guard interval is essentially an identical copy of that in the DFT window, time-shifted
by forward. Fig. 2(d) shows the OFDM signal with the
guard interval upon reception. Let us assume that the signal
has traversed the same dispersive channel, and the same DFT
window is selected containing a complete OFDM symbol for
the fast subcarrier waveform. It can be seen from Fig. 2(d),
a complete OFDM symbol for slow subcarrier is also maintained in the DFT window, because a proportion of the cyclic
prefix has moved into the DFT window to replace the identical
part that has shifted out. As such, the OFDM symbol for slow
subcarrier is an almost identical copy of the transmitted waveform with an additional phase shift. This phase shift is dealt with
through channel estimation and will be subsequently removed

(13)
where is the observation period. Assuming a large number of
subcarriers used, the bandwidth efficiency of OFDM is found
to be
(14)
The factor of 2 accounts for two polarizations in the fiber. Using
a typical value of 8/9 for , we obtain the optical spectral efficiency factor of 1.8 Baud/Hz. The optical spectral efficiency
gives 3.6 b/s/Hz if QPSK modulation is used for each subcarrier. The spectral efficiency can be further improved by using
higher-order QAM modulation [32], [33]. To practically implement CO-OFDM systems, the optical spectral efficiency will
be reduced due to the need of (i) pilot symbols and subcarriers
for channel and phase estimation, and (ii) a sufficient frequency
guard band between WDM channels to accommodate for laser
frequency drift. This guard band can be avoided by using orthogonality across the WDM channels, which will be further
discussed in the next subsection.
5) Orthogonal-Band-Multiplexed OFDM (OBM-OFDM):
The laser frequency drift of WDM channels can be resolved
by locking all the lasers to the common optical standard such
as an optical comb. In so doing, all the subcarriers across
the WDM channels can be orthogonal, i.e., the orthogonality

1564

JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 29, NO. 10, MAY 15, 2011

Fig. 4. IQ modulator for up-conversion of a complex-valued baseband signal


to a real-valued passband signal . The down-conversion follows the reverse process by reversing the flow of and .

(mixer) or IQ modulator/demodulator, which at the up-conversion can be expressed as

(15)

Fig. 3. Optical spectra for (a) wavelength-division-multiplexed CO-OFDM


channels, (b) zoomed-in OFDM signal for one wavelength, and (c) Orthogonalband-multiplexed OFDM (OBM-OFDM) without guard band.

condition of (5) is satisfied for any two subcarriers, even from


different WDM channels. We may consider the wavelength
channel as one of the subbands of the OFDM signal. As shown
in Fig. 3(c), the subcarrier in channel 1 (or subband 1) is orthogonal to another subcarrier in different channel (channel
2 or subband 2). We call this form of wavelength multiplexing
orthogonal-band-multiplexing (OBM-OFDM), namely, the
orthogonality applies to the subcarriers from different channels
or subbands [34]. The term of channel discussed here in
a broader sense could be generated in either RF, or optical,
or even digital domain. An optical or electrical filter with
bandwidth slightly larger than the channel bandwidth can be
used to select the desired channel. Inter-channel interference
is avoided through the orthogonality among OBM-OFDM
subcarriers. Consequently, no frequency guard band is necessary in such a scheme. However this requires DFT window
synchronization across WDM channels or subbands as we
discuss in Section II.B.4. To accommodate optical add/drop, a
small number of such OBM-OFDM subbands can be grouped
together to allow for frequency guard band between individual
group, in a similar manner to coherent WDM [35].
6) Complex and Real Representations of an OFDM Signal:
At the very beginning and end of digital signal processing, the
baseband OFDM signal is represented as a complex value, but
during transmission the OFDM signal becomes a real-valued
signal, more precisely, there is frequency up-conversion and
frequency down-conversion required for this complex-to-real
value conversion, or baseband to passband conversion. Mathematically, such transformation involves a complex multiplier

where the passband signal


is a real-valued signal at the
center frequency of
is the baseband complex-valued
signal, Re and Im stand for real and imaginary parts of a
complex quantity. Traditionally, the IQ modulator can be constructed with a pair of RF mixers and LOs with 90 degree shift
as shown in Fig. 4. The real-to-complex down-conversion of an
OFDM signal follows the reverse process of the up-conversion
by reversing the flow of the baseband signal and RF passband signal in Fig. 4. The mixer design can be implemented
in a mixed-signal integrated circuit (IC) up to 60 GHz with the
state-of-the-art silicon technology [36]. The IQ modulator/demodulator for optical OFDM up/down conversion resembles,
but is relatively more complicated than the RF counterpart [34].
Discrete multi-tone (DMT) is one of the interesting variations
of OFDM modulation [37], [38]. It uses the real-valued signal
even at the baseband. This is achieved through enforcing the
following condition to OFDM signal in (9):
(16)
namely, the information symbols for the subcarriers and

are complex conjugate to each other. The resulting


in (9) is thus a real-valued signal. Because both the transmitted
and received signals are of real value, IQ modulation and demodulation are not necessary. This leads to a much simpler and
cost-effective IC design with minimum RF feature [37], [38].
For this reason, DMT has been widely incorporated in commercial copper-based digital subscriber line (DSL) systems,
such as asynchronous DSL (ADSL) and very-high-data-rate
DSL (VDSL).
C. Frequency Offset and Phase Noise Sensitivity
The frequency offset and phase noise sensitivities have long
been recognized as two major disadvantages of OFDM. Both
frequency offset and phase nose lead to inter-carrier interference
(ICI). Because of its relatively long symbol length compared
to that of the single-carrier, OFDM is prone to both frequency
offset and phase noise [39]. However, it is stressed that these

SHIEH: OFDM FOR FLEXIBLE HIGH-SPEED OPTICAL NETWORKS

1565

two disadvantages obviously have not prevented OFDM from


gaining its popularity in RF communications. Frequency offset
sensitivity can be mitigated through frequency estimation and
compensation [40], [41], and phase noise sensitivity is resolved
primarily via careful design of RF local oscillators that satisfy
the required phase noise specification. Fortunately, the phase
noise specification can be readily met using the state-of-the-art
CMOS ASIC chips [42]. Nevertheless, it is important to understand the impairments from both frequency offset and phase
noise. Revisiting the phase noise problem is especially important to optical OFDM where the laser phase noise is often relatively large, despite the fact that there has been great research
effort of designing lasers specifically for the purpose of low
laser linewidth [43]. The impact of laser phase noise on optical OFDM is thus a critical issue especially for using higherorder constellation to achieve high spectral efficiency modulation [32], [33].
An OFDM signal with frequency offset and phase noise can
be generalized as

In (19), the first, second, and third term respectively corresponds


to the signal, inter-carrier interference (ICI), and the AWGN.
The ICI coefficient
represents the interference between two
subcarriers with a index difference of . Non-vanishing value
of
signifies the finite ICI as a result of either frequency offset
or phase noise. However, for the sake of explanatory simplicity,
we will investigate frequency offset and phase noise sensitivity
separately.
1) Frequency Offset Effect: When only frequency offset effect is considered, we set
to zero, and carry out the integration in (22), we have

(17)

(24)

where
is the frequency offset,
is the phase noise, and
is the additive white Gaussian noise (AWGN). According
to (4), the received information symbol will be
(18)
Substituting

of (17) into (18), we arrive at

(23)
where
, is the normalized frequency offset.
From the expression of (19) and (23), the variance of the interference
due to frequency offset can be computed as

where
is the variance the transmitted information symbol for
each subcarrier, and is the summation of all the ICI terms,
given by

(25)
Assuming that the interference is also additive white Gaussian
noise (AWGN), the effective signal-to-noise ratio (SNR) is
given by

(19)

(26)
where
is the original SNR without the frequency
offset, and
is the variance of the white noise . The analytical form of the bit-error-ratio (BER),
for a QPSK signal is
given by [44]
(27)

(20)
(21)
and are respectively white noise and ICI noise for the th
subcarrier. We have defined the ICI coefficient
as

(22)

We conduct simulation for a 256-subcarrier QPSK modulated OFDM system and compare the numerical simulation result with that obtained using the analytical expression of (27).
The result is shown in Fig. 5. It is observed from that the analytical approximation works well for a SNR below 12 dB and an
error-floor emerges prematurely for the analytical approximation, implying the AWGN approximation of ICI fails to work
well at the regime of high SNR.
2) Phase Noise Effect: Phase noise has been extensively
studied ever since OFDM has begun to be incorporated into the
standards. The fundamental work can be found in publication

1566

JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 29, NO. 10, MAY 15, 2011

where
stands for ensemble average, and is the 3 dB laser
linewidth, more precisely, the combined linewidth including
both transmit and receive lasers. Expressing
,
where
, and is the residual amplitude noise for each
OFDM symbol. From (29) the SNR of
can be represented
as
(31)

Fig. 5. BER performance of 256-subcarrier QPSK modulated QPSK systems


in the presence of different values of normalized frequency offset . Both the
results of numerical simulation and analytical calculation are shown.

and
are respectively the variance of the signal and
where
AWGN noise for the subcarrier, and
is the variance of the
noise including ICI and the amplitude fluctuation of the
.
of 1 is assumed for simplicity. We use SNR penalty
(dB) to
characterize the phase noise impairment given by
dB

(32)

is the SNR without phase noise impairwhere


ment.
(dB) can be also interpreted as the additional SNR
needed to achieve the same effective SNR or BER. The SNR
penalty can be given by [39]
dB

Fig. 6. SNR penalty against the laser linewidth normalized to


for both
4-QAM and 16-QAM OFDM systems. The penalty is measured at a BER of
.

by Pollet et al. [39]. Assume the frequency offset has been compensated and only phase noise is present, the expression of (22)
becomes

(28)
Denote
. is the also called common phase error
(CPE). Substituting (28) into (19), and making some simple rearrangement, we arrive at
(29)
where is the received information symbol after removing the
CPE. It can be seen that the phase noise has two major effects:
(i) CPE, , which rotates the entire constellation. This can be
estimated and the skewed constellation can be rectified through
simple collective rotation as shown in (29) and (ii) the ICI impairment which is the second term in (29), manifested by the
non-vanishing terms of
for nonzero . To assess the ICI
impairment of phase noise, we make further assumption that the
phase noise
observes the Wiener process such that [45]
(30)

(33)

It can be observed from (33) that the phase noise impact on


the OFDM signal is enhanced by the relatively long symbol
period of
compared with the single-carrier signal. Furthermore, the penalty is also proportional to , or the SNR where
the penalty is referenced, implying that higher-order modulation is more sensitive to phase noise as it requires higher SNR.
Fig. 6 shows SNR penalty as a function of linewidth for both
4-QAM and 16-QAM modulations. For 4-QAM modulation, to
limit the phase noise impairment below 1 dB at a BER of
,
the
should be kept below 0.01. Higher modulation such as
16-QAM tightens the requirement of
down to 0.002. Therefore, for coherent optical OFDM system, the laser linewidth is a
critical parameter, especially for higher-order modulation such
as 16-QAM and beyond [32], [33].
III. OPTICAL OFDM FLAVORS
Optical OFDM systems can be divided into five functional
blocks as shown in Fig. 7: (i) RF(or baseband) OFDM transmitter, (ii) electrical-to-optical (E/O) up-convertor, (iii) optical
fiber link, (iv) optical-to-electrical (O/E) down-converter, and
(v) RF (or baseband) OFDM receiver. Dependent on how the
E/O down-conversion is performed, optical OFDM can be classified into two broad categories: coherent optical OFDM (COOFDM) and directed detection optical OFDM (DDO-OFDM).
CO-OFDM represents the ultimate performance in receiver sensitivity, spectral efficiency and robustness against polarization
dispersion, but yet requires the highest complexity in transceiver
design. In the open literature, CO-OFDM was first proposed by
Shieh and Authaudage [16], and the concept of coherent optical
MIMO-OFDM was formalized by Shieh et al. in [46]. The early
CO-OFDM experiments were carried out by Shieh et al. for a
1000 km SSMF transmission at 8 Gb/s [47], and by Jansen et al.

SHIEH: OFDM FOR FLEXIBLE HIGH-SPEED OPTICAL NETWORKS

1567

Fig. 7. Conceptual diagram of generic optical OFDM systems consisted of five functional blocks. For simplicity, only CO-OFDM option is shown in electrical-tooptical up-converters and optical-to-electrical down-converters.

for 4160 km SSMF transmission at 20 Gb/s [48]. Another interesting and important development is the proposal and demonstration of the no-guard interval CO-OFDM by Yamada et al.
in [49] where optical OFDM is constructed using optical subcarriers without a need for the cyclic prefix. Nevertheless, the
fundamental principle of CO-OFDM remains the same, which
is to achieve high spectral efficiency by overlapping subcarrier
spectrum yet avoiding interference by using coherent detection
and signal set orthogonality.
Direct-detection optical OFDM (DDO-OFDM) has many
more variants than the coherent counter part. This mainly
stems from a broader range of applications for direct-detection
OFDM due to its lower cost. For instance, single-side-band
(SSB)-OFDM has been recently proposed by Lowery et
al. and Djordjevic et al. for long-haul transmission [17],
[18]. Tang et al. have proposed an adaptively modulated optical OFDM (AMOOFDM) that uses bit and power loading
showing promising results for both multimode fiber (MMF)
and short-reach SMF fiber link [13], [50]. The common feature for DDO-OFDM is of course using the direct-detection
at the receiver. Dependent on how the O/E up-conversion is
performed, we further divide the DDO-OFDM into two categories: (i) linearly-mapped DDO-OFDM (LM-DDO-OFDM)
where the optical OFDM spectrum is a replica of baseband OFDM, and (ii) nonlinearly-mapped DDO-OFDM
(NLM-DDO-OFDM) where the optical OFDM spectrum does
not display a replica of baseband OFDM. Offset SSB-OFDM
[17], Baseband Optical SSB OFDM [51], RF-tone assisted
OFDM [52], and virtual SSB-OFDM [53] belong to the class
of LM-DDO-OFDM; adaptively modulated optical OFDM
AMOOFDM [13], [50] and compatible OFDM [54] belong the
class of NLM-DDO-OFDM.

One of the major strengths of optical OFDM is its adaptability into a wide range of applications from access to core networks. Both CO-OFDM and DD-OOFDM have been proposed
for long haul transmission [16], [17]. But it is generally agreed
that CO-OFDM will have a major role in long-haul transmission where the receiver sensitivity, PMD sensitivity, and spectral efficiency are of critical importance, whereas DD-OOFDM
may find its niche in metro or access networks where cost is
primary concern. Orthogonal frequency-division multiple access (OFDMA) has been actively investigated for optical access networks [55], [56]. Apart from the conventional silica
MMF, polymer optical fiber is gaining more and more interest
for home networking or inter-cabinet interconnections. In this
application, optical DMT modulation similar to ADSL is proposed for transmission over polymer fibers for cost-reduction
[38], [57]. Optical OFDM is also proposed for optical free-space
indoor transmission via visible light [58] and outdoor transmission [59], [60].
Since this tutorial is slanted toward the potential impact of
optical OFDM on optical networks, we place the focus on optical fiber transmission where the major transmission media is
single-mode fiber (SMF). In the remainder of the paper, unless
specifically mentioned, the optical channel referred to is the
SMF link as opposed other forms of optical channels such as
multimode fibers (MMF) or free space channels. The detailed
description of other optical channels can be found in [12][15],
[58][61].
IV. UNIQUENESS OF THE OPTICAL CHANNEL
There is a fallacy that because of the extensive study on RF
OFDM that has been developed over the last 20 or so years,
optical OFDM will be an effortless one-to-one translation from

1568

JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 29, NO. 10, MAY 15, 2011

the wireless domain to the optical domain. As we shall see in


what follows, a clear understanding of the uniqueness of the optical channel and optical systems makes possible the most efficient design of the O-OFDM systems. Using single-mode-fiber
(SMF) optical communication systems and wireless systems as
examples, we lay out the following differences that have significant ramification on the OFDM design:
i) Channel model: A typical wireless channel undergoes a
Rayleigh process that can be modeled as summation of
multiple paths, given by [30], [31]
(34)
where
is a complex constant,
is the Rayleigh
fading process, and
is the delay for the th path. In
contrast, an optical single-mode fiber can be more conveniently modeled in the frequency domain for the two
polarization components in the fiber expressed as [46]
(35)
(36)

(37)
where
is the phase dispersion due to the fiber chromatic dispersion effect,
is the Jones matrix for the
fiber link representing the polarization dependent effect
including PMD and PDL [61], [62],
is the number of
PMD/PDL cascading elements represented by their birefringence vector and PDL vector
is the Pauli matrix vector [62]. The significance of the channel model is
that the majority of the channel dispersion
can
be first estimated and factored out for the channel estimation. Dynamic dispersion comes from the Jones matrix , but can be effectively reduced to a summation of
only a few taps of FIR model if its mean PMD value is
known. Therefore, dependent on the PMD value and the
data rate, channel estimation can be greatly simplified for
the optical OFDM systems [63].
The frequency-domain representation of the optical
channel goes beyond the superficial pedagogical transformation between frequency-domain and time-domain.
The channel model of (35) and (36) implies that the
channel dispersion in the SMF channel
can be estimated as
(38)
is the signal bandwidth. It follows that the
where
channel dispersion is proportional to the signal bandwidth, in sharp contrast to the wireless channel expression
of (34) where the channel dispersion is independent of
the signal bandwidth. As a result, the channel length, or
the product between the sampling rate and the channel

dispersion has quadratic dependence on the signal bandwidth as opposed to linear in wireless channels. This
implies that the sub-banding technique [34], [64] (such
as OBM-OFDM discussed in Section II) that partitions
the entire band into many subbands will reduce the computational complexity by a factor equal to the number
of subbands. This type of computational complexity
reduction via sub-banding does not apply to the wireless
channel.
ii) Channel nonlinearity: Wireless channel is of free-space,
and therefore does not possess any nonlinearity in the
channel. On the other hand, optical fiber is fairly nonlinear. Coupled with the fiber dispersion, PMD and PDL
effects, the optical channel is arguably more complicated than a wireless channel. Most often, there is no
closed-form analytical solution for nonlinear transmission in the optical fiber, and subsequently the numerical
solution to the nonlinear Schrdinger equation that describes the nonlinear wave propagation in the fiber is
required to analyze the performance [65]. At first glance,
OFDM plagued with high PAPR would not fit for the
optical fiber with high nonlinearity. Fortunately, the fiber
chromatic dispersion serves as the saving grace that
tends to mitigate against the nonlinearity [66], [67], and
recent experiments have shown successful transmission
of CO-OFDM at 100 Gb/s and beyond over 1000 km
SSMF fiber [19][21], [64], [68].
iii) Time variation of the channel characteristics. As important as the frequency dispersion (or frequency selectivity)
of the channel, the time selectivity or dispersion is another determining factor [30], [31]. Time dispersion is
defined as the rate at which the channel characteristics
are changing. In wireless systems, time dispersion is
characterized by the Doppler frequency from the fast
moving mobile users, whereas in fiber optic systems by
the polarization rotation due to mechanical disturbance
of the fiber optic link. The extent of the time selectivity
is characterized by the product of the Doppler frequency
in the wireless systems (or polarization rotation rate in
fiber optic systems) and OFDM symbol length, which
is about 0.04 for Universal Mobile Telecommunications
System (UMTS) or Wireless LAN (WLAN) environment
[31], or 5 10
for fiber optic systems (using 50 ns
symbol length, and 1 kHz polarization rotation rate).
Subsequently, the optical channel can be considered
as quasi-static. Efficient channel estimation algorithms
can be adopted by taking advantage of this important
phenomenon.
iv) Amplifier nonlinearity: This is an important factor that
may have not been commonly recognized. In wireless
systems, the major nonlinearity takes place in the power
amplifier. It is critical to have either high saturation power
RF amplifier, or operate with sufficient back-off. However, in fiber optic systems, the predominant amplifier is
Erbium doped fiber amplifier (EDFA), which is perfectly
linear. This is because the response time of the EDFA is
about millisecond (ms), and therefore any nonlinearity
faster than millisecond would practically vanish. This

SHIEH: OFDM FOR FLEXIBLE HIGH-SPEED OPTICAL NETWORKS

1569

is significant in the sense that in designing CO-OFDM


systems, when confronting the trade-off between optical
loss and RF loss, we would choose the former because
it is more linear. For instance, in the CO-OFDM transmitter design, we would choose to minimise the RF drive
voltage to the optical IQ modulator and optically amplify
the signal to compensate the excess loss of the optical IQ
modulator.
v) Tolerance to the out-of-band emission: In wireless systems, because of the scarcity of the spectrum, the RF
channels are packed as tightly as possible. Therefore,
there is stringent out-of-band emission requirement that is
enforced upon the OFDM transmitter. A stringent transmission mask is usually stipulated that details the maximum relative intensity the transmission emission should
be restricted to [69]. For instance, for the WiFi signal with
channel spacing of 20 MHz, the maximum out-of-band
emission intensity at 11, 20, and 30 MHz is respectively
dB,
dB, and
dB relative to the in-band density. It would be an easy task if an RF filter could be used
to remove any out-of-band emission at the transmitter in a
wireless system. The problem there is that the power amplifier is one of the major contributors to the overall chip
power consumption, and the introduction of an RF filter
following the power amplifier would induce significant
loss and decrease the chip power efficiency, and therefore
is commonly avoided in wireless systems. This places a
stringent requirement to the OFDM transmitter in terms
of the nonlinearity control. In contrast, in optical systems, wavelength division multiplexing (WDM) devices
are commonly used to combine multiple wavelengths,
and any out-of-band emission from the CO-OFDM transmitter is effectively eliminated. As such, the CO-OFDM
transmitter is more tolerant to out-of-band emission. This
fact should be taken advantage of when dealing with the
PAPR reduction in the O-OFDM systems.
V. NONLINEAR TRANSMISSION PERFORMANCE
OPTICAL OFDM

OF

A. Information Spectral Efficiency and Launch Power Limit


The maximum spectral efficiency of optical fiber transmission can be achieved by removing the frequency guard bands
between wavelength channels. The potential cross-talk due
to such dense wavelength packing can be resolved with the
concept of CO-OFDM [16]. In such systems, the CO-OFDM
wavelength channels can be either continuously spaced without
frequency guard band [19], [21], [49], [70], or densely spaced
with extremely small frequency guard band [20], [32]. Most recently, the nonlinear transmission performance of CO-OFDM
systems has been reported including analytical results for
single-channel transmission without consideration of chromatic
dispersion in [71], complete analytical expressions involving
summation of a large number of nonlinearity products in [66],
and computer numerical simulation of CO-OFDM system
performance in [72][74]. It would be of great interest to derive
concise closed-form solutions that capture the dependence of

the nonlinear performance on some major system parameters


such as chromatic dispersion and dispersion compensation
ratio. This body of analytical work on nonlinear system performance was pioneered in [75], [76] where nonlinear launch
power and information capacity are derived in closed form.
However, there are two limitations for the reports [75], [76]:
(i) they assume that nonlinear phase noise is generated independently in different spans, ignoring an important phase array
effect of the four-wave-mixing (FWM) products that accounts
for the interference among multiple spans [66], and (ii) they
only present the results for single polarization and thereby
cross-polarization nonlinear interaction is not included. We
derive closed-form analytical expressions for nonlinear system
performance of dual-polarization CO-OFDM systems using
an approach similar to [77] for single-polarization systems. In
particular, we find a closed-form expression for the nonlinear
noise density
given by [77]
(39)
(40)

(41)
are respectively the loss and chromatic disperwhere and
sion of the transmission fiber, is the residual dispersion ratio of
dispersion compensation, is the span length,
is the number
of spans, is the total signal bandwidth of the transmission sysis the signal (nonlinear noise) power density, i.e.,
tems,
the signal (nonlinear noise) power per unit bandwidth,
is the
corner frequency of the nonlinear noise, and
is the (noise)
enhancement factor accounting for the FWM noise interference
among different spans. In the derivation of (39), we have assumed that FWM noise is a small perturbation to the original
signal, and also subcarrier spacing is much smaller than the
walk-off frequency of the fiber [77].
The total noise is the addition of optical amplified spontaneous emission (ASE) noise and nonlinear FWM noise. As
such, the effective signal-to-noise ratio (SNR) is expressed as
[77]

(42)
(43)
is the ASE noise per polarization,
is the spontawhere
neous noise factor equal to half of the noise figure of the optical
amplifier
is the Planck constant, is the light frequency,
is the ASE noise per span. 2
in the denominator repand
resents the ASE noise for two polarizations.

1570

JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 29, NO. 10, MAY 15, 2011

Under the assumption of Gaussian noise distribution, the information spectral efficiency (defined as the maximum information capacity C normalized to bandwidth B) for dual-polarization is readily given by [78]

(44)
The optimum launch power density is another important parameter and is defined as the launch power density where the maximum SNR takes place. By simply differentiating SNR of (42)
over , and setting it to zero, we obtain the optimum launch
power density, the optimal SNR, and given by

Fig. 8. Information spectral efficiency versus number of 100-km spans with


various dispersion maps. CD: chromatic dispersion. CR: (CD) compensation
ratio.

B. Nonlinearity Mitigation
(45)

(46)
(47)

Equation (47) clearly shows the challenges of improving spectral efficiency by redesigning the fiber system parameters: to increase spectral efficiency by 2 bit/s/Hz, the dispersion needs to
be increased by a factor of 8, or the nonlinear coefficient be
decreased by a factor of 2.8, or number of spans be reduced by
a factor of 2, all of which are difficult of achieve. These challenges resulting from the parameter scaling are not unique to
CO-OFDM but rather fundamental to the fiber optical channel
[75], [76]. In a nutshell, it is of diminishing return to improve
the spectral efficiency by modifying the optical fiber system parameters. The only effective method to substantially improve
the spectral efficiency is to add more dimensions such as resorting to polarization multiplexing that leads almost a factor of
2 improvement, or fiber mode multiplexing by at least a factor
of two or more dependent on the capability of available digital signal processing (DSP) power. Fig. 8 shows the achievable
spectral efficiency for the three systems studied: (I) with CD of
16 ps/nm/without optical dispersion compensation, (II) with CD
of 16 ps/nm/km with 95% optical dispersion compensation, and
(III) with CD of 4 ps/nm/km without optical dispersion compensation. Other systems parameters are: 10-span of 100 km fiber
link; Fiber loss of 0.2 dB/km; nonlinear coefficient
W km ; noise figure of the amplifier of 6 dB; total system
bandwidth of 40 nm. The spectral efficiency for the systems I, II,
and III are respectively 9.90, 8.38, and 8.63 b/s/Hz at 10 spans.
This shows a total capacity of 49.5 Tb/s can be achieved for
10 100 km SSMF uncompensated EDFA-only dual-polarization systems within the C-band.

Despite the fact that CO-OFDM holds great promises of


flexible line rate adaptation, subwavelength accessibility and
bandwidth sub-banding to avoid the electronic bottleneck, fiber
nonlinearity stands in the way as one of the most prominent
obstacles to its practical implementation [66], [73], [79]. The
commonly studied approaches are as follows: (i) pre- and/or
post- compensation where the nonlinear phase noise is compensated at the transmitter or/and receiver [46], [79]; (ii) joint
cross-polarization nonlinearity cancellation, which is similar
to approach (i) but broadens the compensation to nonlinear
polarization rotation. (iii) nonlinear back-propagation where
the nonlinearity is unwrapped by back-propagating the received
signal toward the transmitter digitally [34], [80]; (iv) Volterra
nonlinear compensation where the nonlinearity is approximated
as the Volterra series and the nonlinearity is compensated iteratively [81]. There are several obstacles associated with all
the nonlinear compensation algorithms. First, there is unpleasant trade-off between the computational complexity and
compensation performance. Any good performance usually
requires a multiple of the original computational complexity.
Second, cross-phase modulation (XPM) nonlinearity is difficult
to compensate considering the practical difficulty in aligning
and transferring the large amount of waveform data between
different wavelength channels. In some situations where some
of the neighboring channels could have already been dropped,
their waveform data can not be easily accessed. Finally, PMD
also disperses the signal waveform stochastically, which can
not be predicted by back-propagation. This will limit the effectiveness of all the proposed nonlinear compensation algorithms.
C. Optimal Subband Bandwidth in Nonlinearity Performance
There is a common belief that CO-OFDM has inferior nonlinear performance in the fiber optic channel due to its high
PAPR. However, it has been shown that due to the uniqueness of chromatic dispersion in an optical fiber, properly designed CO-OFDM can in fact possess nonlinearity advantage
over coherent single-carrier (SC) for ultrahigh speed transport
at 100 Gb/s and beyond [67]. To elaborate this point, we use
400 Gb/s transport as an example where the signal occupies a
bandwidth of 120 GHz. We first ask the following important
question: how many subbands should the whole-spectrum of

SHIEH: OFDM FOR FLEXIBLE HIGH-SPEED OPTICAL NETWORKS

120 GHz be partitioned in order to achieve the best nonlinear


transmission performance? We assume here that each band is
filled with a single-carrier signal. Fig. 9 shows the transmission
results for 1000-km inline-dispersion compensated (IDC) and
non-dispersion compensated (NDC) systems studied in [67]. It
can be observed that the best performance is achieved when the
entire spectrum is partitioned into 8 and 32 subbands for IDC
and NDC systems, respectively. This shows that for NDC systems which is considered as the optimal transmission link for the
deployment of 100 Gb/s and beyond transport, the optimal subband bandwidth is about 3 GHz. In other words, when the signal
spectrum is beyond 3 GHz, nonlinear transmission performance
can be improved by partitioning the spectrum into multiple subbands of about 3 GHz. Subsequently, CO-OFDM systems, such
as multiband DFT-spread OFDM [67], or non-guard-interval
OFDM [82] can outperform coherent single-carrier transport in
nonlinearity tolerance at 100 Gb/s and beyond. The reason why
neither small number of subbands nor large number of subbands
achieves optimal nonlinearity performance is as follows: In the
case of small number of subbands, e.g., single carrier, the subband bandwidth becomes too broad. Although the PAPR within
each subband is low at launch, due to large walk off among
frequency components within each subband, the PAPR of each
subband will grow rapidly thus inducing nonlinearity penalty; In
the case of large number of subbands, the subband bandwidth
becomes too narrow. Neighboring subbands interact just as narrowly-spaced OFDM subcarriers, generating large inter-band
nonlinear crosstalk and thereby large penalty due to narrow subband spacing. Subsequently there exists a sweet spot in number
of subbands that gives the optimal nonlinearity performance as
shown in Fig. 9. It is noted that similar results on the optimal
symbol rate for coherent WDM systems has been obtained in
[83].
VI. OFDM FOR FUTURE OPTICAL NETWORKS
A. Software-Defined Optical Transmission (SDOT)
In response to the emergence of plethora of analog and digital
standards in the 1980s, the concept of SDR has been proposed
as a practical solution [1], namely, a software implementation of
the radio transceiver that can be dynamically adapted to the user
environment, instead of relying on dedicated hardware. Not surprisingly, a similar challenge arises in modern optical communications, where multiple advanced modulation formats [84] have
been proposed for the next generation 100 Gb/s Ethernet transport. This may signal a trend toward software-defined optical
transmission (SDOT) in which the transponder can be adapted
or reconfigured to multiple standards, or multiple modulation
formats [33], [61]. In particular, we envisage an intelligent software-defined optical transceiver, a similar concept to SDR, that
can
i) dynamically set up the physical link without human intervention, for instance, measuring the link loss, dispersion,
and setting up dispersion compensation modules;
ii) assign an optimal line rate for the link with a sufficient
margin;
iii) perhaps operate in multiple formats, that is, can elect to
run in either multi-carrier mode, or single-carrier mode;

1571

Fig. 9. Q factor as a function of number of sub-bands of a multiband


DFT-spread OFDM signal for (a) single-channel and 3-channel WDM transmission with 7-dBm launch power, and (b) single-channel transmission with
various input powers. Fiber link is 10 100 km SSMF fiber and IDC uses 96%
compensation.

iv) accurately report the channel conditioning parameters including OSNR, chromatic dispersion, PMD, or electrical
SNR, etc., which may identify the fault or predict link
failures before they occur.
We anticipate that electronic digital signal processing (DSP) enabled SDOT will lead to a fundamental paradigm shift from the
inflexible optical networks of today to robust, plug-and-play optical networks in the future. The introduction of SDOT places
focus on automation and reconfigurability and will inevitably
bring down the maintenance and operational cost, all of which
are critical to ensure the sustainability of information infrastructure that can scale up cost-effectively with explosive bandwidth
demand.
A conceptual diagram of SDOT is shown in Fig. 10. A salient
difference from conventional optical systems is the presence of
DAC/ADC and DSP in the architecture of the SDOT. The entire communication system is partitioned into analog and digital
domains by DAC/ADC. For optimization and application purposes, there are low-speed interactions among DSP, DAC/ADC,
and the front end. Again, SDOT promotes the migration from
analog to digital to enhance optical transmissions via dynamic
adaptation to the transmission channel and reconfiguration to an
appropriate modulation format.
B. Performance of Adaptable Optical Networks
Optical networks have been known to be rigid where the
line rate and channel spacing of each connection are very often
fixed and can not be adjusted dynamically. The introduction of
OFDM in optical communications can potentially play a profound role in creating next-generation optical networks offering

1572

JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 29, NO. 10, MAY 15, 2011

Fig. 10. Conceptual diagram of software-defined optical transmission (SDOT).

Fig. 12. Performance comparison between traditional fixed data rate and proposed adaptive data rate OFDM technology for the NOBEL-US network. The
loss ratio is compared for a statistical factor of 0.6 [86].

Fig. 11. SNR of OFDM subcarriers for (a) uniform loading, and (b) mixed
loading [85]. Power loading for the subcarriers has already used in (b).

adaptable line rate, subwavelength accessibility, and plug-andplay operation including automatic installation, monitoring, and
maintenance. We view the need for flexible optical networks as
the primary driver toward OFDM based high-speed transport.
In what follows we will review the current status of applying
OFDM techniques in flexible optical transmission systems and
networks.
Fig. 11 shows an experiment result of 10.7 Gb/s CO-OFDM
transmission over 1000-km SSFM fiber, with uniform 4-QAM
loading and mixed 4- and 8-QAM loading [85]. The distribution of the SNR across the subcarriers is shown in Fig. 11(a),
obtained through OFDM receiver signal processing. The SNR
values of the middle subcarriers are greater than those at the two
edges. It seems that most of the subcarriers with high SNR can
be further upgraded to high-order QAM. This corresponds to
the optical network operation scenarios for which at the beginning of the system life, there may be ample system margin to
spend, or for a connection with shorter reach, which both will
produce high SNR for the OFDM subcarriers. Because of the excessive SNR margin, we are able to load additional data into the
OFDM signal, to be exact, another OC-48 payload onto original
OC-192. In order to achieve that, a nominal 50% of the subcarriers should be loaded with 8-QAM from original 4-QAM. The
data rate increases from 10.7 Gb/s to 13.3 Gb/s. Fig. 11(b) shows
the SNR of the OFDM subcarriers with a mixture of 4-QAM
and 8-QAM loading, with 8-QAM filled in the middle of the
OFDM spectrum. The insets show the constellation diagrams
associated with each bit-loading band. The clear constellations
show that additional data has been successfully carried on this
OFDM transmission system. Note that we have achieved this by
using the same optical and electrical bandwidth and the same
launch power of
dBm. The adjustment only involves the
software embedded in the transceiver without a need for hardware adjustment [85].

The reach-dependent capacity of optical networks enabled by


optical OFDM is studied by using the NOBEL-US network in a
report by Bocoi et al. [86]. In OFDM enabled networks, the link
data rate can be dynamically adjusted according to the available
SNR, approximately 1 bit/s/Hz improvement for every 1/2 reduction of the reach. The detailed network model and associated methodology can be found in [86]. The loss ratio between
the total data lost and the total routed demands in the network
is used as a metric for performance comparison. The statistical
factor, i.e., the standard deviation normalized to the demand is
assumed to be 0.6. Fig. 12 shows the performance comparison
between OFDM-based and traditional network [86]. It can be
seen that the OFDM-based network significantly outperforms
the traditional one, for instance, at the loss ratio of 0.001, the allowable traffic capacity of OFDM-based network enjoys a 45%
gain.
The future optical network that fully embraces the flexibility
of optical OFDM has been elegantly elaborated by Jinno et
al. in [87] where they propose a novel, spectrum-efficient,
and scalable optical transport network architecture called
SLICE. As shown in Fig. 13, compared to the rigid fixed-grid
wavelength spaced optical network, the SLICE architecture facilitates sub-wavelength, super-wavelength, and multiple-rate
data traffic capability in a highly spectrum-efficient manner.
The bandwidth provision extends the adaptability from bandwidth-constraint assignment to adaptable bandwidth enabled by
bandwidth-variable wavelength cross-connects [87]. The subwavelength accommodation of OFDM based optical networks
offers the accessibility of ultrahigh speed signal with cost-effective lower-rate transponder [19]. This enhanced bandwidth
granularity can not be easily accommodated with conventional
single-carrier modulation format. It is worth pointing out that
the demonstrated subwavelength capability has been so far
limited to the transmitter and receiver only [19] and it is yet to
be seen how subwavelength capability can be exploited at an
ROADM node. Considering the increasing interest in achieving
power-efficiency towards green internet [88], there is a great
incentive to engage in adaptive optical networks where the
network capacity is better utilized via link-dependent rate adaptation and the associated electronic digital signal processing
is minimized by allowing subwavelength accessibility. The
impact of OFDM-enabled flexible optical networks on the DSP

SHIEH: OFDM FOR FLEXIBLE HIGH-SPEED OPTICAL NETWORKS

1573

Fig. 13. Spectrum assignment in SLICE: (a) conventional optical path network, and (b) SLICE [87].

requirements is application specific, the detailed discussion of


which is outside the scope of this paper. However, we would
like to make the following two general observations: (i) a flexible optical network requiring subband digital signal processing
does not increase the computation complexity as explained in
Section IV. (i), and (ii) to achieve such flexibility does increase
system complexity because of needing a feedback path from
receiver to transmitter for dynamic adaption.
VII. OFDM FOR 1 TB/S ETHERNET TRANSPORT
As 100 Gb/s Ethernet has become increasingly a commercial reality, the next logical pressing issue is the migration path
toward 1 Tb/s Ethernet transport. There have been pioneering
Tb/s transmission experiments employing optical time-division
multiplexing (OTDM) [89], [90]. But there are three key problems that have not been addressed very well by the OTDM approach. First, OTDM relies on precise timing alignment of many
tributaries for multiplexing and de-multiplexing. Whether the
TDM is performed in optical domain now, or perhaps in electronic domain in the future, the timing accuracy in the order
of femtosecond is challenging if not impossible. Second, the
demonstrated transmission reach of the Tb/s OTDM system is
limited, due to its extreme sensitivity to chromatic- and polarization-dispersion that calls for high-order optical compensation for either dispersion, which is costly if not infeasible.
Third, because of the short pulse employed for each tributary,
it is quite questionable whether migration to OTDM-based Tb/s
system will lead to any improvement in optical spectral efficiency. Recently, CO-OFDM has been proposed and demonstrated [19][21] as a promising alternative pathway toward
Tb/s transport that possesses high spectral efficiency, resilience
to tributary timing alignment and most important of all, the
channel dispersion.
The optical bandwidth required for conveying Tb/s information is about 340 GHz, dependent on the detailed OFDM design,
which translates to about 170-GHz electrical bandwidth. Here,
for the sake of simplicity, we have assumed polarization multiplexing and QPSK modulation for CO-OFDM systems. Such

Fig. 14. Conceptual diagram of the three-layer Opto-electronic integrated circuit (IC) hierarchy for a Tb/s CO-OFDM transceiver. Each box represents an
OFDM band at 100 Gb/s. The multiplexing is performed from bottom layer to
top layer whereas the demulitpelxing is in a reversed order.

broad electronic bandwidth is unlikely to be supported in the


next decade. Therefore, some kind of bandwidth down-scaling
is mandatory. We have proposed orthogonal-band-multiplexing
(OBM) scheme to sub-band the OFDM at the system level [19].
To migrate from 100 Gb/s to 1 Tb/s Ethernet poses a huge challenge as it requires 10 times the number of OFDM bands. To
solely rely on the optical OBM-OFDM entails many coherent
optical transmitter and receiver pairs, which may be cost-prohibitive. It is thus imperative to introduce an immediate integration process in the RF layer. Fig. 14 illustrates the concept
of such three-layer integration circuit (IC) architecture for a Tb/s
CO-OFDM transceiver. As an example, the CO-OFDM system
is assumed to transmit at a rate of 1.2 Tb/s, which is partitioned
into 12 bands. The entire mixed-circuit IC design involves three
layers of integration as follows: (i) in the baseband layer, baseband OFDM is generated at the rate of 100 Gb/s, (ii) in the RF
layer, each of the four basebands is multiplexed up to an RF carrier and combined electrically to a 400 Gb/s RF signal, and (iii)
in the photonics layer, the output signal from the RF layer will be
modulated onto three frequency-locked wavelengths and combined optically into a 1.2 Tb/s optical signal. The baseband layer
and RF layer are ideally implemented in mixed-circuit CMOS
ASICs, leveraging the recent progress made in millimeter-wave
CMOS technology [36]. The photonic integration circuits (PIC)

1574

JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 29, NO. 10, MAY 15, 2011

Fig. 15. Experimental setup for real-time CO-OFDM reception [92]. Inside the frame box is the signal processing block including one FPGA evaluation board
and 4 ADC boards, the picture of which is shown to the right of the experimental setup.

within the photonics layer or the integration of photonics and


CMOS is of great importance for the 1 Tb/s transport. Compared with OTDM, the CO-OFDM-based Tb/s systems employ
multiplexing and de-multiplexing of each OFDM band with RF
and wavelength combiners, eliminating the need for extremely
tight timing alignment. Furthermore, CO-OFDM has shown extreme dispersion resilience [19][21], without using complicated higher-order optical dispersion compensation.
VIII. REAL-TIME OPTICAL OFDM DEMONSTRATION
The computational complexity of OFDM is about the same
as coherent single-carrier that uses frequency-domain equalization [91]. Given that 100-Gb/s coherent single-carrier is almost
close to its commercial availability, it appears that a real-time
OFDM transceiver ASIC at 100 Gb/s should be in the near
horizon as well. Unfortunately, optical OFDM requires DACs
at about 50 GS/s which are not needed in coherent single-carrier and with the entire industry being focused on delivering
single-carrier 100-Gb/s transponders, there has not been significant activity in development of high-speed DACs. Relying on
off-shelf DACs that are bandwidth limited to around 10 GS/s,
the real-time demonstration of optical OFDM is lagging behind
that of coherent single-carrier [2]. Nevertheless, from a technological point of view, DACs are in fact easier to design than
ADCs, and therefore there is no fundamental technical hurdle to
their commercial manufacturability once there is a demand. As
we illustrated in Section VI, we anticipate a bright prospect for
optical OFDM in the emerging flexible optical networks. We
believe that the trend of migration from single-carrier to multicarrier in optical fiber communications is inevitable, but the
question of when this will happen is out of the domain of this
paper.
Due to lack of access to expensive ASIC platform, the realtime demonstration of optical OFDM has relied on the field-pro-

grammable gate arrays (FPGA) platform, and this has become


one of the hot topics in the field of optical OFDM. The motivation behind the FPGA-based real-time activity is to study unique
OFDM signal processing in general and to verify any novel algorithms in particular. The first real-time optical OFDM was
demonstrated at 3.1 Gb/s by Yang et al. in [92], followed by
a real-time dual-polarization CO-OFDM receiver at 6 Gb/s in
[93]. The experimental setup of [92] is shown in Fig. 15. A picture of the main real-time DSP block is shown at the right of
Fig. 15 consisting of 4 E2V ADC boards and one Altera Stratix
III FPGA evaluation board. A much higher line rate at 54 Gb/s
has been demonstrated by using multi-band structure and measuring at 3.6 Gb/s per band in real-time [94]. Simultaneously
there have been various real-time FPGA-based activities in optical OFDM transmitter at 8.36 Gb/s for long-haul transmission [95] and at 12.1 Gb/s per band for 109 Gb/s transmission
[96]. Rapid progress has been reported in direct-detection optical OFDM transceiver for 5.25-Gb/s 128-QAM-encoded [97]
and 11.25-Gb/s 64-QAM-encoded optical OFDM transmission
[98]. Most recently, a record real-time OFDM at 40 Gb/s was
demonstrated using state-of-the-art DACs at 50 Gs/s and 16
Virtex-5SX95T chips in a proof-of-concept OFDMA-PON experiment [99]. It is reasonable to anticipate that 100-Gb/s realtime optical OFDM is in the near horizon. These research works
are critical to investigate feasibility of novel DSP algorithms
and the dynamic behaviors of OFDM transceivers, all of which
provide valuable references to eventual CMOS ASIC implementation [94].
ACKNOWLEDGMENT
The author would like to express heartfelt thanks to many
colleagues for their insightful discussion and technical support
in the works of optical OFDM including, but not limited to R.
Tucker, R. Evans, G. Pendock, B. Krongold, A. Nirmalathas, S.

SHIEH: OFDM FOR FLEXIBLE HIGH-SPEED OPTICAL NETWORKS

Skafidas, H. Bao, W. Chen, X. Yi, Y. Tang, F. Zhang, Y. Ma,


Q. Yang, S. Chen, A. Al Amin, A. Li, K. P. Ho, I. Djordjevic,
X. Liu, S. Chandrasekhar, Y. K. Chen, N. Kaneda, F. Buchali,
M. Nazarathy, D. Marom, and T. Kawanishi. The author also
would like to thank the anonymous reviewers for many excellent suggestions.

REFERENCES
[1] J. Mitola, The software radio architecture, IEEE Commun. Mag., vol.
33, no. 5, pp. 2638, 1995.
[2] H. Sun, K. Wu, and K. Roberts, Real-time measurements of a 40 Gb/s
coherent system, Opt. Exp., vol. 16, pp. 873879, 2008.
[3] S. J. Savory, Digital coherent optical receivers: Algorithms and subsystems, IEEE J. Sel. Topics Quantum Electron., vol. 16, no. 5, pp.
11641179, Sep./Oct. 2010.
[4] I. Djordjevic, M. Arabaci, and L. Minkov, Next generation FEC for
high-capacity communication in optical transport networks, J. Lightw.
Technol., vol. 27, no. 16, pp. 35183530, Aug. 2009.
[5] R. W. Chang, Synthesis of band-limited orthogonal signals for multichannel data transmission, Bell Syst. Tech. J., vol. 45, pp. 17751796,
1966.
[6] R. W. Chang, Orthogonal Frequency Multiplex Data Transmission
System, U.S. Patent 3 488 445, Jan. 6, 1970.
[7] K. Kitayama, Highly spectrum efficient OFDM/PDM wireless networks by using optical SSB modulation, J. Lightw. Technol., vol. 16,
no. 16, pp. 969976, Jun. 1998.
[8] Q. Pan and R. J. Green, Bit-error-rate performance of lightwave hybrid AM/OFDM systems with comparison with AM/QAM systems in
the presence of clipping impulse noise, IEEE Photon. Technol. Lett.,
vol. 8, pp. 278280, 1996.
[9] Q. Shi, Error performance of OFDM-QAM in subcarrier multiplexed
fiber-optic transmission, IEEE Photon. Technol. Lett., vol. 9, no. 6,
pp. 845847, Jun. 1997.
[10] R. You and J. M. Kahn, Average power reduction techniques for
multiple-subcarrier intensity-modulated optical signals, IEEE Trans.
Commun., vol. 49, no. 12, pp. 21642171, Dec. 2001.
[11] B. J. Dixon, R. D. Pollard, and S. Iezekeil, Orthogonal frequency-division multiplexing in wireless communication systems with multimode
fiber feeds, IEEE Trans. Microw. Theory Techn., vol. 49, no. 8, pp.
14041409, Aug. 2001.
[12] N. E. Jolley, H. Kee, R. Rickard, J. M. Tang, and K. Cordina, Generation and propagation of a 1550 nm 10 Gb/s optical orthogonal frequency division multiplexed signal over 1000 m of multimode fibre
using a directly modulated DFB, in Proc. OFC, Anaheim, CA, 2005,
Paper OFP3.
[13] J. Armstrong and A. J. Lowery, Power efficient optical OFDM, Electron. Lett., vol. 26, no. 6, pp. 370372.
[14] J. M. Tang, P. M. Lane, and K. A. Shore, High-speed transmission
of adaptively modulated optical OFDM signals over multimode fibers
using directly modulated DFBs, J. Lightw. Technol., vol. 24, no. 1, pp.
429441, Jan. 2006.
[15] A. Lowery and J. Armstrong, 10 Gbit/s multimode fiber link using
power-efficient orthogonal-frequency-division multiplexing, Opt.
Exp., vol. 13, pp. 1000310009, 2005.
[16] W. Shieh and C. Athaudage, Coherent optical orthogonal frequency
division multiplexing, Electron. Lett., vol. 42, pp. 587589, 2006.
[17] A. J. Lowery, L. Du, and J. Armstrong, Orthogonal frequency division
multiplexing for adaptive dispersion compensation in long haul WDM
systems, in Proc. Opt. Fiber Commun. Conf., Anaheim, CA, 2006,
Paper PDP 39.
[18] I. B. Djordjevic and B. Vasic, Orthogonal frequency division multiplexing for high-speed optical transmission, Opt. Exp., vol. 14, pp.
37673775, 2006.
[19] Y. Ma, Q. Yang, Y. Tang, S. Chen, and W. Shieh, 1-Tb/s per channel
coherent optical OFDM transmission with subwavelength bandwidth
access, in Proc. OFC, San Diego, USA, 2009, Paper PDP C1.
[20] R. Dischler and F. Buchali, Transmission of 1.2 Tb/s continuous
waveband PDM-OFDM-FDM signal with spectral efficiency of 3.3
bit/s/Hz over 400 km of SSMF, in Proc. OFC, San Diego, CA, 2009,
Paper PDP C2.

1575

[21] S. Chandrasekhar, X. Liu, B. Zhu, and D. W. Peckham, Transmission of a 1.2-Tb/s 24-carrier no-guard-interval coherent OFDM superchannel over 7200-km of ultra-large-area fiber, in Proc. Eur. Conf.
Optical Commun., Vienna, Austria, 2009, Post-deadline Paper PD2.6.
[22] X. Liu, S. Chandrasekhar, B. Zhu, P. Winzer, A. Gnauck, and D.
Peckham, 448-Gb/s reduced-guard-interval CO-OFDM transmission over 2000 km of ultra-large-area fiber and five 80-GHz-Grid
ROADMs, J. Lightw. Technol., vol. 29, no. 4, pp. 483490, Feb.
2010.
[23] M. S. Zimmerman and A. L. Kirsch, AN/GSC-10 (KATHRYN) variable rate data modem for HF radio, AIEE Trans., vol. 79, pp. 248255,
1960.
[24] R. R. Mosier and R. G. Clabaugh, Kineplex, a bandwidth-efficient
binary transmission system, AIEE Trans., vol. 76, pp. 723728, 1958.
[25] S. B. Weinsten and P. M. Ebert, Data transmission by frequency-division multiplexing using the discrete fourier transform, IEEE Trans.
Commun., vol. COM-19, no. 5, pp. 628634, Oct. 1971.
[26] P. Duhamel and H. Hollmann, Split-radix FFT algorithm, IET Elect.
Lett., vol. 20, pp. 1416, 1984.
[27] K. Lee, C. T. D. Thai, and J. K. Rhee, All optical discrete Fourier
transform processor for 100 Gbps OFDM transmission, Opt. Exp., vol.
16, pp. 40234028, 2008.
[28] D. Hillerkuss, M. Winter, M. Teschke, A. Marculescu, J. Li, G. Sigurdsson, K. Worms, S. B. Ezra, N. Narkiss, W. Freude, and J. Leuthold,
Simple all-optical FFT scheme enabling Tbit/s real-time signal processing, Opt. Exp., vol. 18, pp. 93249340, 2010.
[29] R. Peled and A. Ruiz, Frequency domain data transmission using reduced computational complexity algorithms, in Proc. IEEE ICASSP,
Denver, CO, 1980, pp. 964967.
[30] S. Hara and R. Prasad, Multicarrier Techniques for 4G Mobile Communications. Norwood, MA: Artech House, 2003.
[31] L. Hanzo, M. Munster, B. J. Choi, and T. Keller, OFDM and
MC-CDMA for Broadband Multi-User Communications, WLANs and
Broadcasting. New York: Wiley, 2003.
[32] H. Takahashi, A. A. Amin, S. L. Jansen, I. Morita, and H. Tanaka,
8 66.8-Gbit/s coherent PDM-OFDM transmission over 640 km of
SSMF at 5.6-bit/s/Hz spectral efficiency, in Proc. Eur. Conf. Opt.
Commun., Brussels, Belgium, 2008, Paper Th.3.E.4.
[33] X. Yi, W. Shieh, and Y. Ma, Phase noise effects on high spectral efficiency coherent optical OFDM transmission, J. Lightw. Technol., vol.
26, no. 10, pp. 13091316, May 2008.
[34] W. Shieh, H. Bao, and Y. Tang, Coherent optical OFDM: Theory and
design, Opt. Exp., vol. 16, pp. 841859, 2008.
[35] A. D. Ellis and F. C. G. Gunning, Spectral density enhancement
using coherent WDM, IEEE Photon. Technol. Lett., vol. 17, no. 2,
pp. 504506, Feb. 2005.
[36] C. Doan, S. Emami, A. Niknejad, and R. Brodersen, Millimeter-wave
CMOS design, IEEE J. Solid-State Circuits, vol. 40, no. 1, pp.
144155, Jan. 2005.
[37] P. S. Chow, J. M. Cioffi, and J. A. C. Bingham, A practical discrete
multitone transceiver loading algorithm for data transmission over
spectrally shaped channels, IEEE Trans. Commun., vol. 43, no. 2/3/4,
pp. 773775, Feb./Mar./Apr. 1995.
[38] S. C. J. Lee, F. Breyer, S. Randel, H. P. A. van den Boom, and A. M. J.
Koonen, High-speed transmission over multimode fiber using discrete
multitone modulation [invited], J. Opt. Netw., vol. 7, pp. 183196,
2008.
[39] T. Pollet, M. Van Bladel, and M. Moeneclaey, BER sensitivity of
OFDM systems to carrier frequency offset and wiener phase noise,
IEEE Trans. Commun., vol. 43, pp. 191193, 1995.
[40] P. H. Moose, A technique for orthogonal frequency division multiplexing frequency offset correction, IEEE Trans. Commun., vol. 42,
no. 2/3/4, pp. 29082914, Feb./Mar./Apr. 1994.
[41] T. M. Schmidl and D. C. Cox, Robust frequency and timing synchronization for OFDM, IEEE Trans. Commun., vol. 45, no. 12, pp.
16131621, Dec. 1997.
[42] T. Maeda, N. Matsuno, S. Hori, T. Yamase, T. Tokairin, K. Yanagisawa, H. Yano, R. Walkington, K. Numata, N. Yoshida, Y. Takahashi,
and H. Hida, A low-power dual-band triple-mode WLAN CMOS
transceiver, IEEE J. Solid-State Circuits, vol. 41, no. 11, pp.
24812490, Nov. 2006.
[43] R. Wyatt and W. J. Devlin, 10 kHz linewidth 1.5 um InGaAsP external
cavity laser with 55 nm tuning range, IET Elect. Lett., vol. 19, pp.
110112, 1983.
[44] J. Proakis, Digital Communications, 3rd ed. New York: WCB/McGraw-Hill, ch. 5.

1576

[45] K.-P. Ho, Phase-Modulated Optical Communication Systems. New


York: Springer, 2005, ch. 4.
[46] W. Shieh, X. Yi, Y. Ma, and Y. Tang, Theoretical and experimental
study on PMD-supported transmission using polarization diversity in
coherent optical OFDM systems, Opt. Exp., vol. 15, pp. 99369947,
2007.
[47] W. Shieh, X. Yi, and Y. Tang, Transmission experiment of multi-gigabit coherent optical OFDM systems over 1000 km SSMF fiber,
Electron. Lett., vol. 43, pp. 183185, 2007.
[48] S. L. Jansen, I. Morita, N. Takeda, and H. Tanaka, 20-Gb/s OFDM
transmission over 4 160-km SSMF enabled by RF-pilot tone phase
noise compensation, in Proc. Opt. Fiber Commun. Conf., Anaheim,
CA, 2007, Paper PDP15.
[49] E. Yamada, A. Sano, H. Masuda, T. Kobayashi, E. Yoshida, Y.
Miyamoto, Y. Hibino, K. Ishihara, Y. Takatori, K. Okada, K. Hagimoto, T. Yamada, and H. Yamazaki, Novel no-guardinterval PDM
CO-OFDM transmission in 4.1 Tb/s (50 88.8-Gb/s) DWDM link over
800 km SMF including 50-GHz spaced ROADM nodes, in Proc.
Opt. Fiber Commun. Conf., San Diego, CA, 2008, Paper PDP8.
[50] X. Q. Jin, J. M. Tang, P. S. Spencer, and K. A. Shore, Optimization
of adaptively modulated optical OFDM modems for multimode fiberbased local area networks [invited], J. Opt. Netw., vol. 7, pp. 198214,
2008.
[51] D. F. Hewitt, Orthogonal frequency division multiplexing using
baseband optical single sideband for simpler adaptive dispersion
compensation, in Opt. Fiber Commun. Conf., Anaheim, CA, 2007,
Paper OME7.
[52] W. R. Peng;, X. Wu, V. R. Arbab, B. Shamee, L. C. Christen, J. Y.
Yang;, K. M. Feng, A. E. Willner, and S. Chi, Experimental demonstration of a coherently modulated and directly detected optical OFDM
system using an RF-Tone insertion, in Proc. Opt. Fiber Commun.
Conf., San Diego, CA, 2008, Paper OMU2.
[53] W. Peng, X. Wu, V. R. Arbab, B. Shamee, J. Y. Yang, L. C. Christen,
K. M. Feng, A. E. Willner, and S. Chi, Experimental demonstration
of 340 km SSMF transmission using a virtual single sideband OFDM
signal that employs carrier suppressed and iterative detection techniques, in Proc. Opt. Fiber Commun. Conf., San Diego, CA, 2008,
Paper OMU1.
[54] M. Schuster, S. Randel, C. A. Bunge, S. C. J. Lee, F. Breyer,
B. Spinnler, and K. Petermann, Spectrally efficient compatible
single-sideband modulation for OFDM transmission with direct
detection, IEEE Photon. Technol. Lett., vol. 20, no. 9, pp. 670672,
May 2008.
[55] L. Xu, D. Qian, J. Hu, W. Wei, and T. Wang, OFDMA-based passive optical networks (PON), in Proc. Technol. Digest. IEEE/LEOS
Summer Topical Meetings, 2008, pp. 159160.
[56] D. Qian, J. Hu, P. Ji, T. Wang, and M. Cvijetic, 10-Gb/s OFDMA PON
for delivery of heterogeneous services, in Proc. OFC, 2008, Paper
OWH4.
[57] S. C. J. Lee, F. Breyer, S. Randel, M. Schuster, J. Zeng, F. Huijskens,
H. P. A. van den Boom, A. M. J. Koonen, and N. Hanik, 24-Gb/s
transmission over 730 m of multimode fiber by direct modulation of
an 850-nm VCSEL using discrete multitone modulation, in Proc. Opt.
Fiber Commun. Conf., Anaheim, CA, 2007, Paper PDP6.
[58] J. Grubor, O. C. G. Jamett, J. W. Walewski, S. Randel, and K.-D.
Langer, High-speed wireless indoor communication via visible light,
in ITG Fachbericht 198. Berlin und Offenbach: VDE-Verlag, 2007,
pp. 203208.
[59] N. Cvijetic, D. Qian, and T. Wang, 10 Gb/s free-space optical transmission using OFDM, in Proc. OFC/NFOEC, 2008, Paper OThD2.
[60] I. B. Djordjevic, B. Vasic, and M. A. Neifeld, LDPC-coded OFDM
for optical communication systems with direct detection, IEEE J. Sel.
Topics Quantum Electron., vol. 13, no. 5, pp. 14461454, Sep./Oct.
2007.
[61] Shieh and Djordjevic, OFDM for Optical Communications. New
York: Academic , 2010.
[62] N. Gisin and B. Huttner, Combined effects of polarization mode
dispersion and polarization dependent losses in optical fibers, Opt.
Commun., vol. 142, pp. 119125, 1997.
[63] X. Liu and F. Buchali, Intra-symbol frequency-domain averaging
based channel estimation for coherent optical OFDM, Opt. Exp., vol.
16, pp. 2194421957, 2008.
[64] W. Shieh, Q. Yang, and Y. Ma, 107 Gb/s coherent optical OFDM
transmission over 1000-km SSMF fiber using orthogonal band multiplexing, Opt. Exp., vol. 16, pp. 63786386, 2008.
[65] G. Agrawal, Nonlinear Fiber Optics, 3rd ed. San Diego, CA: Academic, 2001.

JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 29, NO. 10, MAY 15, 2011

[66] M. Nazarathy, J. Khurgin, R. Weidenfeld, Y. Meiman, P. Cho, R.


Noe, I. Shpantzer, and V. Karagodsky, Phased-array cancellation of
nonlinear FWM in coherent OFDM dispersive multi-span links, Opt.
Exp., vol. 16, pp. 1577715810, 2008.
[67] Y. Tang, W. Shieh, and B. S. Krongold, DFT-spread OFDM for fiber
nonlinearity mitigation, IEEE Photon. Technol. Lett., vol. 22, no. 16,
pp. 12501252, Aug. 2010.
[68] S. L. Jansen, I. Morita, T. C. W. Schenk, and H. Tanaka, 121.9-Gb/s
PDM-OFDM transmission with 2-b/s/Hz spectral efficiency over 1000
km of SSMF, J. Lightw. Technol., vol. 27, no. 3, pp. 177188, Feb.
2009.
[69] Supplement to IEEE Standard for Information Technology Telecommunications and Information Exchange Between SystemsLocal and
Metropolitan Area NetworksSpecific Requirements. Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY)
Specifications: High-Speed Physical Layer in the 5 GHz Band, IEEE
Std 802.11a-1999, 1999.
[70] G. Goldfarb, G. F. Li, and M. G. Taylor, Orthogonal wavelength-division multiplexing using coherent detection, IEEE Photon. Technol.
Lett., vol. 19, no. 24, pp. 20152017, Dec. 2007.
[71] A. J. Lowery, S. Wang, and M. Premaratne, Calculation of power limit
due to fiber nonlinearity in optical OFDM systems, Opt. Exp., vol. 15,
pp. 1328213287, 2007.
[72] M. Mayrock and H. Haunstein, Monitoring of linear and nonlinear
signal distortion in coherent optical OFDM transmission, J. Lightw.
Technol., vol. 27, no. 16, pp. 35603566, Aug. 2009.
[73] X. Liu, F. Buchali, and R. W. Tkach, Improving the nonlinear tolerance of polarization-division-multiplexed CO-OFDM in long-haul
fiber transmission, J. Lightw. Technol., vol. 27, no. 16, pp. 36323640,
Aug. 2009.
[74] Y. Tang and W. Shieh, Coherent optical OFDM transmission up to 1
Tb/s per channel, J. Lightw. Technol., vol. 27, no. 16, pp. 35113517,
Aug. 2009.
[75] P. P. Mitra and J. B. Stark, Nonlinear limits to the information capacity
of optical fibre communications, Nauture, vol. 411, pp. 10271030,
2001.
[76] J. Tang, The channel capacity of a multispan DWDM system employing dispersive nonlinear optical fibers and an ideal coherent optical
receiver, J. Lightw. Technol., vol. 20, no. 7, pp. 10951101, Jul. 2002.
[77] X. Chen and W. Shieh, Closed-form expressions for nonlinear transmission performance of densely spaced coherent optical OFDM systems, Opt. Exp., vol. 18, pp. 1903919054, 2010.
[78] C. E. Shannon, A mathematical theory of communication, Bell Syst.
Tech. J., vol. 27, 1948.
[79] A. J. Lowery, Fiber nonlinearity pre- and post-compensation for longhaul optical links using OFDM, Opt. Exp., vol. 15, pp. 1296512970,
2007.
[80] E. IP, Nonlinear compensation using backpropagation for polarization-multiplexed transmission, J. Lightw. Technol., vol. 28, pp.
939951, 2010.
[81] R. Weidenfeld, M. Nazarathy, R. Noe, and I. Shpantzer, Volterra nonlinear compensation of 100 G coherent OFDM with baud-rate ADC,
tolerable complexity and low intra-channel FWM/XPM error propagation, in Proc. OFC, San Diego, CA, 2010, Paper OTuE3.
[82] T. Kobayashi, A. Sano, E. Yamada, E. Yoshida, and Y. Miyamoto,
Over 100 Gb/s electro-optically multiplexed OFDM for high-capacity
optical transport network, J. Lightw. Technol., vol. 27, no. 16, pp.
37143720, Aug. 2009.
[83] A. D. Ellis, I. Tomkos, I. A. K. Mishra, J. Zhao, S. K. Ibrahim, P. Frascella, and F. C. G. Gunning, Adaptive modulation schemes, in Proc.
Dig. LEOS Summer Topical Meeting, Newport Beach, CA, 2009, Paper
TUD3.2.
[84] P. J. Winzer and R. J. Essiambre, Advanced optical modulation formats, in Chapter in Optical Fiber Telecommunications V: B: Systems
and Networks, I. P. Kaminow, T. Li, and A. E. Willner, Eds. New
York: Academic, 2008.
[85] Q. Yang, W. Shieh, and Y. Ma, Bit and power loading for coherent
optical OFDM, IEEE Photon. Technol. Lett., vol. 20, no. 15, pp.
13051307, Aug. 2008.
[86] A. Bocoi, M. Schuster, F. Rambach, M. Kiese, C. Bunge, and B.
Spinnler, Reach-dependent capacity in optical networks enabled
by OFDM, in Proc. OFC, San Diego, CA, 2009, Paper OMQ4,
OMQ4.
[87] M. Jinno, H. Takara, B. Kozicki, Y. Tsukishima, Y. Sone, and S. Matsuoka, Spectrum-efficient and scalable elastic optical path network:
Architecture, benefits, and enabling technologies, in IEEE Commun.
Mag., Nov. 2009, pp. 6673.

SHIEH: OFDM FOR FLEXIBLE HIGH-SPEED OPTICAL NETWORKS

[88] J. Baliga, K. Hinton, and R. S. Tucker, Energy consumption of the


internet, in Proc. 32nd Australian Conf. Opt. Fibre Technol. COINACOFT, Jun. 2007, pp. 13.
[89] M. Nakazawa, Tb/s OTDM technology, in European Conference
on Optical Communications, Amsterdam, Netherland, 2001, Paper
Tu.L.2.3.
[90] H. G. Weber, S. Ferber, M. Kroh, C. Schmidt-Langhorst, R. Ludwig,
V. Marembert, C. Boerner, F. Futami, S. Watanabe, and C. Schubert,
Single channel 1.28 Tbit/s and 2.56 Tbit/s DQPSK transmission,
Electron. Lett., vol. 42, pp. 178179, 2006.
[91] B. Spinnler, Equalizer design and complexity for digital coherent
receivers, IEEE J. Sel. Opt. Quantum Electron., vol. 16, no. 5, pp.
11801192, Sep./Oct. 2010.
[92] Q. Yang, S. Chen, Y. Ma, and W. Shieh, Real-time reception of
multi-gigabit coherent optical OFDM signals, Opt. Exp., vol. 17, pp.
79857992, 2009.
[93] S. Chen, Q. Yang, Y. Ma, and W. Shieh, Real-time multi-gigabit receiver for coherent optical MIMO-OFDM signals, J. Lightw. Technol.,
vol. 27, no. 16, pp. 36993704, Aug. 2009.
[94] N. Kaneda, Q. Yang;, X. Liu, S. Chandrasekhar, W. Shieh, and Y. K.
Chen, Real-time 2.5 GS/s coherent optical receiver for 53.3-Gb/s subbanded OFDM, J. Lightw. Technol., vol. 28, no. 4, pp. 494501, Feb.
2010.
[95] Y. Benlachtar, G. Gavioli, V. Mikhailov, and R. I. Killey, Experimental investigation of SPM in long-haul direct-detection OFDM systems, Opt. Exp., vol. 16, pp. 1547715482, 2008.
[96] F. Buchali, R. Dischler, A. Klekamp, M. Bernhard, and D. Efinger,
Realisation of a real-time 12.1 Gb/s optical OFDM transmitter and
its application in a 109 Gb/s transmission system with coherent reception, in Proc. ECOC, 2009, Paper PDP 2.1.

1577

[97] X. Q. Jin, R. P. Giddings, E. Hugues-Salas, and J. M. Tang, Real-time


demonstration of 128-QAM-encoded optical OFDM transmission with
a 5.25 bit/s/Hz spectral efficiency in simple IMDD systems utilizing
directly modulated DFB lasers, Opt. Exp., vol. 17, pp. 2048420493,
2009.
[98] R. P. Giddings, X. Q. Jin, E. Hugues-Salas, E. Giacoumidis, J. L. Wei,
and J. M. Tang, Experimental demonstration of a record high 11.25
Gb/s real-time optical OFDM transceiver supporting 25 km SMF
end-to-end transmission in simple IMDD systems, Opt. Exp., vol. 18,
pp. 55415555, 2010.
[99] D. Qian, T. Kwok, N. Cvijetic, J. Hu, and T. Wang, 41.25 Gb/s realtime OFDM receiver for variable rate WDM-OFDMA-PON transmission, in Proc. OFC, San Diego, CA, 2010, Paper PDPD9.

William Shieh (S96M96) received the M.S. degree in electrical engineering


and the Ph.D. degree in physics from the University of Southern California,
Los Angles, in 1994 and 1996, respectively.
Since 2004, he has been with the Department of Electrical and Electronic Engineering, University of Melbourne, Melbourne, Australia. His current research
interests include OFDM techniques in both wireless and optical communications, coherent optical communication systems, and optical packet switching.
He has published more than 110 journal and conference papers, and submitted
14 U.S. patents (nine issued) covering areas of optical OFDM, polarization controller, wavelength stabilization in WDM systems, and Raman amplifier-based
systems and subsystems.
Dr. Shieh has been elected a Fellow of the Optical Society of America (OSA).

You might also like