CO 2013-14

Modulation formats
António Teixeira
Motivations
 High speed electronics is becomming accessible
and available everywhere
 There is, both form the user or the operator, a
stress in lowering the cost per bit
◦ Sustain the price payed however increasing the data
rate
 Two paradigm shifts:
◦ Point to point links are not enough, therefore systems
should provide true network functionalities enabled
by mesh- capable ROADMs
◦ Progress in electronics has now met the optics
progressat the 40Gbit/s, allowing nowadays ultra high
spectral density
Enablers
 From the engeneering perspective
◦ The requirements
 High spectral efficiencies
 Usage of the advanced available electronics
◦ The current trends path
 Advanced modulation formats
 Coding
 Digital equalization
Research areas
 Per channel data rates of 10Gbit/s
◦ Electronic signal processing
 Simple FFE at the receiver
 MLSE (maximum likelihood sequence estimation)
 Starting is pre-distortion at the transmiter
 Coherent detection with electronic processing (make use of
phase information)
◦ General Target: increase robustness to fiber transmission
as GVD and/or PMD
 Per channel data rates 40Gbit/s+
◦ Electronic equalization with Low complexity electronic or
optical FFE (feed Forward equalizers) structures
 Targeting mitigation of linear and nonlinear effects:
 Modulation formats
 Line coding
SMF Bands

Under study

Preliminary usage –
Maintenance
Grids
Fibers and Effects
Cables and parameters
Modulator
 Eg. 40 GHz Phase and Intensity Modulators
 Mach Zehnder Modulator Structure http://www.iocc-optics.com/pkg3.jpg
http://www.fiberoptics4sale.com/wordpress/wp-content/uploads/2009/07/schematicofmachzehndermodulator.jpg

http://www.rie.shizuoka.ac.jp/~hsdhome/list/SPIE/fig4.jpg

http://www.ece.mcmaster.ca/faculty/nikolova/imgs/mach/pic.jpg
Field Power
transmission power

Vbias

field
transmission

Vbias
Exercises
Which modulations formats can this signal be
P representing? AM PSK
0 p 0 0 p p p
How to generate RZ PSK 33% ?
t And RZ PZK 50%?

Which are the main differences in amplitude

and frequency?

How to get no phase modulation and 66%?

How to generate NRZ PSK?

Which are the main differences
Modulators
IQ modulators
MZM
MZM
Transmitters
8 DPSK Example
16 QAM Example
16QAM examples
16 Square QAM
Modulation formats
Modulation formats
Modulation formats
 Reception

DLI – Delay Line interferometers

BD – Balanced Detector
Coherent reception
Coherent reception Gain from the LO
Homodyne/heterodyne
Classification of modulation formats
 Data Modulation Formats (DMFs)
 Symbol Alphabet Size
 Pulsed Modulation RZ Versus NRZ
Data Modulation Formats (DMFs)
 three physical attributes that can be used to
carry information:
◦ 1) intensity; (+++ common)
◦ 2) phase (including frequency) (++common); and
◦ 3) polarization.(--common)
 Active polarization management at the receiver (pol
changes throughout porpagation and time
 Also the RX sensitivity is similar
 However it can be used as a help dimension
 Improve normal propagation
 WDM- two orthogonal contiguous channels in different
polarizations have improved Xtalk resilience.
 In the same wavelength two channels can be transmited in an
“independent manner”.
Symbol Alphabet Size
 Using multilevel signaling
◦ M data bits are encoded on log2(M) symbols and are then transmitted
at a reduced symbol rate of R/M, where R is the bit rate
◦ Multiple amplitude (M-ASK)
 substantial back-to-back RX sensitivity penalty compared to binary on/off keying
(OOK)
◦ Multiple phase
 promising multilevel optical modulation formats is differential quadrature phase-
shift keying
 Benefits
◦ higher spectral efficiencies
◦ Exceeding electrical bandwidth
◦ lower symbol rates at a fixed data rate
 High CD tolerance
 Low DSP requirements
 costs
◦ reduced tolerance to noise
Symbol coding
 modulation formats with more than two symbols, correlative
coding and pseudo-multilevel modulation have received
substantially more attention in optical communications than
multilevel formats
◦ all symbols are transmitted at the bit rate
◦ additional degrees of freedom gained by using an increased symbol
alphabet
◦ introducing memory into the modulation scheme, also referred to as
line coding
 pseudo-multilevel DMFs (eg. CSRZ)
◦ more than two symbols are used to represent a single bit,
◦ the assignment of redundant symbols to transmitted bits is data
independent
 correlative coding DMFs
◦ assignment of symbols depends on the transmitted data information
◦ most important subcategory is partial-response DMFs
“1” “1”
“0”
Pulsed Modulation RZ Versus NRZ
Modulator technologies
 Three basic modulator technologies are widely in
use today
◦ directly modulated lasers
◦ electroabsorption modulators,
◦ and Mach–Zehnder modulators (MZMs).
 Eg.
 MZMs are almost exclusively used for transport
systems at 40 Gb/s and above, owing to
◦ their well-controllable modulation performance and
◦ the possibility of independently modulating intensity
and phase of the optical field
INTENSITY MODULATION
FORMATS
 NRZ-OOK
◦ The simplest
◦ using an MZM to generate NRZ, the modulator is biased at 50%
transmission
 the nonlinear compression of the sinusoidal MZM transfer function at
high and low transmissions can suppress overshoots and ripple
 RZ-OOK
◦ typically employ a pulse carver
 turn an optical NRZ format into its RZ equivalent.
 Typical carving:
 1) sinusoidally driving an MZM at the data rate between minimum and maximum
transmission results in optical pulses with a full-width at half-maximum of 50% of the
bit duration (a duty cycle of 50%), as shown in Fig. 5 (dashed);
 2) sinusoidally driving an MZM at half the data rate between its transmission minima
produces a pulse whenever the drive voltage passes a transmission maximum, as
visualized in Fig. 5 (solid). This way, duty cycles of 33% can be realized;
 3) sinusoidally driving a MZM at half the data rate between its transmission maxima
results in pulses with 67% duty cycle and with alternating phase. The resulting format
is called CSRZ (Section IV-C).
Pulse carver
INTENSITY MODULATION
FORMATS(2)
 CSRZ
◦ pseudo-multilevel modulation format,
characterized by reversing the sign of the optical
field at each bit transition
◦ is most conveniently realized by sinusoidal drive
of an MZM pulse carver at half the data rate
between its transmission maxima
 Since the optical field transfer function (dashed) of the
MZM changes its sign at the transmission minimum,
phase inversions between adjacent bits are produced
 the carrier at the optical center frequency vanishes,
giving the format its name
 spectrum exhibits characteristic tones at ±R/2
CSRZ
Correlative Coding and Partial-
Response Formats
 DB [Phase-Shaped Binary Transmission (PSBT) and Phased
Amplitude-Shift Signaling (PASS)]
◦ a phase change occurs whenever there is an odd number of 0-
bits between two successive 1-bits,
◦ main benefit of DB signals is their higher tolerance to CD and
narrow-band optical filtering compared to binary signaling
formats,
◦ TXs use a differentially precoded version of the data signal [43]
at the input
 exhibits a level change for every 0-bit contained in the original data
sequence and prevents error propagation after detection
 precoded sequence is converted into a three-level electrical signal by
means of severe electrical low-pass filtering (digitally or delay-and-add
filter (best)or analog filter with 3dB@25%BitR (taylorable eg to CD) )
 MZM designed for one fourth of the desired data rate may be used to combine the
functionality of low-pass filtering and modulation
 In the optical domain: Passing a binary signal with levels {−1, +1}, also known as phase-
shift keying (PSK, Section V-A), through a narrow-band optical bandpass filter produces
optical DB
DB
Correlative Coding and Partial-
Response Formats (2)
 AMI [DB Carrier Suppressed (DCS)]
◦ the phase changes for each 1-bit (even for adjacent 1-
bits), independent of the number of 0-bits in between
◦ typically implemented in RZ form (RZ-AMI).
◦ To obtain RZ-AMI, the data modulator is then
followed by an RZ pulse carver
◦ first generates binary (phase or intensity) modulation
in NRZ form ({−1, +1} or {0, +1}) and passes this
signal through an optical delay-and-subtract filter,
implemented by an optical delay interferometer (DI)
with delay τ ≤ T, where T is the bit period.
◦ Modified duobinary is a partial response format that
is generated using a delay-and-subtract circuit with a
2-bit delay (τ = 2T)
VSB
 Vestigial Sideband (VSB)and Single Sideband (SSB)
◦ where one sideband is completely suppressed, and in VSB signaling,
where an optical filter with a gradual rolloff is offset from the optical
carrier frequency to suppress major parts of one sideband
◦ A real-valued (chirp-free) double sideband signal is thus converted into
a complex-valued (chirped) VSB or SSB format.
◦ While SSB filtering is hard to implement in practice due to difficulties in
realizing the appropriate optical or electrical filter functions, optical VSB
has been successfully demonstrated on NRZ-OOK, RZ-OOK and
CSRZ-OOK
◦ In a WDM system,VSB filtering can either be done at the TX (i.e.,
before or in combination with multiplexing the WDM channels) or at
the RX (i.e., after or in combination with demultiplexing).
 Filtering at the TX allows for ultimate spectral compression and highly spectrally
efficient WDM transmission
 The advantage from VSB when filtered at the RX comes from reduced WDM
channel crosstalk for the desired sideband if unequal WDM channel spacing is
employed
Different modulation formats
DIFFERENTIAL PHASE
MODULATION FORMATS
 To detect information carried by the optical
field’s phase by means of square-law
detection, phase-to-intensity converting
elements have to be inserted into the optical
path prior to the photodiode
 Due to the absence of an optical phase
reference at such a (noncoherent) RX, the
phase reference has to be provided by the
signal itself: Each bit acts as a phase
reference for another bit, which is at the
heart of all DPSK formats
 Binary DPSK
◦ A 1-bit is encoded onto a π phase change, whereas a 0-bit is represented by the
absence of a phase change
◦ main advantage from using DPSK instead of OOK comes from a 3-dB RX sensitivity
improvement
 We are transmitting also in the 0’s.
◦ To perform optical phase modulation, one can either use a straight-line phase
modulator (PM) or an MZM
 A PM modulates the phase along a circle in the complex plane, leaving constant the intensity of
the phase-modulated light.
 However, since the optical phase directly follows the electrical drive signal, the speed of phase transitions is
limited by the combined bandwidth of driver amplifier and PM, and any overshoot or ringing in the drive
waveform manifests itself in phase distortions
 An MZM, which is symmetrically driven around zero transmission, modulates along the real axis
through the origin of the complex optical field plane [cf. Fig. 2(c)],
 which always produces exact π phase jumps at the expense of residual optical intensity dips at the locations
of phase transitions
 Since exact phase modulation is more important for DPSK than a constant optical intensity, practical DPSK
TXs are most conveniently implemented using an MZM as a PM
◦ Note the absence of a 0-bit rail in the eye diagrams, which is characteristic of phase-
modulated formats.
◦ the 3-dB sensitivity advantage
◦ of DPSK is only seen for balanced detection;
Balanced DPSK Rx
DI
 a DI is inserted in the optical path at the RX to convert the
differential phase modulation into intensity modulation.
◦ a DI splits the phase-modulated signal into two paths, which experience
a delay difference equal to the bit duration T (25 ps at 40 Gb/s) to let
two neighboring bits interfere at the DI output.
 At the DI output port A (the destructive port), the two optical fields interfere
destructively whenever there is no phase change, and constructively whenever
there is a phase change between subsequent bits, thus converting phase
modulation into intensity modulation
 Maintaining good interference is the most critical aspect in the design of DPSK RXs.
 Due to energy conservation within the DI, the second DI output port B (the
constructive port) yields the logically inverted data pattern
◦ In principle, one of the two
◦ DI output ports is sufficient to fully detect the DPSK signal (single-
ended detection).
 However, the 3-dB sensitivity advantage of DPSK is only seen for balanced
detection;
 The reason for the
 superior performance of balanced detection compared to single ended detection is the non-
Gaussian statistics of square-law detected Gaussian noise,
 the formats seen at ports A and B are AMI and DB
Generation from binary data
DQPSK
DQPSK
 It transmits the four phase shifts {0,+π/2,−π/2, π} at a symbol rate
of half the aggregate bit rate
 DQPSK TX is most conveniently implemented by two nested
MZMs operated as PMs
◦ Using this TX structure, one first takes advantage of the exact π phase
shifts produced by MZMs, independent of drive signal overshoot and
ringing
◦ this TX structure requires only binary electronic drive signals, which are
much easier to generate at high speeds than multilevel drive waveforms
◦ a pulse carver can be added to the structure to produce RZ-DQPSK
 DQPSK optical spectrum is identical to that of DPSK, but the
DQPSK spectrum is compressed in frequency by a factor of 2 due
to the halved symbol rate for transmission at fixed bit rate
◦ high spectral efficiencies in WDM systems
◦ increased tolerance to CD
◦ longer symbol duration compared to
◦ binary modulation formats makes DQPSK more robust to PMD
 DQPSK Rx
 At the RX, the DQPSK signal is thus first split into two equal parts,
◦ and two balanced RXs of the form depicted in Fig. 8(a), but
◦ with differently biased DIs, are used in parallel to simultaneously demodulate
the two binary data streams contained in the DQPSK signal
 DI delay has to equal the symbol duration for DQPSK demodulation,
◦ which is twice the bit duration
 This RX structure has a six times lower tolerance to frequency drifts
between transmit laser and DI compared to DPSK
 Eg. at 40 Gb/s and for a 1-dB penalty,
◦ DPSK tolerates :
 ±1.2 GHz of laser-to-DI frequency mismatch,
◦ whereas DQPSK :
 only allows for ±200 MHz.
◦ In general, feedback-controlled DI tuning within the RX is needed for both
DPSK and DQPSK.
OPTICAL NETWORKING WITH ADVANCED
MODULATION FORMATS
 Loss, Amplification, and
Noise
 Filter Narrowing and
WDM Crosstalk
 CD
 PMD
 Fiber Kerr Nonlinearity
◦ Nonlinear Transmission
Regimes
◦ Types of Nonlinear
Distortions
◦ Signal–Signal Interactions
◦ Dispersion Mapping
◦ Signal–Noise Interactions
 Nonlinearities and
Modulation Formats
 signal power falls below the detectability threshold of optical RXs,
◦ which makes systems require optical amplification.
 OAs can either be designed as
◦ lumped elements (e.g., erbium-doped fiber amplifiers), or
◦ distributed by introducing gain along the transmission fiber (e.g., distributed Raman amplification).
 The main impact of optical amplification is the generation of amplified spontaneous
emission (ASE)
◦ optical source of Gaussian noise.
◦ the ASE spectrum is typically constant (“white”) across the signal spectrum.
◦ that the statistical properties of ASE can be modified by nonlinear interactions durin
◦ If multiple OAs are concatenated to
◦ periodically compensate for fiber loss, ASE builds up in the system, captured by the optical signal-to-
noise ratio (OSNR)g fiber propagation
 The OSNR is typically defined as the average optical signal power divided by the ASE power, measured in both
polarizations and in a (bit rate independent) 12.5-GHz optical reference bandwidth
◦ In well designed fiber-optic RXs, the two beat noise terms dominate all other noise terms (e.g.,
thermal noise), and the RX is called beat-noise limited.
◦ The noise performance of a beat-noise limited RX is fully characterized by OSNRreq, which is the
OSNR required for a specified target bit error ratio (BER)
7% overhead FEC (typical for terrestrial systems) [winzer, JLT,2006]
General comments
 First, RZ formats in general require 1–3 dB less OSNR for
identical BER than their NRZ equivalents [83].
 For beat-noise limited RXs, this is mostly due to the reduced
impact of intersymbol interference (ISI) on RZ formats.
 RZ-AMI turns out to be particularly well performing among
all OOK formats, which has also been proven experimentally
 The spectrally narrow DB format exhibits a back-to-back
penalty of typically 1–2 dB compared to OOK formats. This
penalty is generated by the V-shaped eye opening [cf. Fig.
4(d)], leading to poor 0-bit detection performance.
 Using DPSK instead of intensity modulation, OSNR
requirements are significantly reduced.
 The gain of balanced-detection DPSK over OOK is generally
independent of the target BER and typically amounts to
around 3 dB (cf. RZ-OOK and RZ-DPSK in Table II).
B. Filter Narrowing and WDM
Crosstalk
 Apart from important SE-dependent nonlinearity
considerations the two dominant impairments arising
from dense WDM channel spacing in optically routed
networks are
◦ coherent WDM crosstalk and
 Similar to signal–ASE beat noise, coherent WDM crosstalk
between an optical signal field and the co-polarized residual
optical field of a neighbouring WDM channel after WDM
demultiplexing produces beat interference at square-law
detection
 Careful mux and demux design has to ensure that WDM
crosstalk is suppressed as much as possible, whereas signal
distortions through overly narrow optical filtering are kept to a
minimum.
◦ filter narrowing.
 this concatenation narrows the overall optical filter bandwidth
and distorts the signal.
Classification of Non-Linearities
Inter and intra channel impairments
PMD
PMD
OFDM
 The signal is transmited over several components
◦ Eg. Each carreir is modulated by each FFT component of
the signal
 OFDM is a special multi-carrier modulation (MCM),
in which its Nsc sub-carriers are orthogonal between
themselves.
 The MCM concept works as follows:
◦ the main high rate stream of data is divided into Nsc
parallel low rate streams,
◦ each low rate stream is transmitted in each sub-carrier,
◦ at the receiver side, the data is recovered from the sub-
carriers and the high rate data stream is recovered.
OFDM
Amplitude
spectrum
Transmitter and receiver
Frames OFDM
Cyclic pre/post and Extension
 Typical values

OFDM parameters
 OFDM Symbol duration

 OFDM bandwidth

Gi guard interval between consecutive OFDM symbols

Nb number of bits transmitted on each sub-carrier on one OFDM symbol
Nsp number of data symbols contained in one OFDM frame
cTs ratio between guard interval duration (Gi) and OFDM symbol duration (Ts)
Nsc total number of sub-carriers of the OFDM system
su ratio between used sub-carriers and FFT size in the OFDM system
Db Bit-rate per OFDM stream