Professional Documents
Culture Documents
Contents
Section 1 DWDM Overview 1.1 DWDM Technology Background 1.2 DWDM Principles Overview 1.3 DWDM Equipment Operating Modes 1.3.1 Two-fiber bi-directional transmission 1.3.2 Single fiber bi-directional transmission 1.3.3 Add and drop of optical signals 1.4 Application Modes of DWDM 1.5 Advantages of DWDM Section 2 DWDM Transmission Media 2.1 Optical Fiber Structures 2.2 Types of Optical Fiber 2.3 Basic Features of Optical Fiber 2.3.1 Physical Dimension (Mode field diameter) 2.3.2 Mode Field Concentricity Error 2.3.3 Bend Loss 2.3.4 Attenuation Constant 2.3.5 Dispersion Coefficient 2.3.6 Cutoff Wavelength 2.4 Types and Properties of Optical Fiber Cable 2.4.1 Types of Optical Fiber Cable 2.4.2 Properties of Optical Fiber Cable Section 3 DWDM Key Technologies 3.1 Lasers 3.1.1 Laser Modulation Modes 3.1.2 Wavelength Stability and Control of Laser 3.2 Erbium-doped Optical Fiber Amplifier (EDFA) 3.2.1 EDFA Operating Principle 3.2.2 Applications of EDFA 3.2.3 Gain Control of EDFA 3.2.4 Limitations of EDFA 3.3 DWDM Components 3.3.1 Optical Grating Type DWDM Component 3.3.2 Dielectric Film Type DWDM Component 3.3.3 Fused Conical Type DWDM Component 3.3.4 Integrated Optical Waveguide Type DWDM Component 3.3.5 Performances of DWDM Components
i
1-1 1-1 1-2 1-3 1-3 1-4 1-5 1-5 1-5 2-1 2-1 2-2 2-3 2-3 2-4 2-4 2-4 2-5 2-6 2-6 2-6 2-6 3-1 3-1 3-1 3-4 3-5 3-5 3-7 3-9 3-13 3-14 3-15 3-16 3-17 3-17 3-18
DWDM Principle
Section 4 DWDM Networking Design 4.1 Some Network Element Types of DWDM 4.1.1 Optical Terminal Equipment (OTM) 4.1.2 Optical Line Amplification Unit (OLA) 4.1.3 Optical Add/drop Multiplexing Unit (OADM) 4.1.4 Electrical Regeneration Unit (REG) 4.2 General Constitution of DWDM network 4.2.1 Point-to-point Networking 4.2.2 Chain Type Networking 4.2.3 Ring Type Networking 4.2.4 Network Management Information Channel Backup and Interconnection Capability 4.3 Factors To Be Considered in DWDM Networking 4.3.1 Dispersion Limited Distance 4.3.2 Power 4.3.3 Optical Signal-to-Noise Ratio 4.3.4 Other Factors 4.4 DWDM Network Protection 4.4.1 Protection Based on single Wavelength 4.4.2 Optical Multiplex Section (OMSP) Protection 4.4.3 Applications in Ring Networks 4.5 Analysis to The Examples 4.5.1 Networking Diagram (Physical Network Stations) 4.5.2 Networking Diagram (considering the dispersion limited distance of the lasers to divide the regenerator sections of the network) 4.5.3 Networking Diagram (considering the power of optical amplifiers to divide the optical regenerator sections) 4.5.4 Networking Diagram (considering OSNR)
4-1 4-1 4-1 4-2 4-3 4-4 4-5 4-5 4-6 4-6 4-7 4-9 4-9 4-10 4-11 4-14 4-23 4-23 4-25 4-26 4-27 4-27 4-27 4-28 4-29
ii
DWDM Principle
Objectives: To master the concepts of DWDM. To know the background and technology characteristics of DWDM.
1-3
DWDM Principle
2. Time Division Multiplexing (TDM) TDM is a commonly used method for capacity expansion, e.g. multiplexing of the primary group to the fourth group of the traditional PDH, and STM-1, STM-4, STM-16 and STM-64 of current SDH. TDM technology can enhance the capacity of optical transmission information in duplication and greatly reduce the circuit cost in equipment and line. Moreover, it is easy to extract specific digital signals from the data stream via this multiplexing method. It is especially suitable for networks requiring the protection strategy of self-healing rings. However, TDM method has two disadvantages. Firstly, it affects services. An overall upgrade to higher rate levels requires to replace the network interfaces and equipment completely. Thus the equipment in operation must be interrupted during the upgrade process. Secondly, rate upgrade lacks of flexibility. Let's take SDH as an example, when a system with a line rate of 155Mbit/s is required to provide two 155Mbit/s channels, the only way is to upgrade the system to 622Mbit/s even though two 155Mbit/s are unused. For TDM equipment of higher rate, the cost is relatively high. Furthermore, 40Gbit/s TDM equipment has already reached the rate limitation of electronic devices. Even the nonlinear effects of 10Gbit/s rate in different fiber types will set various limitations to transmission. Currently, TDM is a commonly used capacity expansion method. It can implement capacity expansion via continuous system rate upgrade. When certain rate level is reached, other solutions must be found because of characteristic limitations of devices, lines, etc. All the basic transmission networks, whether using SDM or TDM to expand the capacity, adopt traditional PDH or SDH technology, i.e. utilizing optical signals on a single wavelength for transmission. This transmission method is a great waste of optical capacity because the bandwidth of optical fiber is almost infinite when compared to the single wavelength channel we currently use. We are worrying about the jam of networks, on the other hand huge network resources are being wasted. DWDM technology emerged under this background. It greatly increases the network capacity, makes full use of the bandwidth resources of optical fibers and cuts down the waste of network resources.
1-4
DWDM Principle
1-5
DWDM Principle
The diagram of DWDM system structure and optical spectrum is shown in Figure 1-1. At the transmit end, optical transmitters output optical signals of different wavelengths whose accuracy and stability meet certain requirements. These signals are multiplexed via an optical wavelength multiplexer and sent to an erbium-doped optical fiber power amplifier (it is mainly used to compensate the power loss aroused by the multiplexer and enhance the launched power of the optical signals). After amplification, this multi-channel optical signal is sent to the optical fiber for transmitting. In the midway optical line amplifiers can be installed or not according to practical conditions. At the receiver end, the signals are amplified by the optical pre-amplifier (it is mainly used to enhance receiving sensitivity and prolong transmission distance) and sent to the optical wavelength de-multiplexer which separates them into the original multi-channel optical signals.
OTU
M
O TU
U X
D M U X
OTU
Optical pre-amplifier
Optical spectrum
Single channel
Optical spectrum
Wavelength
Wavelength
1-6
DWDM Principle
Optical source
1
Detector
1
WDM N
Optical source
N
WDM
1 N
Detector
N
N+1
Detector
1
Optical source
1
N+1
2N
WDM
Detector
N
WDM
1 N
Optical source
N
2N
This kind of DWDM system can effectively exploit the huge bandwidth resources of optical fiber and expand the transmission capacity of a single optical fiber in several or tens of times. In long-haul networks, capacity can be expanded by adding wavelengths gradually according to the demands of practical traffic, which is very flexible. This is, under the condition that the actual fiber dispersion isn't known, also an approach to use multiple 2.5Gbit/s systems to implement ultra-large capacity transmission, avoiding adopting ultrahigh speed optical systems.
1-7
DWDM Principle
Optical source
1
Detector
1
Optical source
N
Detector
N
WDM
1
WDM
N
N+1
Detector
N+1
Optical source
N+1 2N
N+1
N+1
2N
Detector
2N
Optical source
2N
2N
Figure 1-3 The DWDM system which adopts single fiber bi-directional transmission
Single fiber bi-directional transmission allows a single fiber to carry full duplex channels and, generally, saves one half of the fiber components of unidirectional transmission. Since signals transmitted in the two directions do not interact and create FWM (Four-Wave Mixing) products, its total FWM products are much less than two-fiber unidirectional transmission. However, the disadvantage of this system is that it requires a special measure to deal with the light reflection (including discrete reflection resulted by optical connectors and Rayleigh backward reflection of the fiber) to avoid multi-path interference. When the optical signal needs to be amplified to elongate prolong transmission distance, components such as bi-directional optical fiber amplifier and optical circulator must be adopted, but their noise factor is a little worse.
OADM
OADM
Detector
Optical source
1
1
Detector
Optical source
2
2
1-8
DWDM Principle
By utilizing optical add/drop multiplexer (OADM), optical signals of the wavelengths can be added and dropped in the intermediate stations, i.e. implementing add/drop of optical paths. This method can be used to implement ring type networking of DWDM systems. At present, OADM can only be made as fixed wavelength add/drop device (as shown in Figure 1-4) and thus the flexibility of this operating mode is limited.
The feature of open DWDM system is that it has no special requirements for multiplex terminal optical interfaces as long as they meet the optical interface standards defined in ITU-T G.957. The DWDM system adopts wavelength conversion technology to convert the optical signal of multiplex terminal into specific wavelength. Optical signals from different terminal equipment are converted into different wavelengths meeting the ITU-T recommendation, then multiplexed. Integrated DWDM system, without adopting wavelength conversion technology, requires that the optical signal wavelengths of the multiplex terminal meets DWDM system specifications. Different multiplex terminal transmits different wavelengths meeting the ITU-T recommendation. Thus, when connected to the multiplexer, these wavelengths occupy different channels and multiplexing is implemented. Different application modes can be adopted according to the demands of engineering. In practical applications, open DWDM and integrate DWDM can be mixed.
1-9
DWDM Principle
generation optical fiber communication technology based on dense-WDM (DWDM) has emerged. DWDM technology has the following features: 1. Ultra-large capacity The transmittable bandwidth of currently commonly used conventional fiber is very wide, but the utilization ratio is still low. By using DWDM technology, the transmission capacity of a single optical fiber is increased by several, tens of or even hundreds of times when compared to the transmission capacity of single wavelength systems. Recently, NEC Company, Japan, implemented a 13220Gbit/s experimental DWDM system with a transmission distance of 120km. This system, with a total bandwidth of 35nm (1529nm~1564nm) and a channel spacing of 33GHz, can transmit 40 million telephone calls. 2. Data rate "transparency" DWDM systems conduct multiplexing and de-multiplexing in terms of optical wavelength differences and are independent to signal rates and modulation modes, i.e. transparent to the data. Hence, they can transmit signals with completely different transmission characteristics and implement combination and separation of various electrical signals, including digital signals and analog signals, and PDH signals and SDH signals. 3. Utmost protection of the existing investment during system upgrade During the expansion and development of the network, it is an ideal approach to implement capacity expansion without the need to rebuild the optical fiber cables and with the only requirement of replacing the optical transmitters and receivers. This is also a convenient way to introduce broad-band services (such as CATV, HDTV and B-ISDN). Furthermore, any new services or new capacity can be introduced simply by adding an additional wavelength. 4. High flexibility, economy and reliability of networking When compared to the traditional networks using electrical TDM networks, new communication networks based on DWDM technology are greatly simplified in architecture and have clear network layers. Dispatching of various services can be implemented simply by adjusting the corresponding wavelengths of the optical signals. Because of the simple network architecture, clear layers and convenient service grooming, the flexibility, economy and reliability of networking are obvious.
1-10
DWDM Principle
5. Compatibility with all optical switching It is foreseeable that, in the realizable all optical networks in the future, processing such as add/drop and connection of all telecommunication services is implemented by changing and adjusting the optical signal wavelengths. So DWDM technology is one of the key technologies to implement all optical networks. Moreover, DWDM systems can be compatible with future all optical networks. It is possible to implement transparent and highly survivable all optical networks based on the existing DWDM system.
1-11
DWDM Principle
Objectives: To master basic structures and types of optical fibers. To know basic characteristics of optical fibers.
Coating
Cladding
Core
n2
n1
2-12
DWDM Principle
n2 n1
n2 n(r)
Thickness of the core and refractive indexes of the core material and cladding material are critical to the properties of the fiber. Figure 2-2 shows three typical optical fibers. As can be seen from this figure, there are two typical refractive index distributions in the fiber core-cladding cross-section. One is that the refractive index radial distributions of the core and the cladding are uniform, and the change of refractive index at the core-cladding boundary is a step function. This fiber is called step-index fiber. The other one is that the refractive index of the core is not a constant. It gradually decreases as the radial coordinate of the core increases until it equals to the index of the cladding. Hence this fiber is called graded-index fiber. The common feature of this two fiber cross-section is that the refractive index of the core n1 is larger than that of the cladding n2. This is also a necessary condition for the optical signal to transmit in the fiber. For a step-index fiber, total internal reflection can occur at the core-cladding boundary and the light wave can propagate along the core. For a graded-index fiber, the continuous refraction occurs to the light wave in the core, forming a light ray similar to the sine-wave through the fiber axis and guiding the light wave to propagate along the core. The tracks of the two light rays are shown in Figure 2-2. With the difference of the diameter size of the core of step-index and graded-index fibers, the number of modes transmitted in the fiber is different. Hence, step-index fiber or gradedindex fiber can be classified into single mode fiber and multimode fiber according
2b 2a
2b 2a
2b 2a
n2 n1
2-13
DWDM Principle
to the number of transmission modes. This is also a classification method of optical fiber. The core diameter of a single mode fiber is very small and, generally, less than 10m, and the core diameter of a multimode fiber is relatively large and often equal to 50m. However, there is little difference between the profiles of these two types of fiber. The diameters of fibers with a plastic jacket are less than 1mm.
2-14
DWDM Principle
dispersion point is near 1510m and its effective area is up to 72 square m. Therefore, it can effectively overcome the nonlinear affects and is especially suitable for DWDM system applications based on 10Gbit/s.
2.3
DWDM Principle
2-16
DWDM Principle
G.652, the attenuation constant at 1310nm and 1550nm should be less than 0.5dB/km and 0.4dB/km, respectively.
Technical details: The following figure shows the dispersion characteristics of several types of fiber.
2-17
DWDM Principle
G.655 fiber with positive dispersion coefficient Dispersion coefficient (ps/nm km)
G.653 fiber
17
G.652 fiber
1310
1550
Wavelength
(nm)
2-18
DWDM Principle
1250nm in jumper cable shorter than 2m. For G.655 1480nm in 22m cable, 1470nm in primary
2.4
2-19
DWDM Principle
2-20
DWDM Principle
Objectives: To understand the requirements and solutions of DWDM optical resources. To understand DWDM optical amplification technology. To understand DWDM multiplexing and de-multiplexing technology.
3.1 Lasers
Laser, whose function is to generate laser, is an important component of DWDM system. At present, lasers used in DWDM system are semiconductor laser LD (Laser diode). The operating wavelengths of DWDM systems are relatively dense. Generally, the wavelength spacing is from several nanometer to sub-nanometer. Hence, the laser diode is required to operate in a standard wavelength and possess good stability. On the other hand, the non-electrical regeneration distance of DWDM systems is increased from 50~60km of single SDH system transmission to 500~600km. lasers of the DWDM system are required to adopt lasers more advanced in technology and excellent in performance in order to elongate the dispersion limited distance of the transmission system and overcome fiber nonlinear effects {such as stimulated Brillouin scattering (SBS), stimulated Raman scattering (SRS), self-phase modulation (SPM), cross-phase modulation (XPM), modulation instability and four-wave mixing (FWM)}. In summary, lasers of DWDM system have two major features: 1. Relatively large dispersion tolerance; 2. Standard and stable wavelength.
3-21
DWDM Principle
For DWDM system without optical line amplifier, direct modulated lasers can be considered to save the cost. 2. Indirect modulation Indirect modulation: This modulation method is also called external modulation, i.e. modulating the laser indirectly and adding an external modulator in its output path to modulate the lightwave. In fact, this modulator works as a switch, as shown in Figure 3-1.
3-22
DWDM Principle
Optical modulator
The constant laser is a highly stable source continuously emitting lightwave with fixed wavelength and power. It isn't affected by the electric modulation signal during emitting, so no modulating frequency chirp occurs and the line breadth of its optical spectrum keeps at minimum. According to the electric modulation signal, the optical modulator processes the highly stable light from the constant laser light in a way of either passing through or blocking. During the modulation process, the spectrum characteristics of the lightwave won't be affected. This guarantees the quality of the spectrum. Lasers adopting indirect modulation are relatively complex with high loss and cost, but its modulating frequency chirp is very low. It can be used in systems whose transmission rate is 2.5Gbit/s and transmission distance longer than 300km. Hence, in DWDM systems with optical line amplifiers, the lasers at the transmit end are generally indirectly modulated. Commonly used external modulators are photoelectric modulator, acoustooptic modulator and waveguide modulator. The basic operating principle of photoelectric modulator is crystal linear photoelectric effect. Photoelectric effect refers to the phenomenon that electric field causes the variation of the refractive index of a crystal. A crystal that is able to generate the photoelectric effect is called photoelectric crystal. Acoustooptic modulator is made by utilizing the acoustooptic effect of the dielectric. Acoustooptic effect refers to the phenomenon that the dielectric changes under the pressure of an acoustic wave when it propagates through the dielectric. This change causes the variation of the refractive index of the dielectric and affects the transmission characteristics of the lightwave.
3-23
DWDM Principle
Waveguide modulator is manufactured from titanium (Ti) diffused LiNbO2 substrate material on which waveguide is made via photoetching method. It has many advantages such as small in dimension, light in weight and facile for optical integration. According to the integration and separation conditions of the laser and the external modulator, external modulated lasers can be classified into two categories: integrated external modulated laser and separated external modulated laser. As a maturing technology, integrated external modulation becomes the development trend of DWDM lasers. The commonly used modulator is electroabsorption modulator which, small and compact and integrated with the laser, meets most application requirements in performances. Electroabsorption modulator, a kind of loss modulator, operates at the boundary wavelength of the material absorption region. When the modulator isn't biased, the wavelength from the laser is out of the absorption range of the modulator material. Thus the launched power of this wavelength is maximum and the modulator is turned on. When the modulator is biased, the boundary wavelength of the material absorption region shifts and the wavelength from the laser is within this region. Thus the launched power is minimum and the modulator is turned off, as shown in Figure 3-2.
Biased Unbiased
Absorption region Absorption region
1 0 2
is the absorption side wavelength of unbiased modulator is the absorption side wavelength of biased modulator 0 is the operating wavelength of the constant light source
1 2
Electroabsorption modulator can be manufactured by utilizing the same technical process as semiconductor laser. Therefore, it is easy to integrate the laser and the modulator, suitable for batch production. So its development speed is high. For example, InGaAsP optoelectronic integrated circuit monolithically integrates a
3-24
DWDM Principle
laser and an electroabsorption modulator on a single chip that is put on a thermoelectric cooler (TEC). This typical optoelectronic integrated circuit is called electroabsorption modulated laser (EML). It can support transmission of 2.5Gbit/s signal over 600km, far exceeding the transmission distance of directly modulated lasers. Its reliability is similar to that of standard DFB lasers with an average life span of 20 years. Separated external modulated laser generally uses constant output laser (CW) + LiNbO3 Mach-Zehnder external modulator. This modulator separates the light input into two equal signals that, respectively, enter the two branches. These two branches employ electrooptic material whose refractive index changes with the magnitude of the external electrical signal applied to it. Change of the refractive index of the optical branches will result in variation of the signal phases. Hence, when the signals from the two branches recombine at the output end, the combined optical signal is an interfering signal with varying intensity. Via this method, the information of the electrical signal is transferred onto the optical signal. Thus optical intensity modulation is implemented. The frequency chirp of separated external modulated laser can be zero. Moreover, its cost is relatively low when compared to electroabsorption modulated external laser.
specific wavelength with an adjustable range of 1.6nm. The chip temperature is adjusted by changing the drive current of the cooler and using a thermal resistance as feedback. Thus the chip temperature is stabilized and stays at a constant value.
3-25
DWDM Principle
According to the correspondent characteristics of wavelength and chip temperature, distributed feedback laser (DFB) controls its wavelength by controlling the temperature of the laser chip to achieve wavelength stability. For DFB laser, the wavelength-temperature coefficient is about 0.02nm/ central wavelength meets the requirement within the range of 15 -35 and its . This
temperature feedback control method completely depends on the chip temperature of the DFB laser. At present, MWQ-DFB laser technical process can guarantee that the wavelength deviation meets the requirements of DWDM system in the life span (20 years) of the laser. Except for the temperature, laser drive current can also affect the wavelength. The sensitivity is 0.008nm/mA, smaller than the affect of the temperature in one order. In some cases, its effect is negligible. Additionally, package temperature may also affect the device wavelength (e.g. temperature conduction brought by wires from the package to laser platform and inward radiation from the package shell will also affect the device wavelength). In a well-designed package, its effect can be controlled to minimum. The above methods can effectively solve the problem of short-term wavelength stability. However, they are incapable of dealing with long-term wavelength variation caused by factors such as laser aging. It is ideal to directly utilize a wavelength sensitive component for wavelength feedback control of the laser. The theory is shown in Figure 3-3. Standard wavelength control of this type of scheme and reference frequency disturbance wavelength control are promising and being developed.
Optical output LD
LD control circuit
3-26
DWDM Principle
Thinking: Why does the DWDM system set strict requirements to the wavelength stability?
Signal input
WDM
PD
EDF
Signal output
TAP
PD Optical detector
As shown in Figure 3-4, signal light and pump light from the pumping laser are combined via a DWDM multiplexer, then they are sent to the erbium-doped fiber
3-27
DWDM Principle
(EDF). The two pumping lasers form a two-stage pump. Excited by the pumping light, the EDF yields the amplification function. Therefore, the function of amplifying the optical signal is implemented. 1.Erbium-doped optical fiber (EDF) Erbium-doped optical fiber (EDF), doped with Er3+ of a given density, is the kernel of the optical fiber amplifier. To illustrate its amplification principle, we need to begin with the energy level diagram of Er3+. The outer-shell electrons of Er3+ have three-level structure (E1, E2 and E3 in Figure 3-5), where E1 is ground state, E2 is metastable state and E3 is high level, as shown in Figure 3-5.
E3 excited state Decay Pump light 1550nm signal light E2 metastable state
When high energy pumping lasers are used to excite the EDF, lots of bound electrons of the erbium ions are excited from the ground state to the high level E3. However, the high level is not stable and erbium ions are soon dropped to the metastable state E2 via a radiationless decay process (i.e. no photon is released). E2 level is an metastable energy band on which particles' survival span is relatively long. Particles excited by the pumping light gather on this level via nonradiative transition. Thus, population inversion distribution is implemented. When an optical signal of wavelength 1550nm passes through this erbium-doped fiber, particles in the metastable state are transited to the ground state via stimulated radiation and generate photons identical to the photons of the incident signal light. This greatly increases the quantity of the photons in the signal light, i.e. implementing the function of continuous amplifying the signal light transmitted in the EDF. 2. Optical coupler (WDM)
3-28
DWDM Principle
Optical coupler, as its name implies, has function of coupling. It couples the signal light and the pumping light and sends them into the erbium-doped fiber. It, also called optical multiplexer, usually employs optical fiber fusible cone multiplexer. 3. Optical isolator (ISO) Optical isolator (ISO), a kind of component utilizing Faraday magnetooptical effect, allows only unidirectional light transmission. Along the light path, the functions of the two isolators are as follows: The input isolator can block the backward ASE in the EDF, keep it from interfering the transmitters of the system and from generating larger noise when it is reflected at the input end and reenter the EDF. The output isolator prevents the amplified optical signal, when reflected at the output end, from reentering the EDF, consuming particles and affecting the amplification characteristics of the EDF. 4. Pumping laser (PUMP) Pumping laser, the energy source of EDFA, provides energy for amplifying the optical signal. Generally, it is a semiconductor laser with output wavelength of 980nm or 1480nm. When passing through the EDF, the pumping light pumps the erbium ions from low level to high level. Thus population inversion is formed. When the signal light passes through, the energy will be transferred to it. Hence, optical amplification is implemented. 5. Optical splitter (TAP) The optical splitter used in the EDFA is a one by two component. Its function is to tap off a small part of the optical signal for monitoring the optical power of the main channel. 6. Optical detector (PD) The PD is an optical power detector. Its function is to convert the received optical power into photocurrent via photoelectric conversion. Hence, it monitors the input and output optical power of the EDFA module.
DWDM Principle
amplifier is to boost the launched power and elongate transmission distance by enhancing the optical power injected into the fiber (generally above 10dBm). So in some documents, it is also named as power booster amplifier. Here, its noise characteristic requirement is not high. The major requirement is linear power amplification characteristic. Generally, booster amplifier works in the saturation range of gain or input power in order to enhance the conversion efficiency from pumping source power to optical signal power.
Repeating section
D W D M
equip ment
BA
D W D M
e q u ip m e n t
2. Line amplifier (LA) Line amplifier is located in the middle of the whole regenerator section, as shown in Figure 3-7. This is an application form to insert the EDFA into the optical fiber transmission link and amplify the signal directly. A regenerator section can be configured with multiple line amplifiers according to the demands. Line amplifier is mainly applied in long-haul communication or CATV distribution networks. Here, the EDFA is required to have high small-signal gain and low noise factor.
Repeating section
D W D M
equip ment
LA
D W D M
equip ment
3-30
DWDM Principle
3. Pre-amplifier (PA) Pre-amplifier is located at the end of the regenerator section but in front of the optical receiving equipment, as shown in Figure 3-8. The main function of this amplifier is to amplify the small signal attenuated along the link and enhance the receiving sensitivity of the optical receiver. Here the main problem is noise. The main noise in EDFA is amplified spontaneous emission (ASE). This noise makes the optoelectronic detector output three noise components, i.e. extra shot noise due to the increase of optical power, signal-ASE beat noise and ASE-ASE beat noise. By using a narrow-band optical filter (1nm bandwidth), most ASE-ASE beat noise can be filtered and extra shot noise can be reduced. But the signal-ASE beat noise can't be filtered. Despite of this, the noise characteristic of EDFA is greatly improved by adopting the optical filter. The pre-amplifier greatly improves the sensitivity of receivers employing direct detection. For example, the sensitivity of an EDFA receiver of 2.5Gbit/s can be up to -43.3dBm. An improvement of about 10dB is achieved when compared to the receivers employing direct detection without EDFA.
R e p e a t in g s e c tio n
D W D M
e q u ip m ent
PA
O p tic a l f ib e r c o n n e c t o r
Tricks:
3-31
DWDM Principle
BA, PA and LA differ from each other in that their locations in the DWDM network are different. However, the most important difference lies in their input optical power and gain: BA: relatively high input optical power and low gain; PA: relatively low optical power and low gain, similar to BA; LA: relatively low input optical power, similar to PA, but its gain larger than BA.
3-32
DWDM Principle
Technically, the range of 1525nm~1540nm in EDFA gain curve is called blue band area and the range of 1540nm ~1565nm is called red band area. Generally, red band area is preferential when the transmission capacity is less than 40Gbit/s.
3-33
DWDM Principle
Technical details: Performance comparison of EDFA gain unflatness and flatness is given in Figure 3-10.
2. EDFA gain-locking EDFA gain-locking is an important problem because the WDM system is a multiwavelength working system. When certain wavelengths are dropped, their energy will be transferred to those undropped signals due to gain competition. Thus the power of other wavelengths increases. At the receive end, abrupt increment of the electrical level is possible to cause error. In limiting case, if seven wavelengths of eight wavelengths are dropped, all the energy will concentrate to the one wavelength left and its power may be up to about 17dBm. This will result in strong nonlinear effects or receiving power overload of the receiver, and this will also cause lots of errors. There are many gain-locking technologies for EDFA. One typical method is to control the gain of pumping laser. The internal monitoring electric circuit of the EDFA controls the output of the pumping source by monitoring the input-output power ratio. When some signals of the input wavelengths are dropped, the input power will decrease and the output-input power ratio will increase. Via the
3-34
DWDM Principle
feedback circuit, the output power of the pumping source will be reduced in order to keep the gain (output/input) of the EDFA. Hence, the total output power of the EDFA is reduced and the output signal power is kept stable. The process is shown in Figure 3-11.
INPUT
OUTPUT
TAP
PIN
PUMP
TAP
PIN
Non-linear control
Another method is saturation wavelength. At the transmit end, except for the eight operating wavelengths, system sends another wavelength as saturation wavelength. In normal cases, the output power of this wavelength is very small. When some line signals are dropped, the output power of the saturation wavelength will automatically increase in order to compensate the energy of the lost wavelengths and maintain the output power and gain of the EDFA. When the multi-wavelength line signals are restored, the output power of the saturation wavelength will correspondingly decrease. This method directly controls the output of the saturation wavelength laser, so its speed is faster than controlling the pumping source.
3-35
DWDM Principle
Technical details:
falling wavelength
>1dB
Figure 3-12 NO Gain-locking when EDFA falling wavelength and adding wavelength
Falling wavelength
<0.5dB
adding wavelength
<0.5dB
Figure 3-13 Having Gain-locking when EDFA falling wavelength and adding wavelength
3-36
DWDM Principle
1. Non-linearity problem Although enhanced by adopting EDFA, the optical power is not the higher the better. When it reaches a certain level, the optical fiber will generate nonlinear effects (including Raman scattering and Brillouin scattering). Especially, EDFA has greater affect to stimulated Brillouin scattering (SBS). Nonlinear effects greatly limits the amplification performance of the EDFA and the implementation of long distance repeaterless transmission. 2. Optical surge problem EDFA can enhance the input optical power rapidly. However, since its dynamic gain variation is slow, optical surge will occur at the moment when the input signal power jumps, i.e. a peak occurs to the output optical power. The optical surge phenomenon is especially obvious in the case of EDFA cascading. The peak power can be up to a few watts and is possible to damage the O/E converter and the end surface of the optical connector. 3. Dispersion problem Although the problem of attenuation limited repeaterless long haul transmission is solved after adopting EDFA, the total dispersion increases as the distance becomes longer. Thus the former attenuation limited system turns into dispersion limited system.
3-37
DWDM Principle
1 2
WDM
(a) Multiplexer
1 2
WDM
(b) De-multiplexer
There are many methods to manufacture DWDM components each of which has its own features. At present, there are four types of widespread commercial DWDM components: interference light filter type, optical fiber coupler type, optical grating type and arrayed waveguide grating (AWG) type.
3-38
DWDM Principle
output (in)
Diffraction grating
1 2
n input (out)
The blazed grating type filter has excellent wavelength selectivity and can reduce the wavelength spacing to about 0.5nm. Moreover, the grating type component is parallel operated and its insertion loss doesn't increase with the number of wavelengths multiplexed. Hence large number of multiplexing channels can be achieved. At present, multiplexing of 131 wavelengths with a spacing of 0.5nm is implemented and the isolation is good. For a wavelength spacing of 1nm, the isolation is up to 5dB. The disadvantage of blazed grating is relatively large insertion loss, generally 3~8dB. Moreover, it is very sensitive to polarization and its optical channel bandwidth-to-spacing ratio isn't ideal. So the optical spectrum utilization ratio isn't high enough. And the wavelength fault-tolerance requirement for the laser and DWDM component is relatively high. Additionally, its temperature drift varies with the thermal expansion coefficient and refractive index of the material. Typically, the component temperature shift is relatively high, approximately 0.012nm / . If temperature control measures are adopted, the . So temperature control temperature shift can be reduced to 0.0004nm / measures are feasible and necessary. This optical grating requires high manufacturing accuracy and is not suitable for mass production. Hence, it is generally applied in experimental scientific research. Except for the above-mentioned optical fiber component, the manufacturing technology for optical fiber Bragg grating filter is gradually maturing. It is manufactured employing the interference of high power ultraviolet light beams to form periodic variation of refractive index at the optical fiber core. The accuracy can be up to 10000 lines per centimeter, as shown in Figure 3-16. Fiber Bragg grating can be feasibly designed and manufactured with low cost. It has very low
3-39
DWDM Principle
insertion loss and stable temperature characteristic. Its intraband filtering characteristic is flat and out-of-band is very steep ( rolling slope is better than 150dB/nm and out-of-band suppress ratio is up to 50dB). This component can be directly melted with the optical fiber of the system. So it can be fabricated into band-pass or band-stop filter with small channel spacing. At present, it is extensively applied in DWDM system. However this kind of optical fiber grating has relatively narrow wavelength range, only applicable to single wavelength. The benefit it brings in is that the filters can be added or removed according to the number of wavelength used. So the application is flexible.
DWDM Principle
interference type filter can be made to band pass certain wavelength range and band stop the other wavelength range, forming the required filter characteristics. The structural principle of the film interference type filter is shown in Figure 3-17. The main features of dielectric film filter DWDM component are as follows: miniaturization and structural stability of the component can be implemented via design, the signal pass-band is flat and polarization-independent, and its insertion loss is low and channel isolation is good. The disadvantage is that the number of channels can't be large. The specific characteristics are related to its structure. For instance, if the film filter type DWDM component utilizes soft material, its wavelength may be changed under the environmental influence because the filter can easily absorb moisture. When employing hard dielectric film material, the temperature stability is better than 0.0005nm/ . Additionally, this kind of component has relatively long design and manufacturing process and low volume of production. And if epoxy resin is used along the light path, it is not easy to achieve high isolation and narrow bandwidth. In DWDM systems, when only 4 to 16 wavelengths are involved, this type of DWDM component is relatively ideal.
1-4
1 filter
3 4 Glass
3-41
DWDM Principle
form a cone and slightly twisting them to fuse them together. Because the cores of different fibers are extremely close, the required coupling power can be obtained via evanescent wave coupling on the conical region. The second type of coupler employs grinding and polishing methods to remove part of cladding of the optical fiber so that only a thin cladding layer is left. Then two optical fibers processed via the same method are butt jointed and coated a layer of index matched solution between them. Thus the two fibers can couple via the evanescent wave in the cladding and obtain the demanded coupling power. Fused conical type DWDM component is simple to manufacture and is extensively applied.
Waveguide grating
1 2
3-42
DWDM Principle
Table 3-1 Comparison of various DWDM components Componen t type Refractive grating type Dielectric film type Fusible cone type Integrated optical waveguide type Mechanis m Angular diffraction Interferenc e/absorptio n Wavelengt h dependent Plane waveguide Mass producti on Average Channel spacing (nm) 0.5~10 Number of channels 131 Crosstalk (dB) Insertion loss (dB) 3~6 Main disadvant ages Temperat ure sensitive Small number of channels Small number of channels Large insertion loss
-30
Average
1~100
2~32
-25
2~6
10~100
2~6
10~45)
0.2~1.5
1~5
4~32
-25
6~11
3-43
DWDM Networking
Objectives: To understand the basic concepts of DWDM networking. To master the configuration of different network elements of DWDM. To master some factors to consider during DWDM network design. To know general protection mechanisms of DWDM networks.
1~16 (32)into a DWDM main optical channel via the multiplexer, amplifies the
optical power of the main channel, and then adds an optical supervisory channel
.
At the receiver end, firstly OTM extracts the optical supervisory channel . Then the DWDM main optical channel is amplified and de-multiplexed into STM-16 signals of the 16/32 different wavelengths. The signal flow of OTM is shown in Figure 4-1.
4-44
DWDM Networking
RI
A S
RO
W P A
M M
D 16 D 32
R W C S D H A
RM
TM
S C 1 A W B A M M
16 M 32
TO
TI
T W C
RI
ROA
W P A
A M W B A
TI
TO
S M
RM1 TM1
S C 2 A W B A West M W P A
TM2
RM2
TO
TI
RO
RI
East
4-45
DWDM Networking
The whole equipment is installed in a subrack. In the figure, each direction employs a pair of WPA+WBA to conduct optical line amplification. It can also use single WLA or WBA to conduct unidirectional optical line amplification.
4-46
DWDM Networking
RM1
SC2
RI TM1
TM2 RM2
SCC R R W W C C 1 2
OHP T T W W C C 1 2
TO
TO
S C A
WPA
WBA TI RO
RO TI WBA
MR2
S C A
RI
MR2
1 2 T T W W C C
1 2 R R W W C C
WPA
2. OADM equipment consists of two back-to-back OTMs Two back-to-back OTMs are used to form an OADM equipment which can add/drop wavelengths. This mode is more flexible when compared to the static OADM which uses a board to conduct wavelength conversion. It can add/drop any of wavelengths from 1 to 16/32, more feasible for networking. If a signal channel isn't added/dropped at this station, it can directly access the TWC of the same wavelength via the D16/D32 output port and then enter the M16/M32 board in the other direction. The signal flow of the OADM consisting of two back-to-back OTMs is shown in Figure 4-4.
RI
A RO S
W P A
M M
TM RM
D 16
D 32
R W C S D H A S D H
T W C
M 16
M 32
A
TI TO
W B A M
TM
RM
TM
S C 2 /1 W B A
C
RM
TO
TI
A M 16 M
M 32
T W C west
R W C east
W D P 16 M A
D 32
A A
RO
RI
4-47
DWDM Networking
Figure 4-4 The signal flow of the OADM consisting of two back-to-back OTMs
A RO S
W P A
M M
TM RM
D 16
D 32
T W C M
M 16
M 32
TI
TO
W B A M
TM RM
RM
TM
S C 2 /1 W B A
TO
TI
A M 16 M
M 32
A T W C west east W D P 16 M A
D 32
A A
RO
RI
Figure 4-5 The signal flow of the electrical regeneration equipment REG
Note:
4-48
DWDM Networking
Other basic network unit types of DWDM equipment are included in these types, with similar functions and same position in the network. They only differ in names. The following contents related to the network element or board are described by using OptiX BWS 320G equipment configuration and board.
SDH
OTM
SDH
OTM
4-49
DWDM Networking
SDH
OTM
OADM
SDH
SDH
OTM
4-50
DWDM Networking
1~8
OADM
1~8
OADM
OADM
1~8
OADM
1~8
4-51
DWDM Networking
NMS NE NE
Management information
GNE
Management information
NE NE
Figure 4-9 Network management information channel backup in ring networking (when certain transmission section fails)
However, when both of the two ends of certain office in a fiber section fail or certain transmission section in point-to-point and chain networking fails, network management information channel will fail. Consequently, network management administrators won't be able to get the supervisory information of failed office and do some operation with it.. To avoid this circumstance, network management information should use the backup channel. The SDH network elements can provide backup network management information channel by using data communication network. Between two network elements in need of protection, a network management information backup channel can be established by accessing the data communication network via routers. When the network is normal, network management information is transmitted by the main supervisory channel, as shown in Figure 4-10.
4-52
DWDM Networking
NMS
Router
Router
Figure 4-10 Network management information channel backup (in normal case)
When the main channel fails, network elements automatically switch the management information to the backup channel to guarantee that the network management system can supervise and operate the entire network. The whole switching process is conducted automatically, not requiring manual intervention. A network management information channel backup is illustrated in Figure 4-11.
NMS
Router
Router
4-53
DWDM Networking
Figure 4-11 Network management information channel backup (in case of main channel failure)
Attention: In network planning, different routes should be selected respectively for backup supervisory channel and main channel. Otherwise backup function won't be implemented. HUAWEI OptiX BWS 320G equipment provides various data interfaces (e.g. RS232 and Ethernet interface) for management information channel interconnection among different DWDM networks and between DWDM and SDH. This implements unified network management for different transmission equipment. Figure 4-12 illustrates management information channel interconnection among different transmission equipment.
NMS
4-54
DWDM Networking
4-55
DWDM Networking
regenerator system, will aggravate the impairment caused by the dispersion in the system. In some subsystems of optical amplifier, a type of passive dispersion compensation device can be assembled with the optical amplifier to form an amplifier subsystem. This subsystem will add limited dispersion to the system. And the dispersion coefficient, inverse to the optical fibers of the system, will reduce the system dispersion. This device can be installed together with an EDFA to compensate the loss related to the passive dispersion compensation function. Additionally, adopting G.655 optical fiber and G.653 optical fiber is favorable for dispersion reduction. If nonlinear impairments are considered thoroughly, G.655 optical fiber has optimal over-all properties in long haul transmission. 4. Consideration in network design In DWDM network design, firstly the whole network is divided into several regenerator sections, letting the length of each section less than the dispersion limited distance of the laser. Hence, the performance of the whole network can tolerate the effect of dispersion.
Tricks:
When we calculate dispersion during DWDM network design, the typical dispersion coefficient at 1550nm window is 17ps/nm.km because optical fibers employed in the world are primarily G.652 fiber. But in engineering design, 20ps/nm.km is adopted for budget.
4.3.2 Power
Long distance transmission of optical signal requires that the signal power is enough to compensate the attenuation of the optical fiber. Generally, the attenuation coefficient of G.652 optical fiber at 1550nm window is about 0.25dB/km. When factors such as optical connectors and optical fiber redundancy are taken into consideration, the combined optical fiber attenuation coefficient is generally less than 0.275dB/km. During practical calculation, power budget is only conducted for two pieces of adjacent equipment in the transmission network instead of conducting unified power budget for the whole network. The distance (attenuation) between two
4-56
DWDM Networking
pieces of adjacent equipment in the transmission network is called regeneration distance (attenuation).
S Station A
R L
Pout
P in
Station B
As shown in above figure, S is the transmit reference point of station A, R is the receive reference point of station B and L is the transmission distance between point S and point R. Then: regeneration distance = Pout - Pin) /a
Pout: channel output power of point S (in dBm). The optical power of point S is related to the configuration of point A. Pin: permissive channel minimum input power of point R (in dBm). a: optical fiber cable attenuation per kilometer (dB/km) (using 0.275dB/km according to ITU-T recommendations. It contains the effect of various factors, including connectors and redundancy).
DWDM Networking
caused by gain saturation effect. Similarly, the effective gain distribution of the third optical amplifier will also change. This effect will be transmitted downstream to the next optical amplifier. Even though narrow band filter is used in each optical amplifier, ASE noise will also accumulate. This is because the noise exists in the signal frequency band. Optical signal-to-noise ratio (OSNR) is defined as: OSNR = channel optical signal power / channel noise optical power 2. Transmission limitation ASE noise accumulation affects the system SNR because SNR degradation of the received signal is mainly caused by ASE related beat noise. This type of beat noise linearly increases with the number of optical amplifiers. Hence, error rate degrades as the number of optical amplifiers increases. Besides, the noise is accumulated in exponential form to the gain amplitude of the amplifier. As the result of optical amplifier gain, the ASE noise frequency spectrum will have a wavelength peak after accumulating in many optical amplifiers. To specially point out, when adopting closed all-optical ring, the ASE noise will infinitely accumulate if cascading infinite number of amplifiers. Although in systems with filters ASE accumulation is remarkably decreased due to the filters, intraband ASE will still increase as the number of optical amplifiers increases. Hence, SNR will degrade with the increase of amplifiers. 3. Methods of ASE reduction ASE noise accumulation may decrease as the interval of optical amplifiers reduces (when the total gain is equal to the total transmission channel attenuation) because ASE accumulates in exponential form with the increase of the gain amplitude of the amplifier. One of the following filter technologies can further reduce unexpected ASE noise, i.e. adopting ASE noise filter or utilizing self-filtering effect (self-filtering method). Self-filtering method is suitable for systems with tens of or more optical amplifiers. This method adjusts the signal wavelength to the self-filtering wavelength in order to reduce the ASE noise received by the detector, similar to using a narrow band filter. This is most effective when the approach of reducing optical amplifier interval and employing low gain optical amplifier is used to reduce the initial ASE noise. If all-optical DWDM closed ring network is adopted, the self-filtering method isn't suitable. In fact, the peak formed in the whole gain frequency spectrum of the optical amplifier may severely affect system performance. In this case, utilizing
4-58
DWDM Networking
ASE filtering method can utmost reduce ASE noise accumulation. This is achieved via the approach of filtering the DWDM channels which are not sent to the network node before being switched out the node. For systems with a few optical amplifiers, self-filtering method is not as effective as ASE filtering method. ASE filtering method can flexibly select signal wavelengths and has other advantages. Characteristics of the filter must be carefully selected because cascading filter has narrower pass-band than the signal filter (unless it has a rectangular frequency band). 4. Considerations for OSNR in DWDM network design (Note: If you feel that the contents of this part is a bit abstruse, it is OK to skip it and read next chapter) For different network applications, OSNR requirements are almost the same, with slight differences as shown in Table 4-1.
4-59
DWDM Networking
Table 4-1 OSNR comparison Amplifier cascading type 32-channel 822dB system (8*80km) 32-channel 530dB system (5*100km) 32-channel 333dB system (3*120km) Minimum OSNR (dB) 22 20 20
OSNR is one of the most important factors that affect the DWDM system error performance. For a DWDM system with multiple cascading optical line amplifiers, the noise power is dominated by amplified spontaneous emission (ASE) noise. 1)ASE noise accumulation of cascading optical line amplifier The mathematical model of the ASE noise accumulation in multiple cascading optical line amplifiers is illustrated in Figure 4-14.
G1 L1
G2 L2
GN-1 LN-1
GN
EDFA1
EDFA2
EDFAN-1
EDFAN
In Figure 4-14, GN is the gain of EDFAN (in linear unit); LN is the optical fiber cable attenuation of the regenerator section N (in linear unit). The total ASE noise power = The ASE noise power generated by EDFAN + (The ASE noise power generated by EDFAN-1LN-1 GN) + + (The ASE noise power generated by EDFA2L2 G3 N-1 LN-1 GN) + (The ASE noise power generated by EDFA1L1 G2 N-1 LN-1 GN) ( Equation 4-1) 2)Noise of a single EDFA
4-60
DWDM Networking
The ASE noise power per unit frequency band generated by a single optical amplifier, PASE is PASE=2NSP G-1) h (4-2) where NSP is the spontaneous noise coefficient of the EDFA; G is the internal gain of the EDFA; h is the Planck constant;
This calculation method can meet the requirements of the general engineering design. However, the following conditions must also be satisfied except for the above-mentioned assumptions.
4-61
DWDM Networking
4-62
DWDM Networking
Input power
3) Methods of reducing the effect In a system whose laser line breadth is apparently larger than the Brillouin bandwidth or whose signal power smaller than the threshold power, SBS impairment won't occur. 2. Stimulated Raman Scattering (SRS) 1) Principle SRS is a broadband effect related to the interaction of light with silicon atom vibration modes. SRS makes the signal wavelength works as a Raman pump of the channels of longer wavelength or the Raman-shifted light of spontaneous scattering. In any circumstance, the signals of shorter wavelength will always be weakened by this process. At the same time, the signals of longer wavelength will be enhanced. 2) Transmission limitation SRS may occur in both single wavelength systems and multi-wavelength systems. In systems with a single wavelength and no line amplifier, the signals may be impaired by this effect when its power is greater than 1W. However, in multiwavelength systems of relatively wider channel spacing, the channels of shorter wavelength will lose a portion of power to the higher-wavelength channels due to
4-63
DWDM Networking
the effect of SRS, leading to a degradation of the signal-to-noise ratio performance. This may limit the total capacity of systems with fixed total number of channels, channel spacing, mean launched optical power and total system length. The SRS dispersion threshold of systems adopting G.653 optical fiber is slightly lower than that of systems employing G.652 fiber because G.653 fiber has smaller equivalent core area. SRS won't cause practical degradation effect for single wavelength systems. However, it may limit the capacity of DWDM systems. 3) Method of reducing the effect In single wavelength systems, optical filters can be used to filter the unwanted frequency components. However, up to now there are no practical technologies for multi-wavelength systems to eliminate the effect of SRS. The effect of SRS effects can also be released by reducing the signal power. Nevertheless, no apparent SRS limitation has appeared in the carefully-designed DWDM systems implemented at the present time. 3. Self-phase Modulation (SPM) 1) Principle Because of the Kerr effect, instantaneous variations in the power of an optical signal result in self modulation. This effect is called self-phase modulation. In single wavelength systems, SPM effect will broaden the signal's spectrum when changes in the signal's intensity result in variations in its phase, as shown in Figure 4-16. In the normal dispersion zone of optical fiber, signals propagating along the fiber will experience a longer instantaneous widen once the frequency spectral broadening is caused by SPM due to the dispersion. In the abnormal dispersion zone, the dispersion of optical fiber and SPM may compensate with each other. Thus the signal broadening will be smaller.
4-64
DWDM Networking
Figure 4-16 Compression and spectrum broadening of the transmission pulse caused by selfphase modulation
2) Transmission limitation Generally, SPM is relatively apparent only in systems with high accumulated dispersion or ultra long length. Dispersion limited systems may be unable to tolerate SPM effects. In multi-wavelength systems of narrow channel spacing, spectral broadening caused by SPM may lead to interference between adjacent channels. In G.652 optical fiber, SPM of the low chirp intensity modulated signal leads to compression of the pulse. For G.655 optical fiber of abnormal dispersion characteristic, the SPM effect of the signal is a function of the transmitter power. Pulse compression can suppress the dispersion and provide certain dispersion compensation. However, the maximal dispersion limitation and the corresponding transmission distance limitation still exist. Figure 4-16 illustrates compression of the transmission pulse caused by SPM of the low chirp intensity modulated signal in G.652 optical fiber, also it can be regarded as spectral broadening.
4-65
DWDM Networking
3) Method of reducing the effect To adopt G.653 optical fiber and configure the signal channels near the zerodispersion zone will benefit the reduction of SPM effects. For systems employing G.652 optical fiber and less than 100km in length, the effects of SPM can be controlled by using dispersion compensation in appropriate intervals. The SPM effects can also be weakened by reducing the input optical power or configuring the system operating wavelengths over the zero dispersion wavelength of G.655 fiber. 4. Cross-phase Modulation (XPM) 1) Principle In multi-wavelength systems, when variations in light intensity lead to a phase shift, cross-phase modulation will generally broaden the signal spectrum due to the interaction between adjacent channels. The spectral broadening caused by XPM is related to the channel spacing. This is because the dispersion caused by difference in group velocities may lead to the interaction among the pulses which should separately propagate along the optical fiber. In case that XPM results in spectral broadening, the signals will suffer a relatively large instantaneous spectral broadening due to the dispersion effect when propagating along the optical fiber. 2) Transmission limitations Impairment caused by XPM in G.652 fiber-optic systems is more obvious than that in G.653 and G.655 fiber-optic systems. The broadening, caused by XPM, leads to interference between adjacent channels in multi-wavelength systems. 3) Methods of reducing the effect XPM can be controlled by selecting appropriate channel spacing. Study shows that the signal distortion caused by XPM in multi-wavelength systems only occurs between adjacent channels. In a 3-channel system, the signal-to-noise ratio (SNR) of the central channel is nearly equal to that of single channel systems. This is because the channel spacing has increased. Hence, the effect of XPM is negligible if signal channels have appropriate spacing. In simulation experiments for systems with a channel power consumption of 5mw, it is approved that a channel spacing of 100GHz is enough for reducing the effects of XPM. The dispersion penalty caused by XPM can be controlled by adopting dispersion compensation in proper intervals along the system. 5. Four-wave Mixing (FWM) 1) Principle
4-66
DWDM Networking
Four-wave mixing (FWM), also called four-phonon mixing, occurs in the case that two or three lightwaves with different wavelength interact and cause new lightwaves at other wavelengths. These extra wavelengths are so called mixing products or sidebands. This interaction may occur among signals in multiwavelength systems and EDFA ASE noises and between main modes and side modes. In case of 3 signals, the mixing products are shown in Figure 4-17.
f113
F132,312 f221
F1
F2
F3
When channel spacing is equal, these products will right enter the adjacent signal channels. If the phase matched condition is reached between the sideband and the initial signal, these two lightwaves propagating along the optical fiber will generate highly efficient FWM. 2) Transmission limitation Occurrence of FWM sidebands may cause remarkable reduction to the signal power. Even more severely, residual interference occurs when the mixing products directly enter the signal channels. This kind of interference is determined by the interaction between the phases of the signals and the sidebands and indicated by the increase and decrease of the signal pulse amplitude. Residual loss leads to closure of the eye pattern of the receiver and causes bit-error rate (BER) performance degradation. The effect of FWM can be reduced by the breakdown function of frequency spacing and dispersion and the phase matching among lightwaves. G.652 fiber-optic systems suffer less impairment of FWM than those adopting G.653. On the contrary, if a signal is right located at or near the zero-dispersion point, FWM may surge in a relatively short fiber length (i.e. tens of kilometer). Moreover, FWM is sensitive to the channel spacing.
4-67
DWDM Networking
Four-wave mixing may cause severe damage to multi-wavelength systems adopting ITU-T G.652 optical fiber because signals can merely tolerate a very small dispersion. In single channel systems, FWM interaction may occur between signals and ASE noises, as well as the main modes and sidemodes of the transmitters. The ASE phase noises accumulated by optical Kerr effect are superimposed to the signal carriers and cause the broadening of the rear part of the signal spectrum. 3) Method of reducing the effect As mentioned above, FWM band can be suppressed by utilizing the fiber dispersion such as G.655. FWM damage can be released by arranging uneven channel spacing. To lower the power level of G.653 fiber-optic systems can permit multi-wavelength operation, but this will weaken the advantages of the optical amplifiers. To properly suppress the generation of mixing products, a scheme has been proposed (an existing recommendation or new recommendation for future study) to adopt the optical fiber with a minimum permissible dispersion (non-zero dispersion) in the amplification bandwidth of EDFA. It is also a possible scheme to use the non-zero dispersion optical fiber of inverse dispersion characteristic as replacement section. However, this replacement may encounter difficulties during installation, operation and maintenance because of the introduction of another kind of fiber. Some similar methods are discovered to adopt long fiber sections of limited dispersion and short fiber sections of inverse but relatively greater dispersion (for compensation). A scheme has been proposed to adopt uneven and relatively large channel spacing to reduce the nonlinear effects and allow to arrange DWDM systems in G.653 fiber to reduce the effect of FWM. To use uneven channel spacing can guarantee that the mixing products caused by three or more channels won't fall into the wavelengths of other channels. However, the power transfer from the signals to the mixing products (i.e. power loss of the signals) keep fixed due to the configuration of uneven channel spacing, and will still lead to remarkable closure phenomenon of the eye pattern. Increase of the channel spacing can also reduce the effect of FWM. This kind of remission technology may be restricted since the gain spectrum will be narrowed due to the cascading of optical amplifiers and the amplification spectrum will be narrowed due to the access of optical amplifiers. Thinking: At 1550nm window and among three types of optical fiber: G.652, G.655 and G.653, which fiber has the most severe FWM effect? why?
4-68
DWDM Networking
6. Polarization Mode Dispersion (PMD) 1) Principle As we know, the fundamental mode in a circular symmetric dielectric waveguide is dual-degenerate. In a physical optical fiber, this degeneration is separated by birefringence. For polarization-maintaining fibers, birefringence is deliberately introduced. However, for general communication optical fibers, birefringence is an unexpected product which is randomly introduced due to the stress perturbation the fiber suffers. For birefringent optical fibers, the first term generates a group delay time called polarization dispersion. This kind of polarization dispersion leads to a group delay difference between the orthogonal polarization states, as shown in Figure 4-18.
Slow in propagation
Figure 4-18 Occurrence of group delay between the orthogonal polarization states
Although PMD effect randomly changes the polarization state of pulses propagating in optical fiber, a pair of orthogonal states or primary states can be determined, i.e. the signal incident to the fiber at the input end keeps its polarization state at the output end. For the first term, these states are independent to the wavelength. However, in some cases the occurrence of the primary states may be related to the wavelength. This, together with the dispersion of the optical fiber, will lead to further degradation. Birefringence of optical fiber is randomly introduced due to factors such as stress, bending, twisting and temperature. Random birefringence mechanism redetermines the local birefringence axis along the optical fiber and leads to the coupling between polarization modes. The fiber length between this change is called coupling length. The coupling length of an optical fiber refers to the sum of the average value of total local coupling length. 2) Transmission limitation
4-69
DWDM Networking
In digital transmission systems, PMD leads to intersymbol interference. When the total dispersion is equal to 0.4T (T is the bit period), an optical power penalty of about 1dB is introduced. Current study shows that, optical fibers or optical fiber cables are apt to be standardized according to mean PMD, as well as digital transmission systems. It is predicted via computer simulation that the probability for the optical power penalty of a system to exceed 1dB is less than 10-9 if the mean PMD isn't greater than 0.1T. In a long distance amplified systems employing polarization scrambler (a component which deliberately modulates the polarization state of the laser and makes it work in an unpolarized state), PMD leads to the increase of signal polarization. The interaction between polarization dependent loss and polarization hole-burning causes the degradation of system performances. When additional polarization dependent loss occurs in the system, greater secondary loss will be aroused. The secondary effect may generate coupling between PMD and dispersion and increase the statistical component of the dispersion. This field is under study. 3) Method of reducing the effect Since the problem is caused by birefringence, all the efforts for reducing the effects of PMD are related to reducing the birefringence introduced during optical fiber cable manufacturing, such as optimizing optical fiber manufacturing, guaranteeing the concentration of optical fiber, reducing the residual of fiber core and employing accurate cable structure. Typical mean PMD of optical fiber cables is in the following range:
0<( ) < 0.5ps/ km
Another method is to add polarization controllers at the input end and the output end. A polarization splitter is connected after the output polarization controller and used to generate an error signal. The output polarization controller searches this error signal and readjusts the polarization controller to minimize the error signal. At the zero-error signal point, the input polarization state is the primary state of the system. This technology has been used to compensate a 5Gbit/s system. Coherent frequency division multiplexing systems also adopt similar technology. 7. Polarization Dependent Loss (PDL) 1) Principle Polarization dependent loss is caused by dichromatism of optical passive components such as isolator and coupler. When a signal passes through a dichromatic component, its electric field part parallel to the loss axis will suffer
4-70
DWDM Networking
certain attenuation. Like PMD, the axis direction which determines the PDL is randomly changed. 2) Transmission limitation In amplified systems, the amplifiers operate in the power conservative mode. The signal and the noise are affected by PDL. However, the signal and the noise suffer different effects because the noise is unpolarized. The noise can be divided into a component parallel to the signal and another one orthogonal to the signal. Optical amplification may increase the component orthogonal to the signal. Additionally, variations of the signal polarization lead to mode dispersion. Thus, magnitude of the orthogonal component of the noise is time-varied. This will reduce the signalto-noise ratio at the receiver end and cause impairment to the system. 3) Methods of reducing the effects For PMD, it is important to reduce the polarization mode dependent loss of the components. To be pointed out, the effect of polarization mode dependent loss increases with the number of the amplifiers. For example, this requirement is extremely strict in long distance submarine systems. In short distance systems with only several amplifiers, the effect of polarization mode dependent loss is for further study. 8. Polarization Hole-burning (PHB) 1) Principle Polarization hole-burning (PHB) is the result of the anisotropic saturation caused by the polarization saturated signal light incident in the erbium-doped optical fiber. This will reduce the options of stimulated states utilizing the polarization field to locate. Hence, the available gain in the orthogonal direction is relatively large. Although erbium ions are randomly distributed in the glass fiber rod material, dipoles related to the erbium ions are anisotropic in the micro level. When the linear polarization saturated signal is equidirectional to the primary axis of the dipoles, the polarization hole-burning has the greatest effect. However, when the polarization state of the saturated signal is elliptical or circular, its effect decreases. Because the total differential gain is the vector sum of these two effects, both the signal laser and the pumping laser will affect the total effect. The degree of holeburning is in direct proportion to the polarization. Unpolarized saturated signals have no hole-burning problem. This case, as a whole, is similar to the case of circular polarization signal. 2) Transmission limitation
4-71
DWDM Networking
Because it makes the noise formed along the link larger than the noise budget calculated according to the simple linear theory, PHB will affect the performances of the system. The effects are that the signal-to-noise ratio decreases due to the PHB and that the ultimately measured Q value fluctuates under PMD and PDL. Since there are two factors affecting PHB, there are two ways to affect the system performances. The total effect is in direct proportion to the gain saturation and increases with the saturation. Firstly, we consider the effect of the polarized pumping laser. To reach the purpose of the discussion, the pumping polarization can be assumed as fixed. Pumping causes differential gain in the orthogonal polarization axis direction. The noise orthogonal to the pump is greater than the noise equidirectional to the pump. However, the polarization axes of the pumping lasers of the amplifiers along the link are incoherent to each other. The accumulation effect shall be similar to a random walk. The pump which results in PHB can be regarded as a related factor to the PDL of an amplifier. Hence, the noise obtained by averaging the number of the amplifiers should be linear, the same as the budget calculated by the simple linear theory. Signal lasers which cause PHB are slightly different. The lasers are used to propagate signals, so the polarization noise equidirectional to a signal laser obtains the same gain as the signal. However, the noise orthogonal to the signal laser is always orthogonal to the polarization axis of the signal. Hence, the signal increases in a nonlinear mode along the amplified link. The total differential gain caused by PHB will change with the variations of the signal polarization state along the amplified link (caused by PMD). It changes because the hole-burning effect of the signal is related to the pump effect. When staying in their corresponding polarization states, the signal laser and the pumping laser will change the amplitude of the differential gain variation. Hence, although this makes the total noise increase in a nonlinear form, the noise may be timevaried. As mentioned above, the signal-to-noise ratio will decrease and be timevaried. 3) Method of reducing the effect There are several methods for reducing the effect of PHB. It is a feasible method to amplify in the small-signal area, but it is not always possible. In many cases, it can meet the demands. Actually, the simplest method is to adopt unpolarized signals which can be generated via many approaches. The most common approach is to adopt polarization scramble to generate signal. If a phase modulator is used, the polarization state will change between the two orthogonal states at all time. Thus, the signal seems to have no polarization.
4-72
DWDM Networking
This indicates that it's better to arrange the polarization modulation according to a double bit rate because the PDL in the amplifier will be converted from polarization modulation to amplitude modulation. By adopting double bit rate polarization modulation, the amplitude fluctuation stays at the rate above the bandwidth of the detector and is not sensible to the receiver. If this technology is used, the performances of very long distance systems will be improved and reach the expected purpose of high reliability. Polarization modulation has become the standard implementation method for overseas large systems. However, in long distance amplified systems, PMD will result in secondary polarization and cause PHB which leads to the performance degradation of the systems. This effect improves the complex properties of the interaction of polarization effects in amplified links.
MUX
DMUX
MUX
4-73
DWDM Networking
Figure 4-19 1+1 protection based single wavelength and implemented on the SDH layer
This protection system mechanism is similar to the 1+1 MSP of SDH system. All the system equipment needs backup, such as SDH terminal, multiplexer/demultiplexer, optical line amplifier and optical fiber cable line. The SDH signals are permanently bridged to the working system and the protection system. At the receiver end, the status of the SDH signals received by the two DWDM systems are monitored and the more appropriate signal is chosen. This method has high reliability, but its cost is relatively high. In a DWDM system, switching of each SDH channel isn't related to other channels, i.e. when Tx1 of DWDM system 1 fails and switches to DWDM system 2, Tx2 can still operate in DWDM system 1. Once a switching condition is detected, the protection switching must be completed within 50ms. 2. 1:n protection based on single wavelength and implemented on SDH layer DWDM system can implement 1:N protection based on single wavelength and implemented on SDH layer. As shown in Figure 4-20, Tx11, Tx21 and Txn1 share a protection section and, together with Txp1, form a 1: n protection relationship. Tx12, Tx22 and Txn2 share a protection section and, together with Txp2, form a 1: n protection relationship. And so on, Tx1m, Tx2m and Txnm share a protection section and, together with Txpm, form a 1: n protection relationship. The SDH multiplex section protection (MSP) monitors and measures the status of received signals and conducts bridge and selection for the appropriate SDH signals from the protection section.
4-74
DWDM Networking
Tx11 Tx12 Tx1m Tx21 Tx22 Tx2m Txn1 Txn2 Txnm Txp1 Txp2 Txpm
N o te : H e re th e
MUX
DMUX
Rx 11 Rx 12 Rx 1m Rx 21 Rx 22 Rx 2m Rx n1 Rx n2 Rx nm Rx p1 Rx p2 Rx pm
MUX
DMUX
MUX
DMUX
MUX
S D H
DMUX
e q u ip m e n t is
A D M
Figure 4-20 1:n protection based on single wavelength and implemented on SDH layer
In a DWDM system, switching of each SDH channel isn't related to other channels, i.e. when Tx11 of DWDM system 1 fails and switches to DWDM protection system 1, Tx12, Tx13 x1m can still operate in DWDM working system 1. Once a switching condition is detected, the protection switching must be completed within 50ms. 3. 1:n protection based on single wavelength within the same DWDM system Consider a DWDM line which can carry multiple SDH channels. The idle wavelengths in the same DWDM system can function as protection channels.
W orking system Tx1 W orking system Tx2 Txn Txp1 W orking system W orking system MUX LA W DM working system DMUX W orking system Rx1 W orking system Rx2 W orking system Rxn W orking system Rxp1
Figure 4-21 1:n protection based on single wavelength within the same DWDM system 4-75
DWDM Networking
Figure 4-21 shows a DWDM system of n+1 wavelength channels with n working wavelengths and 1 protective wavelength as protection system. However, in practical systems, the reliability of optical fibers and optical fiber cables is worse than that of the equipment. So practically, it does not mean much to provide protection only for the system instead of the line. Once a switching condition is detected, the protection switching must be completed within 50ms.
W orking system Tx1 Tx 2 Tx n MUX 1:2 optical splitter LA W DM working line 1 W DM working line 2 LA Protection system Rx 1 1:2 optical switch DMUX Rx 2 Rxn
Figure 4-22 shows an optical multiplex section protection scheme employing optical splitter and optical switch. In this protection system, only the optical fiber cables and DWDM line systems are backed up, and the SDH terminals and multiplexers in the DWDM system terminal stations aren't backed up. In practical systems, an N: 2 coupler can be used to replace the multiplexer and the 1: 2
4-76
DWDM Networking
splitter. When compared to 1+1 protection, this reduces the cost. OMSP has practical meaning only for two independent optical fiber cables.
WDM terminal
WDM terminal
WDM terminal
WDM terminal
In the protection system shown in Figure 4-24, path protection ring and MSP protection ring of the SDH system can be implemented. The DWDM system only provides "virtual" optical fibers. The protection for each wavelength on SDH layer is independent to the protection mode of other wavelengths. This ring can be twofiber or 4-fiber.
4-77
DWDM Networking
OADM
OADM
OADM
OADM
To employ OADMs with the add/drop multiplexing capability to form rings is another application mode of DWDM technology in ring networks. At present, ring networks formed by OADMs can be classified into two modes. One is wavelength path protection based on single wavelength protection, i.e. 1+1 protection of single wavelength which is similar to the path protection of SDH system. The other is line protection ring which protects the signals of multiplexed wavelengths. When a fiber is cut off, the "loop back" function can be implemented in the two nodes near the fiber cutoff. Thus all the services are protected. This is similar to the MSP of SDH. From the aspect of specific forms, the line protection ring can be divided into two-fiber bi-directional ring and two-fiber unidirectional ring, and four-fiber bi-directional ring can also be formed. In a two-fiber bidirectional ring, half of wavelengths operate as working wavelength and others as protection wavelength.
4-78
DWDM Networking
54KM A B
97KM C
125KM D
110KM E
86KM F
131KM G 138KM
4.5.2 Networking Diagram (considering the dispersion limited distance of the laserlaser to divide the regenerator sections of the network)
The dispersion limited distance of a laser depends on its modulation mode. Generally, the maximum dispersion limited distance of an EA laser can be up to 640km and that of an M-Z external modulated laser can reach 1000km (The line width of M-Z laser is too narrow, unfavorable for overcoming the nonlinear effects of optical fiber). Here we divide the regenerator sections of the network in terms of EA laser which are most commonly used in practical engineering. We analyze the distance between two adjacent stations which have services to add/drop, as shown in Figure 4-26. The distance between A and E is 386km, meeting the dispersion limited distance requirement of the EA laser. The distance between E and N is 1002km. It doesn't meet the EA laser dispersion limited distance requirement of less than or equal to 640km. So it is necessary to divide the distance between E and N. This distance can be divided into two or three shorter regenerator sections. Here we change the optical amplifier station I between E and N into an electrical regeneration station and divide the regenerator section between E and N into two regenerator sections: E---I and I---N, where the distance of regenerator section E---I is 453km and I---N is 549km. After division, the distances meet the EA laser dispersion limited distance requirement of less than or equal to 640km. The networking is shown in Figure 4-26.
4-79
DWDM Networking
Figure 4-26 Networking diagram which considers the dispersion limited distance of the lasers
4.5.3 Networking Diagram (considering the power of optical amplifiers to divide the optical regenerator sections)
According to the related ITU-T recommendations, the regenerating distances between adjacent DWDM stations can be specified as four types: 80km, 100km, 120km and 160km. The 160km standard is adopted only when no line amplifier (optical regenerating station) is used. In the cases that line amplifiers are employed,, the recommended distance is generally less than 120km (33dB). However, networking modes greater than 120km (33dB) can still be adopted as long as the specification such as power of optical amplifiers and OSNR meet the requirements. In applications with line amplifiers, the total launched power of the optical amplifiers is generally not greater than +20dBm (for an 16-wavelength system, the power is +8dBm in each channel), the received power of the preamplifiers isn't less than -30dBm in each channel and the distance between two adjacent stations (for an 16-wavelength system) shouldn't less than 139km (38dB/0.275DB/km). In Figure 4-26, the distance between stations I and J is 176km and exceeds the requirement of the optical amplifier. The distances between other adjacent stations almost meet the requirement. So it is necessary to add an optical regeneratingregeneration station between I and J, dividing I----J into two regenerator sections. As shown in Figure 4-27, the requirement of the amplifier is met after station X is added.
4-80
DWDM Networking
Figure 4-27 Network diagram considering the power of the optical amplifiers
4-81
DWDM Networking
4-82
DWDM Networking
4-83