OSNR sensitivity analysis on a 100 Gb/s PM-QPSK system.
Transcript of OSNR sensitivity analysis on a 100 Gb/s PM-QPSK system.
OSNR sensitivity analysis
on a 100 Gb/s PM-QPSK
system.
Santiago Pacheco Muñoz – Octubre 2013
Abstract
The aim of this thesis is to analyze the OSNR sensitivity on a 100 Gb/s optical
coherent PM-QPSK system against the use of different types of laser sources, looking
for the impact of the associated phase noise at different laser linewidth.
The drive for higher performance in optical fiber systems has focused its
interest in coherent detection, a field revived by advances in digital signal processing
(DSP). DSP-based phase and polarization management techniques make coherent
detection robust and practical. With coherent detection, the complex field of the
received signal is fully recovered, allowing compensation of linear impairments
including chromatic dispersion and polarization-mode dispersion using difital filters.
In addition, fiber nolinearities can also be compensated quasi-exactly.
Coherent PM-QPSK has been attracting considerable attention for 100G long
haul optical interfaces because of its OSNR performance (>2dB improvement over
direct detection formats), spectral efficiency, and tolerance to inter-symbol
interference.
This thesis firstly describes basic principles of fiber optic communications and
digital modulations in order to introduce the system object of our analysis. Then the
use of coherent optical systems is discussed, comparing other kind of system detection
and looking for the benefits of its use.
Thereafter, the Polarization multiplexing – QPSK modulation is introduced,
giving a short evolution on optical modulation and showing the main architecture of a
PM-QPSK transmitter. Subsequently performance and characteristics of 100 Gb/s
PM-QPSK systems are given focusing on its OSNR sensitivity, filtering tolerance and
dispersion.
In a second stage of the thesis, we define a model of a coherent PM-QPSK
system, explaining the principal parameters and modules that conforms the whole
composition. Once our starting point of the analysis, wich is the system, has been
introduced, we start the analysis on the sensitivity against different types of laser
linewidth, performing several simulations using different values on the range of 0 Hz
to 10 MHz and looking for the most interesting results.
In conclusion, this thesis presents, defines, describes and analyses a coherent
100 Gb/s PM-QPSK system in order to define its OSNR sensitivity against changes
on the laser linewidth between 0 Hz – 10MHz.
Contents
1 Fiber optics communication history 1
1.1 The birth of fiber optic systems . . . . . . . . . . . . . . . 2
1.2 Advantages of optical fiber communication. . . . . . . . . 3
2 Principles of fiber optic communication 6
2.1 Basic fiber optic communication system . . . . . . . . . . . 6
2.1.1 Transmission windows. . . . . . . . . . . . . . . . . 7
2.1.2 Fiber optic loss . . . . . . . . . . . . . . . . . . . . . 7
2.1.3 Types of fiber. . . . . . . . . . . . . . . . . . . . . . 8
2.1.4 Dispersion . . . . . . . . . . . . . . . . . . . . . . 10
2.1.5 Types of dispersion . . . . . . . . . . . . . . . . . 12
2.1.5.1 Modal dispersion. . . . . . . . . . . . . . . 12
2.1.5.2 Chromatic dispersion. . . . . . . . . . . . . 12
2.1.6 Digital encoding schemes. . . . . . . . . . . . . . . 13
2.1.7 Multiplexing. . . . . . . . . . . . . . . . . . . . . . 14
2.1.7.1 Time-division multiplexing . . . . . . . . . 14
2.1.7.2 Wavelength-division multiplexing . . . . . 15
2.1.8 Fiber optic sources . . . . . . . . . . . . . . . . . . 18
2.1.9 Direct vs external modulation . . . . . . . . . . . . 19
2.1.10 Fiber optic detectors . . . . . . . . . . . . . . . . . 21
3 Digital modulation 24
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2 Fundamental digital modulation . . . . . . . . . . . . . . 24
3.3 PSK modulations . . . . . . . . . . . . . . . . . . . . . . . . 25
3.3.1 Binary phase-shift keying (bpsk). . . . . . . . . . . . 27
3.3.2 Quadrature phase-shift keying (qpsk). . . . . . . . . 28
4 Coherent optical systems 31
4.1 Why coherent fiber communications. . . . . . . . . . . . . 31
4.2 Coherent vs direct detection. . . . . . . . . . . . . . . . . 32
4.3 Practical difficulties on coherent optical communication
systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.4 High-capacity optical transimissions: a shift to
coherent detection. . . . . . . . . . . . . . . . . . . . . . . 34
4.5 Digital signal processing based coherent optical
communication systems. . . . . . . . . . . . . . . . . . . . 36
4.6 Technologies associated at coherent optical systems:
wdm system evolution. . . . . . . . . . . . . . . . . . . . . 37
4.7 Coherent detection for 40g/100g network deployment. . . . 38
5 Polmux-qpsk modulation and coherent detection 41
5.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . 41
5.2 Evolution on optical modulation. . . . . . . . . . . . . . 41
5.3 Pm-qpsk transmitter and receiver architecture. . . . . . . 42
5.4 100 Gb/s PM-QPSK . . . . . . . . . . . . . . . . . . . . 44
5.4.1 Performance characteristics of 100 Gb/s
pm-qpsk. . . . . . . . . . . . . . . . . . . . . . . . 45
5.4.1.1 OSNR sensitivity. . . . . . . . . . . . . . . 45
5.4.1.2 Optical filtering tolerance. . . . . . . . . . 46
5.4.1.3 Chromatic dispersion tolerance . . . . . . 46
5.4.1.4 Polarization mode dispersion (pmd)
tolerance . . . . . . . . . . . . . . . . . . 47
5.4.2 Upgrading existing transmission links
to 100g. . . . . . . . . . . . . . . . . . . . . . . . . 48
5.4.3 Receiver complexity of digital coherent
detection. . . . . . . . . . . . . . . . . . . . . . . 50
6 Modeling coherent pm-qpsk system 52
6.1 Project layout coherent pm-qpsk system . . . . . . . . . . . 53
6.1.1 Principal parameters and modules description . . . 54
6.2 Inside coherent pm-qpsk system. . . . . . . . . . . . . . . . 57
6.2.1 Project layout polmux qpsk realistic. . . . . . . . . 57
6.2.2 Principal parameters and modules description . . . . 58
6.2.3 Project layout polmux qpsk. . . . . . . . . . . . . . 59
6.2.4 Principal parameters and module description . . . . 59
7 Simulations and results 61
7.1 Initial asumptions. . . . . . . . . . . . . . . . . . . . . . . . . 61
7.2 Final system architecture and reference parameters. . . . . . . 64
7.3 Simulations and analysis. . . . . . . . . . . . . . . . . . . . . . 67
7.3.1 Laser linewidth (phase noise) from 0 hz to 800 khz. . 68
7.3.2 Laser linewidth from 1 mhz to 10 mhz. . . . . . . . 78
7.3.3 Graphic comparison between laser linewidth
from 1 mhz to 10 mhz. . . . . . . . . . . . . . . . . . . . 84
7.4 Changing overhead on = . . . . . . . . . . . . . . . . 86
7.4.1 OSNR sensitivity comparison @ ber=1e-3
overhead [ 2% -1% - 0.5%]. . . . . . . . . . . . . . . . . . 86
7.4.2 = overhead 1%
number symbols training = 277. . . . . . . . . . . . . . . 88
7.4.3 = overhead 0.5%
number symbols training = 138. . . . . . . . . . . . . . . 89
7.5 Changing overhead on = . . . . . . . . . . . . . . . 90
7.5.1 OSNR Sensitivity comparison @ ber=1e-3
overhead [ 2% - 0.5%]. . . . . . . . . . . . . . . . . . 90
7.5.2 = overhead 2%
number symbols training = 555. . . . . . . . . . . . . . . . 92
7.5.3 = overhead 0.5%
number symbols training = 138. . . . . . . . . . . . . . . . 92
7.6 Changing overhead on = . . . . . . . . . . . . . . . . 93
7.6.1 OSNR sensitivity comparison @ ber=1e-3,
overhead [ 2%- 0.5%]. . . . . . . . . . . . . . . . . . . . . 93
7.6.2 = overhead 2%
number symbols training = 555. . . . . . . . . . . . . . . . 94
7.6.3 = overhead 0.5%
number symbols training = 138. . . . . . . . . . . . . . . . 94
8 Conclusion 96
8.1 Main findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
8.2 Near future . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
Bibliography 98
Bibliography 99
Bibliography 100
1
1. FIBER OPTICS COMMUNICATION HISTORY
Information technology has had an exponential growth through the modern
telecommunication systems. Particularly, optical fiber communication plays a vital role
in the development of high quality and high-speed telecommunication systems.
With the explosion of information traffic due to the Internet, electronic commerce,
computer networks, multimedia, voice, data, and video, the need for a transmission
medium with the bandwidth capabilities for handling such vast amounts of
information is paramount. Fiber optics, with its large carrying capacity bandwidth, has
proven to be the solution.
The growth of the fiber optics industry over the past decade has been huge.
Analysts expect that this industry will continue to grow at a tremendous rate well into
the next decade and beyond. Anyone with a vested interest in telecommunication
would be all the wiser to learn more about the tremendous advantages of fiber optic
communication [1].
Figure 1-1. Trends in optical transmission system capacity [1]
2
1.1. THE BIRTH OF FIBER OPTIC SYSTEMS
Communication engineers were interested in optical communication using lasers
in an effective manner from 1960 onwards. A new era in optical communication
started after the invention of laser by Dr. Maiman.
Theodore Maiman is famous for inventing the first functioning laser in the world in
1960. He has been called “the father of the electro-optics industry,” but Maiman
considers himself a scientist and an engineer, with research interests in electro-optics,
lasers, displays, and aerodynamics. In addition to his patent on the first working laser,
the ruby laser, Maiman also holds patents on masers, laser displays, optical scanning,
and laser modulation.
The laser was introduced in 1958 as an efficient source of light. The concept was
introduced by Charles Townes and Arthur Schawlow to show that masers could be
made to operate in optical and infrared regions. Basically, light is reflected back and
forth in an energized medium to generate amplified light as opposed to excited
molecules of gas amplified to generate radio waves, as is the case with the maser.
Laser stands for "light amplification by stimulated emission of radiation."
The light waves from the laser, a coherent source of light having high intensity, high
monochromaticity and high directionality with less divergence, are used as carrier
waves capable of transport large amount of information compared with radio waves
and microwaves [2].
Figure 1-2. Bit rate distance product [2].
Figure 1-2, shows the different communication systems and their bit rate distance
product.
3
In 1960, the first continuously operating helium-neon gas laser is invented and tested.
That same year an operable laser was invented which used a synthetic pink ruby
crystal as the medium and produced a pulse of light.
In 1961, Elias Snitzer of American Optical published a theoretical description of single
mode fibers whose core would be so small it could carry light with only one wave-
guide mode. Snitzer was able to demonstrate a laser directed through a thin glass fiber
which was sufficient for medical applications, but for communication applications the
light loss became too great.
In 1970, the goal of making single mode fibers with attenuation less then 20dB/km was
reached by scientists at Corning Glass Works. This was achieved through doping silica
glass with titanium.
Also in 1970, Morton Panish and Izuo Hayashi of Bell Laboratories, along with a group
from the Ioffe Physical Institute in Leningrad, demonstrated a semiconductor diode
laser capable of emitting continuous waves at room temperature.
In 1973, Bell Laboratories developed a modified chemical vapor deposition process
that heats chemical vapors and oxygen to form ultra-transparent glass that can be
mass-produced into low-loss optical fiber. This process still remains the standard for
fiber-optic cable manufacturing.
In the late 1970s and early 1980s, telephone companies began to use fibers extensively
to rebuild their communications infrastructure [4].
1.2. ADVANTAGES OF OPTICAL FIBER COMMUNICATION
Optical fiber systems have many benefits over metallic-based communication systems.
These advantages include [2]:
· Wider bandwidth:
The information carrying capacity of a transmission system is directly proportional to
the carrier frequency of the transmitted signals.
The optical carrier frequency is in the range 10^13 to 10^15 Hz while the radio wave
frequency is about 10^9 Hz. Thus the optical fiber yields greater transmission
bandwidth than the older communication systems and the data rate or number of bits
per second is increased to a greater extent in the optical fiber communication system.
4
Furthermore data rate or information carrying capacity of optical fibers is enhanced to
many orders of magnitude with the division multiplexing technique.
· optical amplification:
Doped fiber amplifiers (DFAs) are optical amplifiers that use a doped optical fiber as a
gain medium to amplify an optical signal.
The signal to be amplified and a pump laser are multiplexed into the doped fiber, and
the signal is amplified through interaction with the doping ions. The most common
example is the Erbium Doped Fiber Amplifier (EDFA), where the core of a silica fiber is
doped with trivalent Erbium ions and can be efficiently pumped with a laser at a
wavelength of 980 nm or 1,480 nm, and exhibits gain in the 1,550 nm region.
· Low transmission loss:
Due to the usage of the ultra low loss fibers and the erbium doped fibers as optical
amplifiers, one can achieve almost lossless transmission.
In the modern optical fiber telecommunication systems, the fibers having a
transmission loss of 0.2 dB/km are used. Further, using erbium doped fibers amplifiers
over a short length in the transmission path at selective points, appropriate optical
amplification can be achieved. Thus the repeater spacing is more than 100 km. Since
the amplification is done in the optical domain itself, there is no distortion of the signal
only white noise is added.
· Nonconductivity
An optical fiber is a flexible, transparent fiber made of a pure glass (silica) not much
wider than a human hair. It functions as a waveguide, or "light pipe", to transmit light
between the two ends of the fiber.
Another advantage of optical fibers is their dielectric nature. Since optical fiber has no
metallic components, it can be installed in areas with electromagnetic interference
(EMI), including radio frequency interference (RFI). Areas with high EMI include utility
lines, power-carrying lines, and railroad tracks. All-dielectric cables are also ideal for
areas of high lightning-strike incidence.
· Signal security:
The transmitted signal through the fibers does not radiate. Further the
signal cannot be tapped from a fiber in an easy manner.
5
Unlike metallic-based systems, the dielectric nature of optical fiber makes it impossible
to remotely detect the signal being transmitted within the cable. The only way to do so
is by accessing the optical fiber.
Accessing the fiber requires intervention that is easily detectable by security
surveillance. These circumstances make fiber extremely attractive to
governmental bodies, banks, and others with major security concerns.
· Small size and weight:
Fiber optic cables are developed with small radii, and they are flexible, compact and
lightweight. The fiber cables can be bent or twisted without damage.
Further, the optical fiber cables are superior to the copper cables in terms of
storage, handling, installation and transportation, maintaining comparable strength
and durability.
6
2. PRINCIPLES OF FIBER OPTIC COMMUNICATION
2.1. BASIC FIBER OPTIC COMMUNICATION SYSTEM
Fiber optics is a medium for carrying information from one point to another in the form of light. Unlike the copper form of transmission, fiber optics is not electrical in nature. A basic fiber optic system consists of a transmitting device that converts an electrical signal into a light signal, an optical fiber cable that carries the light, and a receiver that accepts the light signal andconverts it back into an electrical signal.
Figure 2-1. Typical fiber optic communication system [3] The complexity of a fiber optic system can range from very simple: single span, like a local area network, to complex: long distance periodically amplified link. For example, the system shown in the figure above, could be built very inexpensively using a visible LED, plastic fiber, a silicon photodetector, and some simple electronic circuitry. On the other hand, a typical system used for long-distance, high-bandwidth telecommunication that employs wavelength-division multiplexing, erbium-doped fiber amplifiers, external modulation using DFB lasers with temperature compensation, fiber Bragg gratings, and high-speed infrared photodetectors could increase the cost. The basic question is “how much information is to be sent and how far does it have to go?” With this in mind we will examine the various components that make up a fiber optic communication system and the considerations that must be taken into account in the design of such systems [2] [3].
7
2.1.1. TRANSMISSION WINDOWS Optical fiber transmission uses wavelengths that are in the near-infrared portion of the electromagnetic spectrum, just above the visible, and thus undetectable to the unaided eye. Typical optical transmission wavelengths are 850 nm, 1310 nm, and 1550 nm. Both lasers and LEDs are used to transmit light through optical fiber. Lasers are usually used for 1310- or 1550-nm single-mode applications. LEDs are used for 850- or 1300-nm multimode applications. There are ranges of wavelengths at which the fiber operates best. Each range is known as an operating window. Each window is centered on the typical operational wavelength [3].
2.1.2. FIBER OPTIC LOSS Loss in a system can be expressed as the following:
where Pin is the input power to the fiber and Pout is the power available at the output of the fiber. For convenience, fiber optic loss is typically expressed in terms of decibels (dB) and can be calculated using the next equation:
Oftentimes, loss in optical fiber is also expressed in terms of decibels per kilometer (dB/km). Optical power in fiber optic systems is typically expressed in terms of dBm, which is a decibel term referred to 1mW. With optical power expressed in dBm, output power anywhere in the system can be determined simply by expressing the power input in dBm and subtracting the individual component losses, also expressed in dB [2] [3].
8
2.1.3. TYPES OF FIBER Three basic types of fiber optic cable are used in communication systems [2] [3]: 1. Step-index multimode
Figure 2-2. Step-index multimode fiber profile [3]. Step-index multimode fiber has an index of refraction profile that “steps” from low to high to low as measured from cladding to core to cladding. Relatively large core diameter and numerical aperture characterize this fiber. The core/cladding diameter of a typical multimode fiber used for telecommunication is 62.5/125 µm. The term “multimode” refers to the fact that multiple modes through the fiber are possible. Stepindex multimode fiber is used in applications that require low bit rate and bandwidth (< 1 GHz) over relatively short distances (< 3 km) such as a local area network or a campus network backbone. The major benefits of multimode fiber are: * it is relatively easy to work with * because of its larger core size, light is easily coupled to and from it * it can be used with both lasers and LEDs as sources * coupling losses are less than those of the single-mode fiber The drawback is that because many modes are allowed to propagate (a function of core diameter, wavelength, and numerical aperture) it suffers from modal dispersion. The result of modal dispersion is bandwidth limitation, which translates into lower data rates.
9
2. Step-index single mode
Figure 2-3. Step-index single mode fiber profile [3]. Single-mode step-index fiber allows for only one path, or mode, for light to travel within the fiber. In a multimode step-index fiber, the number of modes Mn propagating can be approximated by:
Here V is known as the normalized frequency, or the V-number, which relates the fiber size, the refractive index, and the wavelength. The V-number is given by next two equations:
In the equation above, a is the fiber core radius, λ is the operating wavelength, N.A. is the numerical aperture, n1 is the core index, and Δ is the relative refractive index difference between core and cladding. The analysis of how the V-number is derived is beyond the scope of this module, but by reducing the diameter of the fiber to a point at which the V-number is less than 2.405, higher-order modes are effectively extinguished and single-mode operation is possible. The core diameter for a typical single-mode fiber is between 5 μm and 10 μm with a 125 μm cladding. Single-mode fibers are used in applications in which low signal loss and high data rates are required, such as in long spans where repeater/amplifier spacing must be maximized. Because single-mode fiber allows only one mode or ray to propagate (the
10
lowest-order mode), it does not suffer from modal dispersion like multimode fiber and therefore can be used for higher bandwidth applications. However, even though single-mode fiber is not affected by modal dispersion, at higher data rates chromatic dispersion can limit the performance. This problem can be overcome by several methods. One can transmit at a wavelength in which glass has a fairly constant index of refraction (~1300 nm), use an optical source such as a distributed-feedback laser (DFB laser) that has a very narrow output spectrum, use special dispersion-compensating fiber, or use a combination of all these methods. In a nutshell, single-mode fiber is used in high-bandwidth, long-distance applications such as long-distance telephone trunk lines, cable TV head-ends, and high-speed local and wide area network (LAN and WAN) backbones. The major drawback of single-mode fiber is that it is relatively difficult to work with because of its small core size. Also, single-mode fiber is typically used only with laser sources. 3.Graded-index
Figure 2-4. Graded-index fiber profile [3]. Graded-index fiber is a compromise between the large core diameter and N.A. of multimode fiber and the higher bandwidth of single-mode fiber. With creation of a core whose index of refraction decreases parabolically from the core center toward the cladding, light traveling through the center of the fiber experiences a higher index than light traveling in the higher modes. This means that the higher-order modes travel faster than the lower-order modes, which allows them to “catch up” to the lower-order modes, thus decreasing the amount of modal dispersion, which increases the bandwidth of the fiber. 2.1.4. DISPERSION In optics, dispersion is the phenomenon in which the phase velocity of a wave depends on its frequency, or alternatively when the group velocity depends on the frequency. Media having such a property are termed dispersive media. Dispersion is sometimes called chromatic dispersion to emphasize its wavelength-dependent nature, or group-
11
velocity dispersion (GVD) to emphasize the role of the group velocity. Dispersion is most often described for light waves, but it may occur for any kind of wave that interacts with a medium or passes through an inhomogeneous geometry, such as sound waves. A material's dispersion is measured by its Abbe number, V, with low Abbe numbers corresponding to strong dispersion. Dispersion, expressed in terms of the symbol Δt, is defined as pulse spreading in an optical fiber. As a pulse of light propagates through a fiber, elements such as numerical aperture, core diameter, refractive index profile, wavelength, and laser linewidth cause the pulse to broaden. This poses a limitation on the overall bandwidth of the fiber as demonstrated in next figure:
Figure 2-5. Pulse broadening caused by dispersion [3].
The overall effect of dispersion on the performance of a fiber optic system is known as intersymbol interference. Intersymbol interference occurs when the pulse spreading caused by dispersion causes the output pulses of a system to overlap, rendering them undetectable. If an input pulse is caused to spread such that the rate of change of the input exceeds the dispersion limit of the fiber, the output data will become indiscernible [2] [3].
FIGURE 2-6. Intersymbol interference [3].
12
2.1.5. TYPES OF DISPERSION Dispersion is generally divided into two categories: modal dispersion and chromatic dispersion. 2.1.5.1. MODAL DISPERION Modal dispersion is defined as pulse spreading caused by the time delay between lower-order modes (modes or rays propagating straight through the fiber close to the optical axis) and higher-order modes (modes propagating at steeper angles). Modal dispersion is problematic in multimode fiber, causing bandwidth limitation, but it is not a problem in single-mode fiber where only one mode is allowed to propagate.
Figure 2-7. Mode propagation in an optical fiber [3]. 2.1.5.2 Chromatic dispersion Chromatic dispersion is pulse spreading due to the fact that different wavelengths of light propagate at slightly different velocities through the fiber. All light sources, whether laser, have finite linewidths. Because the index of refraction of glass fiber is a wavelength-dependent quantity, different wavelengths propagate at different velocities. Chromatic dispersion consists of two parts: material dispersion and waveguide dispersion.
Material dispersion is due to the wavelength dependency on the index of refraction of glass.
13
Material dispersion and waveguide dispersion can have opposite signs depending on the transmission wavelength. In the case of a step-index single-mode fiber, these two effectively cancel each other at 1310 nm, yielding zerodispersion. This makes very high-bandwidth communication possible at this wavelength. However, the drawback is that, even though dispersion is minimized at 1310 nm, attenuation is not. Glass fiber exhibits minimum attenuation at 1550 nm. Coupling that with the fact that erbium-doped fiber amplifiers (EDFA) operate in the 1550-nm range makes it obvious that, if the zero-dispersion property of 1310 nm could be shifted to coincide with the 1550-nm transmission window, high-bandwidth long-distance communication would be possible. With this in mind, zero-dispersion-shifted fiber was developed [3]. When considering the total dispersion from different causes, we can approximate the total dispersion by Δt tot.
where Δtn represents the dispersion due to the various components that make up the system. The transmission capacity of fiber is typically expressed in terms of bandwidth × distance. The approximate bandwidth of a fiber can be related to the total dispersion by the following relationship:
2.1.6. DIGITAL ENCODING SCHEMES Signal format is an important consideration in evaluating the performance of a fiber optic system. The signal format directly affects the detection of the transmitted signals. The accuracy of the reproduced signal depends on the intensity of the received signal, the speed and linearity of the receiver, and the noise levels of the transmitted and received signal. Many coding schemes are used in digital communication systems, each with its own benefits and drawbacks. The most common encoding schemes are the return-to-zero (RZ) and non-return-to-zero (NRZ). The NRZ encoding scheme, for example, requires only one transition per symbol, whereas RZ format requires two transitions for each data bit. This implies that
14
the required bandwidth for RZ must be twice that of NRZ. This is not to say that one is better than the other. Depending on the application, any of the code formats may be more appropriate than the others. For example, in synchronous transmission systems in which large amounts of data are to be sent, clock synchronization between the transmitter and receiver must be ensured. In this case Manchester encoding is used. The transmitter clock is embedded in the data. The receiver clock is derived from the guaranteed transition in the middle of each bit [3].
Figure 2-8. Different encoding schemes [3]. 2.1.7. MULTIPLEXING The purpose of multiplexing is to share the bandwidth of a single transmission channel among several users. Two multiplexing methods are commonly used in fiber optics [3]: 1. Time-division multiplexing (TDM) 2. Wavelength-division multiplexing (WDM) 2.1.7.1. TIME-DIVISION MULTIPLEXING In electronics, a multiplexer (or MUX) is a device that selects one of several analog or digital input signals and forwards the selected input into a single line. A multiplexer of 2n inputs has n select lines, which are used to select which input line to send to the output. Multiplexers are mainly used to increase the amount of data that can be sent over the network within a certain amount of time and bandwidth. A multiplexer is also called a data selector.
15
In time-division multiplexing, time on the information channel, or fiber, is shared among the many data sources. The multiplexer MUX can be described as a type of “rotary switch,” which rotates at a very high speed, individually connecting each input to the communication channel for a fixed period of time. The process is reversed on the output with a device known as a demultiplexer, or DEMUX. After each channel has been sequentially connected, the process repeats itself. One complete cycle is known as a frame. To ensure that each channel on the input is connected to its corresponding channel on the output, start and stop frames are added to synchronize the input with the output. TDM systems may send information using any of the digital modulation schemes described (analog multiplexing systems also exist).
Figure 2-9. Time-division multiplexing system [3]. 2.1.7.2. WAVELENGTH-DIVISION MULTIPLEXING In wavelength-division multiplexing, each data channel is transmitted using a slightly different wavelength (different color). With use of a different wavelength for each channel, many channels can be transmitted through the same fiber without interference. This method is used to increase the capacity of existing fiber optic systems many times. Each WDM data channel may consist of a single data source or may be a combination of a single data source and a TDM (time-division multiplexing).
16
Figure 2-10. Wavelength-division multiplexing [3] Erbium-doped fiber amplifiers (EDFA)—The EDFA is an optical amplifier used to boost the signal level in the 1530-nm to 1570-nm region of the spectrum. When it is pumped by an external laser source of either 980 nm or 1480 nm, signal gain can be as high as 30 dB (1000 times). Because EDFAs allow signals to be regenerated without having to be converted back to electrical signals, systems are faster and more reliable. When used in conjunction with wavelength-division multiplexing, fiber optic systems can transmit enormous amounts of information over long distances with very high reliability. Dense wavelength-division multiplexing (DWDM) refers to the transmission of multiple closely spaced wavelengths through the same fiber. For any given wavelength λ and corresponding frequency f, the International Telecommunications Union (ITU) defines standard frequency spacing Δf as 100 GHz, which translates into a Δλ of 0.8 nm wavelength spacing. This follows from the relationship :
DWDM systems operate in the 1550-nm window because of the low attenuation characteristics of glass at 1550 nm and the fact that erbium-doped fiber amplifiers (EDFA) operate in the 1530-nm–1570-nm range.
17
Figure 2-11. 8-channel WDM [3]. The ITU grid specifies that each transmitted wavelength in a DWDM system is separated by 100 GHz, systems currently under development have been demonstrated that reduce the channel spacing to 50 GHz and below (< 0.4 nm). As the channel spacing decreases, the number of channels that can be transmitted increases, thus further increasing the transmission capacity of the system.
Figure 2-12. ITU grid [3]
18
2.1.8. FIBER OPTIC SOURCES Two basic light sources are used for fiber optics: laser diodes (LD) and light-emitting diodes (LED). Each device has its own advantages and disadvantages as listed in next table [3].
Figure 2-13. LED VS laser [3]. Fiber optic sources must operate in the low-loss transmission windows of glass fiber. LEDs are typically used at the 850-nm and 1310-nm transmission wavelengths, whereas lasers are primarily used at 1310 nm and 1550 nm. LEDs are typically used in lower-data-rate, shorter-distance multimode systems because of their inherent bandwidth limitations and lower output power. They are used in applications in which data rates are in the hundreds of megahertz as opposed to GHz data rates associated with lasers. Laser diodes (LD) are used in applications in which longer distances and higher data rates are required. Because an LD has a much higher output power than an LED, it is capable of transmitting information over longer distances. Consequently, and given the fact that the LD has a much narrower spectral width, it can provide high-bandwidth communication over long distances. In applications such as wavelength-division multiplexing in which several wavelengths are being transmitted down the same fiber, the stability of the source becomes critical. This usually requires complex circuitry and feedback mechanisms to detect and correct for drifts in wavelength. The benefits, however, of high-speed transmission using LDs typically outweigh the drawbacks and added expense.
19
Figure 2-14. Drive current vs output power for LED and laser [3]. 2.1.9. DIRECT VS EXTERNAL MODULATION Lasers and LEDs used in telecommunication applications are modulated using one of two methods: direct modulation or external modulation. - In direct modulation the output power of the device varies directly with the input drive current. Both LEDs and lasers can be directly modulated using analog and digital signals. The benefit of direct modulation is that it is simple and cheap. The disadvantage is that it is slower than indirect modulation. Direct modulation is used up to 10Gb/s.
Figure 2-15. Direct modulation [3].
20
- In external modulation an external device is used to modulate the intensity or phase of the light source. The light source remains on while the external modulator acts like a “shutter” controlled by the information being transmitted. External modulation is typically used in high-speed applications such as long-haul telecommunication or cable TV head ends. The benefits of external modulation are that it is much faster and can be used with higher-power laser sources. The disadvantage is that it is more expensive and requires complex circuitry to handle the high frequency RF modulation signal.
Figure 2-16. External modulation [3]. External modulation is typically accomplished using an integrated optical modulator that incorporates a waveguide Mach-Zehnder interferometer fabricated on a slab of lithium niobate (LiNbO3). The waveguide is created using a lithographic process similar to that used in the manufacturing of semiconductors. The waveguide region is slightly doped with impurities to increase the index of refraction so that the light is guided through the device.
Figure 2-17. External modulation using Mach-Zehnder waveguide interferometer [3]. Light entering the modulator is split into two paths. One path is unchanged or unmodulated. The other path has electrodes placed across it. Because LiNbO3 is an
21
electro-optic material, when a voltage is placed across the waveguide its index of refraction is changed, causing a phase rotation proportional to the amplitude of the applied voltage. When the light is then recombined, the two waves interfere with one another. If the two waves are in phase, the interference is constructive and the output is on. If the two waves are out of phase, the interference is destructive and the waves cancel each other. The input voltage associated with a 180° phase shift is known as Vπ . The induced phase shift can be calculated using:
where Vin is the voltage applied to the modulator. Lithium niobate modulators are well developed and used extensively in telecommunication applications. Devices are available at both the 1310-nm and 1550-nm wavelengths [3]. 2.1.10. FIBER OPTIC DETECTORS The purpose of a fiber optic detector is to convert light emanating from the optical fiber back into an electrical signal. The choice of a fiber optic detector depends on several factors including wavelength, responsivity, and speed or rise time. Figure 8-30 depicts the various types of detectors and their spectral responses.
Figure 2-18. Detector spectral response [3].
22
The process by which light is converted into an electrical signal is the opposite of the process that produces the light. Light striking the detector generates a small electrical current that is amplified by an external circuit. Absorbed photons excite electrons from the valence band to the conduction band, resulting in the creation of an electron-hole pair. Under the influence of a bias voltage these carriers move through the material and induce a current in the external circuit. For each electron-hole pair created, the result is an electron flowing in the circuit. Typical current levels are small and require some amplification as shown in next figure.
Figure 2-19. Typical detector amplifier circuit [3]. The most commonly used photodetectors are the PIN and avalanche photodiodes (APD). The material composition of the device determines the wavelength sensitivity. In general, silicon devices are used for detection in the visible portion of the spectrum; InGaAs crystal are used in the near-infrared portion of the spectrum between 1000 nm and 1700 nm, and germanium PIN and APDs are used between 800 nm and 1500 nm.
Figure 2-20. Typical photodetector characteristics [3].
- Responsivity : the ratio of the electrical current to the detector’s output optical power
- Quantum efficiency : the ratio of the number of electrons generated by the
detector to the number of photons incident on the detector
23
Quantum efficiency = (Number of electrons)/Photon
- Dark current : the amount of current generated by the detector with no light applied. Dark current increases about 10% for each temperature increase of
1°C and is much more prominent in Ge and InGaAs at longer wavelengths than in silicon at shorter wavelengths.
- Response time : the time required for the detector to respond to an optical input. The response time is related to the bandwidth of the detector by BW = 0.35/tr Parameter: bandwidth of the photodiode. where tr is the rise time of the device. The rise time is the time required for the detector to rise to a value equal to 63.2% of its final steady-state reading [3].
24
3. DIGITAL MODULATION 3.1. INTRODUCTION In electronics, modulation is the process of varying one or more properties of a high-frequency periodic waveform, called the carrier signal, with a modulating signal which typically contains information to be transmitted. The purpose of modulation is usually to enable the carrier signal to transport the information in the modulation signal to some destination. At the destination, a process of demodulation extracts the modulation signal from the modulated carrier. The three key parameters of a periodic waveform are its amplitude ("volume"), its phase ("timing") and its polarization. Any of these properties can be modified in accordance with a low frequency signal to obtain the modulated signal. Typically a high-frequency sinusoid waveform is used as carrier signal, but a square wave pulse train may also be used. In telecommunications, modulation is the process of conveying a message signal, for example a digital bit stream or an analog audio signal, inside another signal that can be physically transmitted. Modulation of a sine waveform is used to transform a baseband message signal into a passband signal, for example low-frequency audio signal into a radio-frequency signal (RF signal). In radio communications, cable TV systems or the public switched telephone network for instance, electrical signals can only be transferred over a limited passband frequency spectrum, with specific (non-zero) lower and upper cutoff frequencies. Modulating a sine-wave carrier makes it possible to keep the frequency content of the transferred signal as close as possible to the centre frequency (typically the carrier frequency) of the passband. The aim of digital modulation is to transfer a digital bit stream over an bandpass channel, or over a limited radio frequency band. 3.2. FUNDAMENTAL DIGITAL MODULATION The most fundamental digital modulation techniques are based on:
Phase-shift keying (PSK), a finite number of phases are used.
Frequency-shift keying (FSK), a finite number of frequencies are used.
Amplitude-shift keying (ASK), a finite number of amplitudes are used.
In the case of Quadrature Amplitude Modulation (QAM), a finite number of at least two phases, joint modulation of phase and amplitude.
25
In QAM, an inphase signal (the I signal, for example a cosine waveform) and a quadrature phase signal (the Q signal, for example a sine wave) are amplitude modulated with a finite number of amplitudes, and summed. It can be seen as a two-channel system, each channel using ASK. The resulting signal is equivalent to a combination of PSK and ASK. In all of the above methods, each of these phases, frequencies or amplitudes are assigned a unique pattern of binary bits. Usually, each phase, frequency or amplitude encodes an equal number of bits. This number of bits comprises the symbol that is represented by the particular phase, frequency or amplitude. If the alphabet consists of M = 2N alternative symbols, each symbol represents a message consisting of N bits. If the symbol rate (also known as the baud rate) is fS symbols/second (or baud), the data rate is NfS bit/second. For example, with an alphabet consisting of 16 alternative symbols, each symbol represents 4 bits. Thus, the data rate is four times the symbol rate. In the case of PSK, ASK or QAM, where the carrier frequency of the modulated signal is constant, the modulation alphabet is often conveniently represented on a constellation diagram, showing the amplitude of the I signal at the x-axis, and the amplitude of the Q signal at the y-axis, for each symbol [15]. 3.3. PSK MODULATIONS All convey data by changing some aspect of a base signal, the carrier wave (usually a sinusoid), in response to a data signal. In the case of PSK, the phase is changed to represent the data signal. There are two fundamental ways of utilizing the phase of a signal in this way:
By viewing the phase itself as conveying the information, in which case the demodulator must have a reference signal to compare the received signal's phase against; or
By viewing the change in the phase as conveying information — differential schemes, some of which do not need a reference carrier (to a certain extent).
A convenient way to represent PSK schemes is on a constellation diagram. This shows the points in the complex plane where, in this context, the real and imaginary axes are termed the in-phase and quadrature axes respectively due to their 90° separation. Such a representation on perpendicular axes lends itself to straightforward implementation. The amplitude of each point along the in-phase axis is used to modulate a cosine (or sine) wave and the amplitude along the quadrature axis to modulate a sine (or cosine) wave.
26
Figure 3-1. Constellation diagram for 8-PSK [15]. In PSK, the constellation points chosen are usually positioned with uniform angular spacing around a circle. This gives maximum phase-separation between adjacent points and thus the best immunity to corruption. They are positioned on a circle so that they can all be transmitted with the same energy. In this way, the module of the complex numbers they represent will be the same and thus so will the amplitudes needed for the cosine and sine waves. Two common examples are "binary phase-shift keying" (BPSK) which uses two phases, and "quadrature phase-shift keying" (QPSK) which uses four phases, although any number of phases may be used. Since the data to be conveyed are usually binary, the PSK scheme is usually designed with the number of constellation points being a power of 2 [15].
Definitions
For determining error-rates mathematically, some definitions will be needed:
Eb = Energy-per-bit Es = Energy-per-symbol = Eb/n with n bits per symbol Tb = Bit duration Ts = Symbol duration N0 / 2 = Noise power spectral density (W/Hz), AWGN Noise. Pb = Probability of bit-error Ps = Probability of symbol-error
Q(x) will give the probability that a single sample taken from a random process with
zero-mean and unit-variance Gaussian probability density function will be greater or
equal to x. It is a scaled form of the complementary Gaussian error function:
.
27
The error-rates quoted here are those in additive white Gaussian noise (AWGN) [15].
3.3.1. BINARY PHASE-SHIFT KEYING (BPSK)
BPSK is the simplest form of phase shift keying (PSK). It uses two phases which are
separated by 180° and so can also be termed 2-PSK. It does not particularly matter
exactly where the constellation points are positioned, and in this figure they are shown
on the real axis, at 0° and 180°. This modulation is the most robust of all the PSKs since
it takes the highest level of noise or distortion to make the demodulator reach an
incorrect decision. It is, however, only able to modulate at 1 bit/symbol (as seen in the
figure) and so is unsuitable for high data-rate applications when bandwidth is limited.
Figure 3-2. Constellation diagram for a BPSK [15].
This kind of modulations need the phase recovery, it is possible. Differential encoding
ease the use but loses some performance.
Implementation
The general form for BPSK follows the equation:
This yields two phases, 0 and π. In the specific form, where fc is the frequency of the
carrier-wave binary data, is conveyed with the following signals:
for binary "0"
28
for binary "1"
Hence, the signal-space can be represented by the single basis function
where 1 is represented by and 0 is represented by . This
assignment is, of course, arbitrary.
This use of this basis function is shown at the end of the next section in a signal timing
diagram. The topmost signal is a BPSK-modulated cosine wave that the BPSK
modulator would produce. The bit-stream that causes this output is shown above the
signal (the other parts of this figure are relevant only to QPSK) [15].
Bit error rate
The bit error rate (BER) of BPSK in AWGN can be calculated as:
or
Since there is only one bit per symbol, this is also the symbol error rate.
3.3.2. QUADRATURE PHASE-SHIFT KEYING (QPSK)
Sometimes this is known as quaternary PSK, quadriphase PSK, 4-PSK, or 4-QAM.
(Although the root concepts of QPSK and 4-QAM are different, the resulting modulated
radio waves are exactly the same.) QPSK uses four points on the constellation diagram,
equispaced around a circle. With four phases, QPSK can encode two bits per symbol,
shown in the diagram with gray coding to minimize the bit error rate (BER).
The mathematical analysis shows that QPSK can be used either to double the data rate
compared with a BPSK system while maintaining the same bandwidth of the signal, or
to maintain the data-rate of BPSK but halving the bandwidth needed. In this latter
case, the BER of QPSK is exactly the same as the BER of BPSK.
As with BPSK, there are phase ambiguity problems at the receiving end, and
differentially encoded QPSK is often used in practice [15].
29
Figure 3-3. Constellation diagram for QPSK with Gray coding. Each adjacent symbol
only differs by one bit [15].
Implementation
The implementation of QPSK is more general than that of BPSK and also indicates the
implementation of higher-order PSK. Writing the symbols in the constellation diagram
in terms of the sine and cosine waves used to transmit them:
This yields the four phases π/4, 3π/4, 5π/4 and 7π/4 as needed.
This results in a two-dimensional signal space with unit basis functions
The first basis function is used as the in-phase component of the signal and the second
as the quadrature component of the signal.
Hence, the signal constellation consists of the signal-space 4 points
The factors of 1/2 indicate that the total power is split equally between the two
carriers.
30
Bit error rate
Although QPSK can be viewed as a quaternary modulation, it is easier to see it as two
independently modulated quadrature carriers. With this interpretation, the even (or
odd) bits are used to modulate the in-phase component of the carrier, while the odd
(or even) bits are used to modulate the quadrature-phase component of the carrier.
BPSK is used on both carriers and they can be independently demodulated [15].
As a result, the probability of bit-error for QPSK is the same as for BPSK:
However, in order to achieve the same bit-error probability as BPSK, QPSK uses twice
the power (since two bits are transmitted simultaneously).
The symbol error rate is given by:
.
If the signal-to-noise ratio is high (as is necessary for practical QPSK systems) the
probability of symbol error may be approximated:
31
4. COHERENT OPTICAL SYSTEMS
4.1. WHY COHERENT FIBER COMMUNICATIONS
One of the standard modulation/demodulation schemes being employed in the
present optical fiber communications is called the intensity-modulation/direct-
detection (IM/DD) scheme. The term IM stems from two facts:
· The light intensity (not the amplitude) is modulated linearly by a signal carrying the
information, usually binary digital signal, to the input signal voltage.
· That basically no attention is paid to the phase of the carrier. The spectral broadening
of the optical carrier is usually much wider than the spread due to modulation.
Figure 4-1. Simplified block diagram of an optical intensity direct detection
modulation [13].
The term DD stems from that the signal is detected directly at the optical stage of the
receiver, neither the frequency conversion (heterodyne or homodyne scheme) nor
sophisticated signal processing at lower frequencies is performed.
The IM/DD system has a great advantage in system simplicity and low cost, but in
some applications of the optical fiber communications in wich a long repeater
separation is our primary concern exist, like optical fiber communications between
continents as example, the improvement of the equivalent receiver sensitivity, the
minimum signal power we can detect, by a coherent modulation/demodulation
scheme may become advantageous, even at the sacrifice of simplicity and low cost.
32
The sensitivity improvement on these coherent schemes is particularly dramatic,
where the silica-fiber loss becomes minimum (around 0.2 dB/km) whereas good
photodetectors for the DD scheme are not available.
The above improvement on the receiver sensitivity is the principal motivation
underlying the effort toward the heterodyne coherent optical fiber communications.
However, it is accepted that the IM/DD systems will never be entirely retired, because
coherent systems are, and will continue to be, relatively expensive.
Figure 4-2. Block diagram of coherent receiver [14]
4.2. COHERENT VS DIRECT DETECTION
Two classes of receiver structures are used to detect optical signals: direct detection
and coherent detection (heterodyne or homodyne).
In coherent detection, an optical local oscillator field is added to the received optical
field, and the sum is detected by a photodetector. The resulting signal is further
processed at base band (homodyne detection) or at an intermediate frequency
(heterodyne detection). Phase and frequency tracking of the signal field by the local
oscillator laser is required. The mixing of the weak signal field and the strong local
oscillator field at the front-end of a coherent receiver provides linear amplification and
converts the optical signal into an electrical ouput with gain by using avalanche
photodiodes (APD), a highly sensitive semiconductor electronic device that exploits
the photoelectric effect to convert light to electricity, raising the signal level well above
the noise level of subsequent electronics. This is why a detector with gain is not
required. Coherent detection can be used on any of the modulations.
33
Figure 4-3. Basic concept of coherent detection [9]
Coherent detection has many advantages over direct detection, it is sensitive to the
phase as well as the amplitude of the optical wave, and offers an inherent ultra-narrow
optical filtering capability useful for dense wavelength-division multiplexing.
The best scheme for coherent detection that gives the best sensitivity [lowest bit-error
rate at a given optical signal-to-noise ratio (OSNR)] is homodyne detection, but this
mode requires the use of special narrow linewidth lasers that are phase locked, which
makes it expensive to implement.
Coherent receivers can linearly down-convert the whole optical signal to a baseband
electrical signal by using heterodyne or homodyne detection, and have the following
advantages against direct detection:
· The shot-noise limited receiver sensitivity can be achieved with a sufficient local
oscillator (LO) power. The LO gives us a signal gain, whereas the LO shot noise
overwhelms the thermal noise of the receiver, thus we can achieve the shot-noise
limited receiver sensitivity.
· Frequency resolution at the intermediate frequency (IF) or baseband stage is high
that we can separate closely spaced wavelength-division multiplexed (WDM) channels
at the electrical stage [14].
4.3. PRACTICAL DIFFICULTIES ON COHERENT OPTICAL COMMUNICATION SYSTEMS
Coherent optical communication systems present much more advantages over those
based on conventional intensity modulation direct-detection (IM/DD) schemes. Among
those advantages, it can be mentioned a larger density of transmission channels.
However, the implementation of these systems requires solving various practical
difficulties. One of them is that the signal and the local oscilator beams must be
34
coherent in order to recover the information contained in the phase of the signal
beam.
On the other hand, it becomes necessary to compensate the slow phase shift, wich is
quite common in these kinds of systems. Finally another important factor to be taken
into account is the signal degradation at the receiver owing to fluctuations of light
state of polarization (SOP) provoked by the fiber.
In particular, in order to solve the latter aspect, different alternative schemes have
been proposed up to date: one of them is the use of active polarization-control
devices, which consist basically of two elements that control two polarization freedom
degrees, ellipticity and rotation. Furthermore it becomes necessary an algorithm to
control at any time the polarization of one of the interferometer arms [14].
Figure 4-4. Signal degradation in optical systems [7].
4.4. HIGH-CAPACITY OPTICAL TRANSIMISSIONS: A SHIFT TO COHERENT DETECTION
High-capacity optical transmission has experienced orders of magnitude growth in
capacity in the past two decades. The capacity growth has been enabled by key
technology breakthroughs such as the erbium-doped fiber amplifier (EDFA),
wavelength-division multiplexing (WDM), dispersion compensation and management.
In addition, advances in modulation formats led to corresponding increases in fiber
transmission capacity in recent years.
In the meantime, spectral efficiency (SE) for fiber-optic transmission has been
increasing steadily as well. High SE can be achieved by using modulation formats in
which more than 1 bit of information is transmitted per symbol.
There are many benefits to employing high-SE modulation formats. The main point is
not in optical amplifier but in electronics using high order modulation formats you
keep same electric band while transmitting more bits. With high-SE formats, the speed
of transceiver electronics can be relaxed. High-SE systems are generally more tolerant
35
to chromatic dispersion and polarization-mode dispersion (PMD), since they increase
the bit rate using the same bandwidth.
Dispersion and PMD tolerance are particularly attractive for high-bit-rate transmission,
as dispersion tolerance is reduced by a factor of 4 for an increase in bit per symbol rate
by a factor of 2.
Early efforts in achieving high SE used direct detection but using a phase-to-amplitude
converter. A constant-intensity (nonlinearity-tolerant) modulation format that has
received great attention is optical differential quaternary phase-shift keying (DQPSK)
with differential detection, which can transmit two bits per symbol, corresponding to a
theoretical SE of 2 bits/s/Hz. To go beyond 2 bit/ s /Hz, polarization-division
multiplexing (PDM) has been suggested to further increase SE in combination with
DQPSK.
However, since the state of polarization of the light wave is not preserved during
transmission, dynamic polarization control is required at the receiver to recover the
transmitted signals, but it is not commercially available.
An alternative approach is to use independent intensity modulation on top of DQPSK,
resulting in eight-level amplitude-phase-shift keying.
In the past few years, research in high-capacity transmission has shifted to coherent
detection, in part to achieve high SE. Coherent detection was the subject of intensive
research in the 1980s because of its high sensitivity.
One of the main reasons that coherent optical communication was abandoned,
starting in the early 1990s,was the invention of EDFAs. Preamplified receivers using
EDFAs achieve sensitivity within a few decibels of that of coherent receivers, thus
making coherent detection less attractive, considering its complexity. Nowadays they
are not so complex with the solution of fast ADC and fast DSP, in addition no optical
PLL is required.
In coherent optical communication, information is encoded onto the electrical field of
the light wave; decoding entails direct measurement of the complex electrical field. To
measure the complex electrical field of the light wave, the incoming data signal (after
fiber transmission) interferes with a local oscillator (LO) in an optical 90° hybrid. If the
balanced detectors in some branches measure the real part of the input data signal,
the other branches, with the LO phase rotated by 90°, will measure the imaginary part
of the input data signal.
Phase and polarization tracking turned out to be the main obstacles for the practical
implementation of coherent receivers. The state of polarization of the light wave is
36
mixed in the fiber, this effect is dynamic and time dependent. Dynamic control of the
state of polarization of the incoming data signal is required so that it matches that of
the LO. Each dynamic polarization controller is bulky and expensive.
For WDM systems, each channel needs a dedicated dynamic polarization controller.
The difficulty in polarization management alone severely limits the practicality of
coherent receivers. Phase locking is challenging as well. All coherent modulation
formats with phase encoding are usually carrier suppressed. Optical PLL works with
carrier suppressed but they are not easily feasible. Therefore, conventional techniques
such as injection locking and optical phase-locked loops cannot be directly used to lock
the phase of the LO. Instead, decision-directed phase-locked loops must be employed.
At high symbol rates, the delays allowed in the phased-locked loop are so small that it
becomes impractical [5].
4.5. DIGITAL SIGNAL PROCESSING BASED COHERENT OPTICAL COMMUNICATION
SYSTEMS
What are the reasons that coherent optical communication is making a
comeback, and why is it possible now? The answer lies in advances in digital
signal processing (DSP).
The performance of digital signal processing equipment has been improved
dramatically over the past two decades, also the achievement of fast ADC are key
components wich makes it feasible to implement the complex signal processing steps
required to synchronize to the received signal in digital domain.
Both phase and polarization management can be realized in the electrical domain by
using DSP. Moreover, coherent detection in conjunction with DSP enables
compensation of fiber-optic transmission impairment, opening up new possibilities
that will likely shape the future of optical transmission technology.
Coherent optical communication system can utilize single or multiple carrier
transmitter and any modulation format, with QPSK being the most popular and higher
order quadrature-amplitude modulation (QAM) and phase-shift keying (PSK) systems
under investigation.
37
4.6. TECHNOLOGIES ASSOCIATED AT COHERENT OPTICAL SYSTEMS: WDM SYSTEM
EVOLUTION
In fiber-optic communications, wavelength-division multiplexing (WDM) is a
technology which multiplexes a number of optical carrier signals onto a single optical
fiber by using different wavelengths. This technique enables multiplication of capacity.
The term wavelength-division multiplexing is commonly applied to an optical
communication (which is typically described by its wavelength), whereas frequency-
division multiplexing
typically applies to a radio carrier (which is more often described by frequency). Since
wavelength and frequency are tied together through a simple directly inverse
relationship, the two terms actually describe the same concept.
By the mid 1990s, the erbium-doped fiber amplifier (EDFA) had made WDM highly
attractive because it could simultaneously amplify many WDM channels. This allowed
the capacity of fiber-optic communication systems to scale in the wavelength domain
by two orders of magnitude compared to single-channel systems [12].
Figure 4-5. Optical system using EDFA and WDM technique [12].
Up until more or less year 2000, achieving a closer WDM channel spacing was a matter
of improving the stability of lasers and of building highly frequency selective optical
filters; pre-2000, the increase in spectral efficiency, was therefore due to
improvements in device technologies.
When 40 Gb/s systems started to enter optical networking at the turn of the
millennium, optical modulation formats and coding became very important, first to
improve sensitivity so that the reach of 40 Gb/s systems would not fall too short of
that of legacy 10 Gb/s systems.
With the simultaneous development of stable 100 GHz and 50 GHz spaced optics, the
modulated optical signal spectra quickly approached the bandwidth allocated to a
single WDM channel, wich took the increase of spectral efficiency from a device design
level to a communications engineering level, and made spectrally efficient modulation
38
important, as it had traditionally been the case in electronic and RF communication
systems.
All indicate that WDM is still scaling in spectral efficiency and capacity at present but
will likely reach fundamental as well as practical limits in the near future. Therefore,
new approaches have to be explored in order to continue the scaling of capacity-
constrained systems. Such approaches could include the use of lower nonlinearity or
lower-loss optical transmission fiber, transmission over extended wavelength ranges,
or even the use of multi-core to increase distance capacity or multi-mode optical fiber.
4.7. COHERENT DETECTION FOR 40G/100G NETWORK DEPLOYMENT
Without a doubt, the networking performance advantages of coherent technology are
considerable. With coherent detection, the phase information of the optical signal is
preserved after electro-optic detection, allowing the optical distortion effects such as
chromatic and polarization mode dispersion (PMD) to be compensated electronically.
This is the only way to do it in digital coherent detection schemes (the preferred
embodiment in the optical industry), as the adaptive equalizer consists of a digital
signal processor (DSP), typically implemented in CMOS ASIC technology, which is low
cost to produce in volume.
Figure 4-6. Technology trends in 40/100G [7].
However, as is typical in this industry, a “one solution fits all” idea that coherent
technology wins in all applications is incorrect, nor is it likely to be in the near future.
Fundamentally, coherent optical systems require much more complex electro-optics
than direct-detection schemes. Typical 40G/100G coherent systems likely will use the
39
polarization-multiplexed quadrature phase-shift keying (PM-QPSK) modulation
scheme. This approach requires:
· two lasers, one at the transmissor and one at the local oscillator
· dual polarization nested Mach-Zehnder modulators
· four driver amplifiers
· four balanced photodiodes
· optical passives for polarization beam combining/splitting and phase diversity
· 2 x 90 degree hybrids
Compare this to differential phase-shift keying (DPSK), which requires a single
laser/driver amp/standard intensity modulator/photodiode and delay interferometer.
The increased complexity of coherent schemes simply translates into increased cost.
One smart thing the industry is doing at 100G is to standardize the modulation scheme
and integrated photonics in the optical internetworking forum (OIF). This certainly
helps the cost structure but, at least in the early years, not enough to offset the more
complex transmit/receive design for coherent.
In some network applications, using coherent detection is the point, even though the
transponder cost is higher but can carry more bits at distances where no other format
goes. For example, the PMD tolerance of direct-detection schemes using 40G DPSK is
around 3 ps mean, or around 8 ps if used with a PMD compensator or for a RZ-DQPSK
modulation format. PMD beyond these levels can easily be outperformed by using
coherent detection. In addition, assuming a clean design with low implementation
penalty, coherent detection offer a 2- to 3-dB OSNR improvement, enabling greater
distance for trans-oceanic submarine or terrestrial ultra long haul (ULH) applications.
The bottom line, though, is that while there are network applications where coherent’s
transponder cost premium can be justified by the reduction in optical-to-eletronical-
to- optical (OEO) regenerators required at the network level, there are others where it
can’t. The marketing and performance advantages of coherent detection make for an
easy sell -- but economic reality means that there will be many metro, regional, long
haul, and even submarine links where 40G DPSK or DQPSK will offer the best
price/performance tradeoff.
Direct detection has dominated 40G deployments to date, with strong demand
forecasts in 2011, 2012, and 2013. Coherent 40G technology will begin deployments in
especially challenging applications such as very high PMD older fibers or trans-oceanic
40
submarine. But wide-scale coherent technology happen with 100G matures, nowadays
100G coherent is commercially available but not yet with full speed deployment.
The performance advantages of coherent really become a “must-have” in the majority
of applications.
For 100G coherent, the use of integrated photonics is also expected to provide a
competitive $/bit/s cost structure at a fairly early stage in the technology life cycle.
Even after 100G availability, 40G direct detection is likely to survive as a dominant
technology in metro/regional networks and smaller national networks where 100G
pipes are still too big to fill efficiently [7].
41
5. POLMUX-QPSK MODULATION AND COHERENT
DETECTION
5.1. INTRODUCTION
The recent progress on high-speed digital signal processing enables the use of digital coherent receivers. A digital coherent receiver basically combines the state-of-the-art in optics and electronics; coherent detection as the theoretically optimum detection principle and a receiver implementation that is equalize for practically any amount of linear transmission impairments. Although this is a theoretically field of optical communications, it mirrors the technical evolution taken years ago in radio and wireline communication. As such, it is likely that digital coherent receivers will rapidly become the undisputed technology of choice in optical transmission systems. The rapid shift away from direct-detection receivers and towards digital coherent receivers is fuelled by a number of technology drivers. Among others, digital coherent receivers have spurred the use of higher order modulation formats, like quadrature phase shift keying [QPSK], polarization-multiplexing, the compensation of linear transmission impairments such as chromatic and polarization-mode dispersion (PMD), the design of Erbium-doped fiber amplifiers without mid-stage access, as well as improvements in optical performance monitoring. Polarization represents a key domain of the optical wave that can be readily exploited. A straightforward example is the recent interest in optical transmission systems that use a polarization-multiplexed (Pol-MUX) data channel. Such pol-muxed transmission doubles the system spectral efficiency (bits/s/Hz) and is more tolerant to fiber dispersive effects. When combined with advanced modulation formats and high-bandwidth, spectrally efficient transmission is possible [1] [11]. 5.2. EVOLUTION ON OPTICAL MODULATION The on and off optical signals were generated in the optical modulator according to the signal information of “1” and “0” in on-off keying (OOK) direct detection modulation. OOK was widely applied to systems with data transmission rates up to 10 Gbit/s because it lets us make a simple optical modulator and receiver configuration, and it has been the only solution for many years.
42
To achieve even greater capacity and long-distance transmission, the important technical issue is how to increase the spectral efficiency while maintaining the tolerance of the optical signal-to-noise ratio. In order to achieve this, optical phase shift keying schemes such as differential phase shift keying (DPSK) and differential quadrature phase shift keying (DQPSK) have been investigated. For example, DQPSK has been applied to the abovementioned 40-Gbit/s/ch WDM systems. In the case of DQPSK, 2-bit signals are modulated and assigned to four optical phases, which results in a two-fold improvement in spectral efficiency compared with OOK. DQPSK is thus able to relax the limitations on both optical bandwidth and transmission distance due to chromatic dispersion. Polarization division multiplexed quadrature phase shift keying (PDM-QPSK) or (PM-QPSK) with coherent detection is being investigated for the next-generation 100-Gbit/s/ch systems. PM-QPSK multiplexes two QPSK signals in the polarization domain. It has the great advantage of mitigating the problems of increasing both the electrical analog amplification bandwidth and the digital signal processing speed because it can reduce the symbol rate to 1/4 of the data transmission rate [8].
Figure 5-1. Different optical modulation formats [8]. 5.3. PM-QPSK TRANSMITTER AND RECEIVER ARCHITECTURE A POLMUX-QPSK (PM-QPSK) transmitter consists of two quadrature modulators and a polarization beam combiner (PBC) to multiplex the two outputs on orthogonal polarizations. At the receiver side, the received optical signal is split in two tributaries with arbitrarily, but orthogonal, polarizations using a second PBS. Both tributaries are subsequently mixed in a 90° hybrid structure with the output of a local oscillator. The outputs of the 90° hybrids (in-phase and quadrature components of both polarizations) are then detected with 4 photodiodes (either balanced or single-ended)
43
and converted to the digital domain using high-speed analog-to-digital converters (ADCs). Next figure shows the constellation diagram of POLMUX-QPSK modulation, represented as a 4-dimensional hypercube. The hypercube is described by the optical phase (in-phase and quadrature) on each of the two polarizations (Φv and Φh), where R and r are the outer and inner radius of the torus (with R > r).
Figure 5-2. Polmux-qpsk constellation diagram [11]. As POLMUX-QPSK modulates 4 bits per symbol, a low symbol rate of over 28 Gbaud is sufficient to obtain a 112-Gb/s line rate. This translates into a 100-Gb/s net data rate when a forward-error correction (FEC) overhead of ~7% and an Ethernet overhead of ~4% are subtracted. The lower symbol rate improves the tolerance to linear transmission impairments, which in turn allows for less stringent requirements on the electrical equalization, as well as making it possible to use lower bandwidth electrical components. The combination of POLMUX-QPSK modulation and coherent detection allows for an OSNR requirement close to the theoretical optimum. At a BER of 1e-3, 111-Gb/s POLMUX-QPSK typically requires an OSNR >15 dB, with 0.1 nm resolution bandwidth. In comparison, the OSNR requirement of filter-tolerant 43-Gb/s DPSK (today’s most widely deployed 40G format) is approximately 13.5 dB. The mere 2 dB difference despite the factor of 2.5x increase in data rate underlines the excellent OSNR performance of POLMUX-QPSK combined with coherent detection [7] [11].
44
5. 4. 100 GB/S PM-QPSK
Since 2005, to keep up with growing traffic demands in core networks, carriers have
upgraded their existing 10 Gb/s designed links with 40 Gb/s channels . As network
demands continue to grow rapidly, carriers will have to explore the possibilities for
further network upgrades. An important question is, with 100 Gb/s on the horizon,
is it practical to upgrade these existing networks with 100 Gb/s channels? A critical
enabler for the 40 Gb/s network upgrades was the introduction of advanced
modulation formats, allowing the retrofit of 40 Gb/s data channels into existing 10
Gb/s dense wavelength-division multiplexed (DWDM) transport systems. Modulation
formats such as optical duobinary (ODB), differential phase shift keying (DPSK), and
differential quadrature phase shift keying (DQPSK) have all been deployed in carrier
networks.
One common property of these modulation formats is the support of 50 GHz DWDM
channel spacing. Earlier 40 Gb/s modulation formats that did not support 50 GHz
channel spacing were not deployed, since a primary driver for deploying higher line
rates is to improve spectral efficiency and thus maximize capacity on existing DWDM
systems and fiber pairs. At 100 Gb/s, improved spectral efficiency to meet Internet and
video traffic growth is again expected to be a key stimulus, with support of 50 GHz
channel spacing still a key requirement. To facilitate ease of networking, tolerance to
transmission through many reconfigurable optical add/drop multiplexer (ROADM)
nodes is also essential, as express channels will often transit through a large number
of these ROADM nodes. Each ROADM permits each wavelength channel to be added,
dropped, or passed through at that node, completely in the optical domain (i.e.,, not
converted to an electronic signal). Thus, each ROADM acts as an optical filter,
constraining the bandwidth of the DWDM signals.
For 100 Gb/s, significant research has been done recently on advanced modulation
formats like 8-PSK/quadrature amplitude modulation (QAM), 16-QAM or 32-QAM.
Coding more than 1 b/symbol is essential to reduce the spectral width of the signal. At
100 Gb/s, it is necessary to code at least 3 b/symbol to narrow the spectrum
sufficiently to operate through 50 GHz filters. For this trial we use the PM-QPSK
modulation format that codes 4 b/symbol (modulating each of two orthogonal
polarization tributaries with both in-phase and quadrature-phase components). The
spectral width of 100 Gb/s PM-QPSK is sufficiently narrow to allow use of powerful
forward error correction (FEC) with a 20 percent overhead. Although the FEC increases
the line rate, symbol rate, and spectral width of the signal, the signal can still
propagate through multiple cascaded 50 GHz ROADMs with adequate performance.
The FEC with higher coding gain enables improved optical signal-to-noise ratio (OSNR)
sensitivity and thus
45
longer reach (propagation distance) between optical-to-electrical-to-optical (OEO)
regeneration points, thereby reducing the network cost.
Another advantage of PM-QPSK is the significant effort on implementation agreements
undertaken recently by the Optical Interworking Forum (OIF). The OIF members are in
or near agreement on many of the necessary hardware blocks and interfaces to
support these modulation techniques. The OIF implementation agreements have not
included FEC and the digital signal processing (DSP) following the coherent detection,
for now leaving these two areas open for vendor innovation. Thanks to Moore’s law,
massive DSP functionality can be integrated today into a single chip, even at 100 Gb/s.
Furthermore, using coherent detection, the full E-field of the signal can be measured
in the receiver, leading to excellent tolerance to linear impairments, such as chromatic
dispersion (CD) and polarization mode dispersion (PMD). These impairments can
now be compensated in the electronic domain, making 100 Gb/s practical even on
older fiber plant [6].
5.4.1. PERFORMANCE CHARACTERISTICS OF 100 GB/S PM-QPSK
5.4.1.1. OSNR SENSITIVITY
Coherent PM-QPSK offers approximately 6 dB improvement in OSNR sensitivity
compared to binary on-off keying (OOK) for the same bit rate. As 100 Gb/s is 10 times
higher capacity than 10 Gb/s, any new 100 Gb/s modulation scheme would ideally
offer a 10 dB performance improvement, providing a comparable OSNR sensitivity to
10 Gb/s OOK.
Although difficult to achieve in practice, part of the performance shortfall can be
recovered by the use of a high-coding-gain soft decision forward error correction (SD
FEC) [10]. Depending on the particular algorithm, soft bit resolution, and overhead rate
selected, another 2–3 dB gain can be realized compared to the typical 7 percent
overhead enhanced FEC codes.
The rest of the shortfall can be made up by reduction in penalty allocations. In 10 Gb/s
OOK systems, there is often a penalty of 1 dB or more allocated for imperfect CD
compensation, and a similar penalty allocated for PMD.
A key advantage of coherent detection is that the electromagnetic phase information
is passed into the electronic domain, so powerful electronic dispersion compensation
(EDC) in the DSP can mitigate the distortions with very low residual penalty. Therefore,
by using 100 Gb/s PM-QPSK with SD FEC and EDC, there is a 6 dB improvement
for coherent detection, a 2–3 dB improvement for SD FEC, and a 1–2 dB improvement
due to reduced CD and PMD penalties. This results in a total improvement of 9–11 dB,
46
approaching the OSNR sensitivity of 10 Gb/s OOK systems, thus allowing 100 Gb/s PM-
QPSK to be deployed with a comparable reach to current 10 Gb/s OOK systems [6].
5.4.1.2. OPTICAL FILTERING TOLERANCE
Due to their 10 Gbaud symbol rate, 10 Gb/s OOK channels have a much narrower
spectral width than the 50 GHz channel filters used in DWDM systems.
This provides excellent tolerance to cascades of ROADMs, with negligible penalty after
transmission. Similarly, to ensure good ROADM tolerance at 100 Gb/s, a sufficiently
low symbol rate is required, since the spectral width of the signal scales with the
symbol rate.
Using 100 Gb/s PM-QPSK (~25 Gbaud) provides a clear advantage over higher symbol
rate formats. Coding even more bits per symbol results in a denser signal constellation
and leads to reduction in OSNR sensitivity. A 100 Gb/s PM-QPSK signal can tolerate
filter bandwidths below 30 GHz with minimal penalty, significantly better than direct
detection DQPSK and OOK formats. This exceptional filtering tolerance allows
deployment through a large number of ROADMs.
Using a higher-level coding scheme, with a resulting drop in OSNR performance, is not
necessary. Reducing the symbol rate has other practical advantages, such as easing the
implementation of the modem in a complementary metal oxide semiconductor
(CMOS) chip, and reducing the bandwidth requirements for the electro-optic
components. However, using higher complexity constellations to further lower the
symbol rates places more stringent requirements on signal and local oscillator laser
line widths and reduces nonlinear phase noise tolerance. All of these trade-offs must
be considered when choosing a modulation format [6].
5.4.1.3. CHROMATIC DISPERSION TOLERANCE
With EDC inside the modem chip, chromatic dispersion (CD) can be compensated
without optical tunable dispersion compensators. The amount of CD that can be
compensated inside the chip is a function of the number of taps in the finite impulse
response (FIR) adaptive filter and the time delay of each tap.
Installations of 10 Gb/s DWDM systems primarily utilize dispersion compensating fiber
(DCF) deployed throughout the network to limit the residual CD at the 10 Gb/s OOK
receiver to within ±400 ps/nm (typically) for long-haul systems.
It is quite straightforward to meet this range in a 100 Gb/s PM-QPSK EDC with a small
number of taps. However, if the system could be designed without DCF, it could have
significant improvements in performance. Usually a small spool of DCF, built in a
47
dispersion compensating module (DCM), is installed with each optical amplifier. DCF is
special fiber with larger CD than the transmission fiber and of opposite sign.
As a result it has higher loss per unit length and a smaller core diameter. Given these
characteristics, a line system with DCF needs more amplification (adding more noise)
and will introduce more penalties due to nonlinear interactions within the DCMs than
a system without DCF.
Figure 5-3. Effect of dispersion, pulse overlap (inter symbol interference) [10].
Since each DCM must be matched to the specific length and type of transmission fiber
in the preceding span, installation and maintenance of DCF-free systems would also be
simplified. In addition, carriers are interested in reducing latency in their networks to
improve the performance of delay-sensitive telecommunications applications, such as
Internet gaming and network storage.
For these reasons, carriers would prefer to eliminate DCF for next-generation transport
networks. Eliminating DCF greatly increases the required dispersion tolerance,
particularly for standard single-mode fiber (SSMF, G.652) links, and greatly impacts the
EDC complexity, as a large number of taps are now required to fulfill the required
dispersion tolerance. Higher EDC complexity increases the chip gate count, which
in turn increases chip power consumption and reduces yield [6] [10].
5.4.1.4. POLARIZATION MODE DISPERSION (PMD) TOLERANCE
The EDC can also compensate PMD without optical PMD compensators. The number of
taps needed for PMD compensation is relatively small, as the pulse energy distortion
from PMD only spills into a few adjacent time slots.
One key parameter for PMD mitigation is that it must be fast enough to track the rapid
polarization dynamics that can occur in carrier networks. This is in contrast to CD
48
compensation, which is more static, changing very slowly and by small amounts due to
variations in the fiber temperature. Fortunately, PMD-mitigating algorithms can handle
very fast changes in either the received state of polarization or the instantaneous value
of the PMD because the tap coefficients are updated at a rate on the order of the clock
frequency of the DSP [10].
Figure 5-4. Pulse spreading due to dispersion effects [10].
5.4.2. UPGRADING EXISTING TRANSMISSION LINKS TO 100G At first sight the advantage of digital coherent receivers might not be entirely straightforward as the transmitter and receiver complexity is generally much higher in comparison to established direct-detection modulation formats, such as 43-Gb/s differential phase shift keying (DPSK) or even 43-Gb/s differential quadrature phase shift keying (DQPSK). However, when we look at the overall system complexity, the advantages of POLMUX-QPSK modulation and digital coherent receivers are much more pronounced. In order to upgrade existing transmission links to a 111-Gb/s line rate, the modulation format should be able to cope with all of the transmission impairments incurred by the already installed equipment. This includes transmission over high-PMD fiber, installed dispersion compensation modules, like dispersion compensating fiber [DCF] or fiber Bragg gratings [FBGs], as well as a limited optical bandwidth through cascaded filtering in photonic cross-connects (PXC).
49
Figure 5-5. PMD tolerance of a 100 Gb/s PM-QPSK system [6]. At a 111-Gb/s line rate the optical spectrum of DQPSK modulation is too broad to fit within a 50-GHz channel grid, making it incompatible to most field-installed transmission systems. 111-Gb/s POLMUX-QPSK modulation combined with a digital coherent receiver, on the other hand, can compensate for dispersion map deviations, PMD , FBG-induced phase ripples, as well as being more tolerant to the optical filtering incurred by cascaded PXCs on a 50-GHz channel grid. This will enable transmission links that cannot support 43-Gb/s line rates using direct detection receivers to upgrade to 111-Gb/s line rates using POLMUX-QPSK and digital coherent receivers. Careful system design is in particular required when POLMUX-QPSK modulated channels co-propagate with other modulation formats on neighboring WDM channels. For example, when 111-Gb/s POLMUX-QPSK co-propagates with 10-Gb/s on-offkeying (OOK) channels at 50-GHz channel spacing, the nonlinear tolerance is reduced by approximately 4 dB, but it depends on the channel spacing. Intermixing of 10-Gb/s OOK and POLMUX-QPSK modulation can therefore severely limit nonlinear transmission over long-haul distances. However, these penalties can be lowered by optimizing the respective channel powers or by spectrally separating the 40G/100G channels from already deployed 10G channels [6].
50
5.4.3. RECEIVER COMPLEXITY OF DIGITAL COHERENT DETECTION Compared to direct-detection receivers, coherent detection and the associated digital signal processing imply a significant shift in system complexity from the transmission link to the transmitter and receiver. The optical components in a POLMUX-QPSK transmitter and receiver require a higher complexity compared to more conventional direct-detection modulation formats. Optical integration might therefore be one of the promising directions to reduce footprint, power consumption and improve optical specifications. For example, a single POLMUX-QPSK Mach-Zehnder modulator at the transmitter or an integrated quad photo-diode array combined with a 90° hybrid structure at the receiver are promising directions of optical integration. In addition, there is a shift in complexity from the optical into the electrical domain. In particular the ADCs are a key component for any digital coherent receiver. Ideally the optical signal is converted to the electrical domain using a factor of two over-sampling, which implies that ~60-Gsample/s ADCs are required to realize a 100G coherent receiver. The design of 60-Gsample/s ADCs that allow for a >18-GHz electrical bandwidth, effective vertical resolution of at least 4 bits, and a power consumption of only a couple of Watts is truly challenging and requires state-of-the-art mixed signal design. The same is true for the receiver-side digital signal processing, which may consist of as many as 40 to 100 million gates and therefore requires state-of-the-art 40 nm or 65 nm CMOS processes. In addition, to limit the power consumption associated with inter-chip communication both the ADCs and digital signal processing are preferably integrated on a single-chip. Finally, an important consideration for 100G transmission is the implementation of advanced FEC coding and de-coding schemes. Due to the 25-Gbaud symbol rate it is possible to add up to 20% FEC overhead without incurring significant optical filtering penalties. The use of low-density parity check codes (LDPC) with ~20% overhead combined with soft-decision decoding enables a coding gain of up to 11 dB at a 10-15 BER, which is a 2-3 dB improvement in effective coding gain over the FEC codes used today at 43-Gb/s line rates [8].
51
Figure 5-6. Measured back to back BER performance versus OSNR on a 126.5 Gb/s coherent PM-QPSK system [8].
52
6. MODELING COHERENT PM-QPSK SYSTEM OptSim is a software tool for the design and simulation of optical communication systems at the phisical level. With state-of-the-art simulation techniques, an easy-to-use graphical user interface and lab-like measurement instruments, OptSim provides unmatched accuracy and usability. High-density optical systems operating over 100 Gbps require advanced transmission schemes for accurate delivery of data over long reaches. Next researches on OSNR levels and BER quality are focused on a PM-QPSK system simulated with an OPTSIM schematic structure. The OPTSIM software has library’s, object models, and compound components that help us to analize and plot diverse qualities of different kind of systems structures. Coherent phase modulation technologies coupled with polarization multiplexing have been developed. In one such approach, polarization-multiplexed quadrature-phase-shift-keying (PM-QPSK), four data signals are used to generate a single optical signal, chere each of its polarizations supports a QPSK-modulated data-signal pair. On this PM-QPSK schematic structure, we can modify several parameters inside different layout levels to set up the transmitter and receiver structures. This chapter shows all the elements that composes the whole structure with some of the most relevant parameters, whereas some of them are changed into the different realizations in order to obtain different configurations.
53
6.1. PROJECT LAYOUT COHERENT PM-QPSK SYSTEM
Figure 6-1. Schematic for the optical transmitter/receiver with a link. At first look at the system, the transmitter it’s composed of nine parallel PM-QPSK transmitters, implementing a 9-channel PM-QPSK WDM system. Each one works at a different center frequency, and generates a single PM-QPSK signal from four 28 Gbps data channels (overhead is included to account for forward error correction), using a channel spacing of 100 Ghz. The maximum transmission frequency is 193.4 Thz, and minimum transmission frequency is 192.6 Thz. These optical signals generated by the transmitters are filtered and combined before a transmission over 1800 km of fiber (without dispersion compensation). The fiber link, is composed by an iteration loop of a single fiber section of 90 Km. After the fiber, some optical noise is added on the optical signal and filtered again. In optical fiber communications, the practice of adding broadband noise from an amplified spontaneous emission (ASE) source is often used to determine a system’s tolerance, particularly the receiver, to OSNR. This practice is referred to as NL within this context. It has been assumed that this practice is equivalent to the effect of noise build up in real amplified transmission system making use of erbium doped fiber amplifiers (EDFA). On the receiver side, one 90 ° hybrid single ended combinator divide the signal into two polarization components and returns the ‘x’ and ‘y’ component of each polarization.
54
This four signals transports codification of the 4 bits (1 bit per 1 signal) that the PM-QPSK modulation transmittes. Typically, the receivers necessary for demodulating a PM-QPSK signal must be able to extract the relevant polarization information from the signal, as well as provide a phase-locked local oscillator. However, the use of digital signal processing (DSP) can dramatically simplify the receiver design. In a DSP-based coherent receiver, the local oscillator doesn’t need to be phase-locked to the signal, nor is complicated polarization handling required. Instead, receiver circuitry is used to convert the received PM-QPSK signal into electrical signals representing the in-phase and quadrature portions of each optical polarization. DSP circuitry is then used to recover the polarization and the phase of the signal. 6.1.1. PRINCIPAL PARAMETERS AND MODULES DESCRIPTION Center frequency (Thz): 193 Bandwidth : 0.9844 Reference Bitrate (Gb/s): 26.75 Samples per bit: 46 Some basic information about modules used on the schematics is given on this point.
55
Transmitter Optical filter Optical combiner
Optical amplifier
Iteration loop
Attenuation 1 = 18 dB
Attenuation 2 = 3 dB
Optical White noise generator
Output power = 9 dBm
Optical Gaussian filter is used -3dB Two-Sided Bandwidth = 40 Ghz
One-sided noise spectral density = 4.57839 dB (mW/Thz)
Iteration of a 90 km optical fiber section Number of iterations = 20 , Total 1800 Km Fiber Loss = 0.22 dB/Km Dispersion at the reference frequency = 3.8 ps/nm/km Fiber PMD = 0.02 ps/km^0.5
Relative intesity noise (RIN) = -155 In general, RIN is normalized to 1 Hz bandwidth
In this case a realistic version of PM-QPSK. Its structure is shown on next schematic
This module implements an optical white noise generator, a source that generates a spectrally flat (over all the simulation bandwidth) Gaussian random process. It can be useful in several situations, like in the debug phase of a simulation or to characterize the transfer function of the user defined filters and the generic optical devices.
Optical attenuator
Single Ended 90-
Degree Hybrid
including Local
Oscillator
The input optical signal is split into the two polarization components by a polarization beam splitter (PBS). The two resulting signal components are sent to two 90 degree hybrid that allows “beating” between local oscillator and incoming signals. Hence, on the 4 output signals we obtain three terms: 1. One CW term (bias) proportional to the LO power 2. One term proportional to the power of incoming signal (interference) 3. One term proportional to the amplitude of incoming signal times the amplitude of the LO: the useful part of the signal
56
PIN photodiode Electrical Filter Trans impedance amplifier
Gain = 46.02
Dcomp_psnm = -6840 ps/km
- 3dB bandwidth = 53.5 Ghz Quantum efficiency = 0.47891 Responsivity = 0.6 A/W
- 3 dB bandwidth = 18.725 Ghz
Electrical dispersion
compensator
Memoryless
Blind Receiver
for Coherent
Polarization
Multiplexed
QPSK
Modulation
This component recovers the Jones matrix of the channel and applies its inverse in order to separate data flows transmitted on orthogonal polarizations. Then it estimates the average local-oscillator-to-signal phase in order to allow separation between in-phase and quadrature signals. After application of these steps error counting is applied on the resulting 4 signals. Resulting estimated BER is displayed for each data flow as well as averaged and related to the aggregate flow.
Provided that the LO power is large with respect to the power of the received signal, the interference terms are small, and, except for bias terms, the output are proportional to the in-phase and quadrature components of x and y component of received optical signals. Inserting photo-detectors at the four output ports we can obtain four electric current signals proportional to the mentioned four components of received optical signal, allowing coherent detection.
57
6.2. INSIDE COHERENT PM-QPSK SYSTEM On this chapter we will see the layouts of TX and RX of PM-QPSK inside the principal section. Figures below shows the “polmux qpsk realistic” layout and “polmux qpsk” layout, this two schematics composes the nine compound component transmissors of our PM-QPSK system. An overview of principal parameters and modules description are also illustrated for all the subsystems, giving some values and descriptions of the most interesting parameters. 6.2.1. PROJECT LAYOUT POLMUX QPSK REALISTIC
figure 6-2. Schematic for TX polmux qpsk realistic The figure above shows the internal structure of the PM-QPSK transmitter, as we see, four data source are introduced to four non-return-to-zero (NRZ) drivers, and these electrical signals are filtered before being introduced to the “polmux qpsk” compound component adding a bias electrical wave.
58
At the output of this component we have an optical signal, wich is the PM-QPSK signal. 6.2.2. PRINCIPAL PARAMETERS AND MODULES DESCRIPTION
compound component
polmux-qpsk
Type of filter: Bessel filter Number of poles = 5 -3 dB Bandwidth = 24.71467
Output power = -20 dBm
This component simulates a
generator of constant level signal.
Data Source (PRBS generator), simulates a pseudo-random or a
deterministic logical signal generator of arbitrary level (number
of bits per symbol).
data source
electrical driver
electrical filter
bias wave generator
Data Source (PRBS generator), simulates a pseudo-random or a
deterministic logical signal generator of arbitrary level (number
of bits per symbol).
Data Source (PRBS generator), simulates a pseudo-random or a
deterministic logical signal generator of arbitrary level (number
of bits per symbol).
optical amplifier
fixed output power
optical amplifier
fixed output power
The layout of this compound component is shown on next
figure.
59
6.2.3. PROJECT LAYOUT POLMUX QPSK
Figure 6-3. schematic for polmux qpsk
This layout shows how we construct the PM-QPSK modulation, by using a laser source
and splitting the signal into two QPSK modulations. After apply a polarization rotator
on one of them and then combining the two signals we obtain the PM-QPSK signal.
This compoud does not refer to a real component, it aggregates artifically laser
modulator and PBC.
6.2.4. PRINCIPAL PARAMETERS AND MODULES DESCRIPTION
Cw = -50 dBm
Laser linewidth = 0 Mhz
laser source
60
Poincare’ Sphere Rotation Angle =
180
degree compound component
qpsk modulation
Optical combiner
Polarization rotator
Optical splitter
Poincare’ Sphere Rotation Angle about Axis S3 = 180 degrees
61
7. SIMULATIONS AND RESULTS 7.1. INITIAL ASUMPTIONS The aim of the following simulations is to understand the impact on OSNR and BER level at reception in a 100 Gb/s PM-QPSK, sometimes modifying one or more parameters of our system. These parameters dictate the available degrees of freedom we have to experiment. With this first set of simulations we want to define sensitivity in normal condition for sensitivity, it means OSNR required for a given BER, so we need the BER vs OSNR diagram. An initial outline of the system to understand what is being done would be:
Figure 7.1. Basic transmission scheme In long-haul WDM systems, erbium-doped fiber amplifiers are used to provide a wide and flat gain spectrum in order to accommodate and amplify as many WDM channels as possible. ASE noise emitted from the erbium-doped fiber (EDF) adds to signal and grows rapidly along cascades of optical fiber span. This ASE noise it’s an unavoidable disturbance that produces on reception a smearing of the constellation signal points, this may change a sample from the detection region, increasing the probability of error. ASE noise influence receivers BER, making detection more difficult as it’s shown on figure below:
Optical fiber link
62
The main point of study is the phase noise level on our system. It was found that the increased linewidths on optical lasers is achieved from a coupling between intensity and phase noise, caused by a dependence of the refractive index on the carrier density in the semiconductor. The output of a single-frequency laser is not perfectly monochromatic but rather exhibits some phase noise. Laser sources phase noise constitutes one of the fundamental limitations for coherent transmission systems employing single-mode optical fiber, wich are sensitive to the phase-modulation noise in the optical carrier and local oscillator waves. Phase noise is directly related to frequency noise, as the instantaneous frequency is essentially the temporal derivate of the phase. We have studied the impact of different laser with different linewidth on system performances, increasing or decreasing the receiver BER level.
Figure 7.2. Constellation regions due to ASE noise
63
Simulations were carried out transmitting 333000 bits for each of the 4 streams at 27.75 Gb/s. The number of symbols in a frame are 27750, this means we are sending a total of 3 frames on the global transmission. The frame length is composed by two sections, first of all we have a header or training section with his own number of symbols, and then it goes the tracking bit-stream. Initially the number of training symbols corresponds to a 2% overhead versus the total number of symbols on the frame, this means a number of 555 symbols, but this value will be changed on some of next simulations in order to see the impact on OSNR and BER levels. Finally, the simulated system we are going to consider, uses a DSP algorithm LMS (Least Mean Squares) adaptive filter. LMS algorithms are a class of adaptive filter used to mimic a desired filter by finding the filter coefficients that relate to producing the least mean squares of the error signal (difference between the desired and the actual signal). It is a stochastic gradient descent method in that the filter is only adapted based on the error at the current time.
One of the most important parameters on LMS algorithm is the step size or µ. The
step size determinates the convergence velocity of the desired signal. If its value is small, then the coefficients change only a small amount at each update, and the filter converges slowly. With a larger step-size, more gradient information is included in each update, and the filter converges more quickly; however, when the step-size is too large, the coefficients may change too quickly and the filter will diverge. On our simulations, we have two different step sizes:
µtraining for the training portion of the frame
µtracking for the tracking portion of the frame
By optimizing both of them, we obtain two more degree of freedom on our system to see the impact on OSNR and BER levels. All the simulations are realized by taking the optsim data and creating a matlab file that uses this results and plots what we are looking for, these matlab codes will be shown on the annex.
64
7.2. FINAL SYSTEM ARCHITECTURE AND REFERENCE PARAMETERS
On the chapter before I have introduced an initial layout of the PM-QPSK optimization
setup and parameters system to be managed.
This initial architecture is changed on the layout top level in order to reduce
simulation times. As example the fiber span iteration is removed but its effects are still
taken into account.
Also, the single-ended 90° hybrid is changed by another with 8 outputs that allows to
use a balanced photodetector configuration.
Figure 7-3. Back-to-back with noise loading
PM-QPSK transmitter Frequency = 193 Ghz Bitrate = 27.75 Gbps Laser Power = -20 dBm
-3dB Two-Sided Bandwith = 50 Ghz
Transmitter
Optical Gaussian Filter
65
Single-ended 90-
degree Hybrid with
4+4 Outputs
The input optical signal is splitted into the two polarization components by a polarization beam splitter (PBS). The two resulting signal components are sent to two 90 degree hybrid that allows “beating” between local oscillator and incoming signals. Hence, on the 4+4 output signals we obtain three terms: 1. One CW term (bias) proportional to the LO power 2. One term proportional to the power of incoming signal (interference) 3. One term proportional to the amplitude of incoming signal times the amplituted of the LO: the useful part of the signal
Amplifier 1 Output Power = 2 dBm
Amplifier 2 Output Power = 5 dBm
One-sided noise spectral density = 4.3309 dB{mW/Thz}
Optical combiner
Optical amplifier
Optical White noise
generator
66
Reference parameters:
Center Frequency = 193 Thz (1553.3288 nm)
Rb = 111 Gb/s (100 GE 64B/66B encoding + 7% FEC overhead)
Rs = 27.75 Gb/s
Frame length = 27750 symbols
Number symbols training = 555 (2% overhead)
Total Timespan simulated #bits @ center frequency = 333000
Compound component with the algorithm
to obtain the BER on each branch and the
total BER of the system.
-3dB bandwidth = 13.875 Ghz
Number of poles = 5
Substracts the two electrial inputs
Gain = 60
-3dB bandwidth = 55.5 Ghz
Responsivity = 0.6 A/W
PIN photodiode
Electrical substractor
Trans impedance
amplifier
Electrical amplifier
BER counter
67
7.3. SIMULATIONS AND ANALYSIS
On these simulations, we are going to see the impact of the use of different kinds of laser, with different kinds of laser linewidth on the system showed before. The procedure is to make diverse simulations changing the laser linewidth value on our system and see the impact on the BER and OSNR levels. Optical signal suffers more than only attenuation. In amplitude, spectrally, temporally signal interaction with light-matter and other signal disturbances such as power reduction, dispersion, polarization and unbalanced amplification. Thus leading to random noise, which causes misalignments, jitter and other disturbances resulting in erroneous bits, the rate of which is known as BER parameter. In order to run this simulations, the first step is to fix the value of the laser linewidth, and make a plot using the Optsim data and a Matlab code programmed to show a plot BER vs OSNR. By using this graphic, we look for the OSNR value at a desired BER = 1E-3. These will be the first approach, because the initial values for the two µ (tracking and training) are fixed at their hypothetical optimal value (log µ = -3.1) without phase noise. Later, we change the initial OSNR value with the one we found, and create another plot that sweeps the two values of µ around their hypothetical optimal values and shows the BER level regions for each combination of values. With the use of this plot, we can found a more realistic value for the two step size by searching the center of the graphic, and we return to the step one another time using these new values and searching the OSNR level for them. Once we have the real OSNR value, we make the second plot another time obtaining the final results.
68
7.3.1. LASER LINEWIDTH (PHASE NOISE) FROM 0 HZ TO 800 KHZ
· = 0 Hz
Optical Signal to Noise Ratio is the measure of the ratio of signal power to noise power in an optical channel. OSNR is important because it suggests a degree of impairment when the optical signal is carried by an optical transmission system that includes optical amplifiers. On this work we find de OSNR required for a given BER (i.e. sensitivity). The blue line corresponds to a direct detection measure (WDD) of the OSNR level at the system, and the red one corresponds to a measure without direct detection (WoDD). As it could be seen, the obtained OSNR level for a PM-QPSK system using 0 Hz source laser linewidth is around 13.8 dB on the WDD measure. Otherwise the WoDD measure is 0.6 dB under, it means a OSNR of 13.2 dB for a BER 1-E3.
69
In this section, the plots of µtracking vs µtraining are realized with WDD data, on sections below, plots using WDD and WoDD are included.
On this plot we can see the BER level regions for the different values of µtracking and training. The number over the colored lines should be interpreted as follows: -3 BER=1E-3. Our purpose is to find the optimals µtracking and training from the plot we have obtained, just to obtain the optimum working point, and the results are as follows:
∆λ (Khz) OSNR WDD @ BER 1E-3
Optimum Log µ training
Optimum Log µ tracking
0 13.8 -3.5 - 4.2
70
· = 10 Khz
∆λ (Khz) OSNR WDD @ BER 1E-3
Optimum Log µ training
Optimum Log µ tracking
10 13.95 -3.5 - 3.75
71
Comparing this results with the firsts we obtain, we can notice our sensitivity has increased 0.15 dB. The optimum working point for µ training remains the same, and the optimum working point for µ tracking varied 0.45 from its initial value.
· = 20 Khz
72
As we can see on this second comparison with the firsts we obtain, we can notice our sensitivity keeps increasing as we increment the laser linewidth, now 0.25 dB. The optimum working point for µ training changed 0.15, and the optimum working point for µ tracking varied 0.4 from its initial value.
· = 50 Khz
∆λ (Khz) OSNR WDD @ BER 1E-3
Optimum Log µ training
Optimum Log µ tracking
20 14.05 -3.35 - 3.8
73
· = 100 Khz
∆λ (Khz) OSNR WDD @ BER 1E-3
Optimum Log µ training
Optimum Log µ tracking
50 14.1 -3.15 - 3.35
74
On this comparison at 100 KHz, we see our initial sensitivity has increased 0.4 dB. The tendency of the optimum working points is to decrease from initial ones, but are still around the hypothetical ones (around -3.1).
∆λ (Khz) OSNR WDD @ BER 1E-3
Optimum Log µ training
Optimum Log µ tracking
100 14.2 -2.95 - 3.35
75
· = 300 Khz
∆λ (Khz) OSNR WDD @ BER 1E-3
Optimum Log µ training
Optimum Log µ tracking
300 14.6 -2.8 - 3.05
76
· = 500 Khz
∆λ (Khz) OSNR WDD @ BER 1E-3
Optimum Log µ training
Optimum Log µ tracking
500 14.95 -2.5 - 2.9
77
· = 800 Khz
∆λ (Khz) OSNR WDD @ BER 1E-3
Optimum Log µ training
Optimum Log µ tracking
800 15 -2.25 - 2.9
78
On the table below, we can see the OSNR evolution when we change the source laser
linewidth.
The OSNR is the ratio between the signal power and the noise power in a given
bandwidth. As we can see, increasing laser linewidth needs a higher level of OSNR to
ensure a BER=1E-3.
As a result, we can see that even optimizing µ’s, the more we increase the laser
linewidth, more power we need to assume the same sensibility (1.2 dB higher at 800
KHz).
LASER LINEWIDTH
(Khz)
OSNR @ BER 1E-3
WDD (dB) with
Optimum µ tracking
and µ training
0 13.80
10 13.95
20 14.05
50 14.10
100 14.20
300 14.60
500 14.95
800 15.00
7.3.2. LASER LINEWIDTH FROM 1 MHZ TO 10 MHZ
On this chapter, we are going to increase the sweep range on the laser source
linewidth just to look for which is the maximum value when we can achieve the
desired BER level of 1-E3.
79
The second plot types are also focused on the WDD Optsim results.
· = 1 Mhz
80
Total Timespan simulated #bits @ center frequency = 550000
The figure above shows a variation of the second plot. This time we changed the total number of simulated bits to see the impact on BER level regions. As it could be seen, it didn´t have a relevant impact, the main form of the figure was the same as it was with the precedent values of bits.
∆λ (Mhz) OSNR WDD @ BER 1E-3
Optimum Log µ training
Optimum Log µ tracking
1 15.2 -2.8 - 2.45
81
· = 2 Mhz
∆λ (Mhz) OSNR WDD @ BER 1E-3
Optimum Log µ training
Optimum Log µ tracking
82
· = 5 Mhz
With this value of laser linewidth, we observe that the signal becomes unstable, and
we can’t ensure a minimum level of BER neither increasing the OSNR. We cannot reach
the desired sensitivity.
This is also reflected at the plot below, where we can observe on the BER level regions
that the minimum desired BER=1E-3 is not achieved.
2 15.8 -2.6 - 2.3
83
On the center of the figure, we can observe a BER level of BER=1E-2.
· = 10 Mhz
Evidently, if a laser linewidth of 5 Mhz didn’t fit our expectations on BER level, if we try
increasing it until 10 Mhz, the results will be worse.
84
7.3.3. GRAPHIC COMPARISON BETWEEN LASER LINEWIDTH FROM 0 HZ TO 10 MHZ
85
On this BER vs OSNR plot, we compare all the different values we have obtained from
0 Hz to 10 Mhz.
This plot shows the WDD data version, also the plot has been done with the obtained
optimum values of µ tracking and µ training for each case.
Until now, the maximum stable value for the laser linewidth is 2 Mhz, and the
difference between the first value at 0 Hz and this one is 2 dB. This is the impact on
OSNR level caused by the phase noise added when we increased the laser linewidth.
This time WoDD data is shown on the plot. Like the graphic above, the impact of changing the laser linewidth from 0 Hz to 2 Mhz resulted in a difference of 2 dB.
LASER LINEWIDTH
(Khz)
OSNR @ BER 1E-3 WDD (dB) with
Optimum µ tracking and µ training
0 13.80
10 13.95
86
20 14.05
50 14.10
100 14.20
300 14.60
500 14.95
800 15.00
7.4. CHANGING OVERHEAD ON =
This time, we are going to focus the laser linewidth of 2 Mhz, and change the number
of symbols of the training sequence to see the impact it has on the BER and OSNR
levels of the system.
The original number of training symbols was 555, representing a 2% of the whole part,
changing its value to 277 we can get a 1% of overhead, also to get an overhead of 0.5%
the number of symbols must be 138.
Longer number of training symbols means less usefull information on the frame, this is
the motivation on that research.
7.4.1. OSNR SENSITIVITY COMPARISON @ BER=1E-3, OVERHEAD [ 2% -1% - 0.5%] Number of training symbols = {555,277,138}
With direct detection (WDD)
Optimum µ tracking and µ training
1000 2000
15.20 15.80
87
Changing the overhead of the frame has a low impact on OSNR @ BER=1E-3.The
shifting of the value is small, about 0.25 dB when we reduce the percentage from 2%
to 0.5%.
A shorter training means less overhead, so it's better as there is no significant
difference in BER vs OSNR, so we prefer a shorter training.
88
Without Direct Detection (WoDD)
On the figure above, we can see this time that the size of the training sequence
doesn’t affect the OSNR sensibility when we don’t use direct detection.
7.4.2. = OVERHEAD 1% NUMBER SYMBOLS TRAINING = 277
· WDD
89
·WoDD
7.4.3. = OVERHEAD 0.5% NUMBER SYMBOLS TRAINING = 138
·WDD
90
·WoDD
Figures above shows the impact on changing the overhead of the frame when laser
linewidth is 2 Mhz.
As we could see, reducing the number of training symbols didn’t affect substantially on
the BER region we want to ensure. On the horizontal axis, we notice a reduction of the
area, on the other hand, looking the vertical axis, we could see the region is 0.1 wider.
7.5. CHANGING OVERHEAD ON =
We have seen 5Mhz and 10Mhz didn’t fit as a possible source laser linewidth, now
3Mhz and 4 Mhz will be under simulation.
We are going to find out if the OSNR level is still stable around the desired BER, and
the evolution of the BER level graphics once setted the optimum µ and OSNR values.
7.5.1. OSNR SENSITIVITY COMPARISON @ BER=1E-3 OVERHEAD [ 2% - 0.5%] Number of training symbols = {555,138}
WDD
Optimum µ tracking and µ training
91
WODD
On this sheet we can observe that the instability point on both graphics is under 2 dB
of distance.
92
7.5.2. = OVERHEAD 2% NUMBER SYMBOLS TRAINING = 555
·WDD
7.5.3. = OVERHEAD 0.5% NUMBER SYMBOLS TRAINING = 138
·WDD
93
On the figure above, we could see the desired BER level region doesn’t exists when
WDD data is used.
Using 2% overhead this time we can observe an improvement in front of 0.5%
overhead plot using WDD data, now decreasing the length of the overhead we don’t
have a BER=1E-3 region.
7.6. CHANGING OVERHEAD ON =
7.6.1. OSNR SENSITIVITY COMPARISON @ BER=1E-3, OVERHEAD [ 2% - 0.5%] Number of training symbols = {555,138}
·WDD
Figure above shows how for a 4 Mhz laser linewidth the instability zone is near the
point we can ensure the 1E-3 BER, about 2 dB with direct detection.
94
7.6.2. = OVERHEAD 2% NUMBER SYMBOLS TRAINING = 555
·WDD
7.6.3. = OVERHEAD 0.5% NUMBER SYMBOLS TRAINING = 138
·WDD
95
With 3 and 4 Mhz laser linewidth we can observe BER level regions are smaller at
BER=1E-3 or indeed sometimes didn’t exists.
This issue added to the factor that instability points are closer ( < 2dB ) to the desired
BER level, makes this type of source laser linewidth less usefull instead of the ones
with a smaller laser linewidth.
96
8. CONCLUSION
8.1. MAIN FINDINGS
We have analysed the sensitivity evolution comparing on a 100 Gbps PM-QPSK system
by the use of different type of lasers with different type laser linewidth. We have
studied the behavior of the system from 0 Hz to 10Mhz laser source linewidths in
order to see the impact of the phase noise.
The best value for the laser linewidth is zero, but our goal is to analize the robustness
to a different type of linewidth optimizing µ tracking and µ training.
We have investigated the most interesting values of the laser linewidth using the
results of each one, and focusing on those which bring us desiderable results. Thanks
to this results, we’ve searched the limit when the laser linewidth brings us unstable
results and becomes unhelpful for our proposals.
Throught the first part of this work, we have introduced the most remarkable elements
and properties of the whole system, and the most common issues working with this
technology.
Then, at the simulation part and always optimizing the two values of µ tracking and µ
training we have found out a 2 dB OSNR difference between 0Hz and 2Mhz laser
linewidth. And from the other side, we have obtained unsuitable results when
increasing beyond 2 Mhz the source laser linewidth, making the results unstable or
similar at this situation.
We have noticed that these optimal values for µ tracking and µ training tend to zero as
you increase the laser linewidth.
Also, by the observation of the simulations changing the overhead, we noticed the
changes on BER level regions were weak, around 0.1-0.2 points at both axis of the
graphics. Changing the number of training symbols of the frame didn’t affects more
than the given value on the overall results.
97
8.2. NEAR FUTURE
Direct detection has dominated 40G/100G deployments to date, with strong demand
forecasts in 2012, and 2013. Coherent 40G technology will begin deployments in
especially challenging applications such as very high PMD older fibers or trans-oceanic
submarine. But wide-scale coherent technology 100G helps us where the performance
advantages of coherent really become a necessity as you want more capacity in the
majority of applications.
For 100G coherent, the use of integrated photonics is also expected to provide a
competitive cost structure at a fairly early stage in the technology life cycle. Even after
100G availability.
A 100-Gbit/s/channel system based on digital coherent technology is considered to be
a promising candidate for next-generation large-capacity long-distance optical
communication systems. The optical components required for such systems, such as a
PM-QPSK optical modulator, integrated receiver, and local light source, are
commercially available, but better systems to get full performance are in development.
Opto-electrical integration technologies, which enable us to construct small, low-cost,
and highly functional optical components, will play an important role in providing cost-
effective transmission equipment for future 100-Gbit/s/ch and post-100-Gbit/s/ch
optical communications.
98
Bibliography
[1] Koichi Murata and Takshi Saida
Optical Device Technologies for Future Network Evolution
NTT technical review, Vol. 9 No.3 mar. 2011
[2] M. ARUMUGAM, Department of Physics, Anna University
Optical fiber communication
Pramana – J.Phys., VOL. 57, Nos 5 & 6, Nov. & Dec. 2001.
[3] Nick Massa
Fundamental of photonics, Module 1.8
Springfield Technical Community College, Massachusetts 2000.
[4] Kazuro Kikuchi
Coherent optical communication systems
Department of Frontier Informatics, University of Tokyo, Kashiwa, Chiba, Japan.
[5] Guifang li
Recent advances in coherent optical communication
Advances in Optics and Photonics 1, University of Central Florida (2009).
99
[6] M. Birk, P. Gerard, R. Curto, Lynn E. Nelson, X. Zhou, P. Magill, T. J. Schmidt
C. Malouin, B. Zhiang, E. Ibragimov, S. Khatana, M. Glavanovic, R. Lofland
R. Marcoccia, R. Saunders, G. Nicholl, M. Nowell, F. Forghieri
Coherent 100 Gb/s PM-QPSK Field Trial
IEEE Communicactions magazine, sept 2010.
[7] Michael Finkenzeller
Delivering 100G per wavelength with today’s DWDM infrastructure
Nokia Siemens Networks - Motivation, Experiments and Standards RIPE 55, Amsterdam.
[8] M. Birk, P. Gerard, R. Curto, Lynn E. Nelson, X. Zhou, P. Magill, T. J. Schmidt
C. Malouin, B. Zhiang, E. Ibragimov, S. Khatana, M. Glavanovic, R. Lofland
R. Marcoccia, R. Saunders, G. Nicholl, M. Nowell, F. Forghieri
Field trial of a real-time, single wavelength, coherent 100 Gbit/s PM-QPSK channel upgrade of an installed 1800km link
IEEE Xplore, July 2010
[9] Tomas Urra B.
Sistemas Ópticos Coherentes
UTFSM, Noviembre 2005
[10] Lars Risby
Basic principles of fiber optic systems
Adva Optical Networking, 2006.
100
[11] D. vand den Borne, V. Sleiffer, M.S. Alfiad, S. L. Jansen, T. Wuth
POLMUX-QPSK modulation and coherent detection: the challenge of long-haul 100 G transmission
IEEE Xplore, ECOC 2009, 20-24 September, 2009, Vienna, Austria
[12] Razali Ngah
Optical time division multiplexing for optical communication system
IEEE Xplore, Research vote no: 78025
[13] Steve Hranilovic and Frank R. Kschischang
Optical Intensity-Modulated Direct Detection channels: Signal Space and Lattice Codes
IEEE Transactions on information theory, vol 49. NO.6 June 2003
[14] Y. Ben Ezra, B.I. Lembrikov, Avi Zadok, Ran Halifa and D. Brodeski
All-Optical Signal Processing for High Spectral Efficiency (SE) Optical Communication
Intech, Optical Communication ed. by Narottam Das, ISBN 978-953-51-0784-2
[15] www.wikipedia.org