`
`FIGURE 22.7
`Dispersion
`compensation in
`WDM system,
`showing difference
`in cumulative
`dispersion.
`
`FIGURE 22.8
`Cumulative
`dispersion plotted
`along a WDM
`system, which uses
`short lengths o f
`dispersion-
`compensating fiber.
`Total dispersion at
`the three
`wavelengths diverges
`along the length o f
`the fiber because
`the compensation
`inevitably is
`imperfect.
`
`MASIMO 2014
`PART 9
`Apple v. Masimo
`IPR2020-01526
`
`
`
`Optical Networking System Design
`
`Nonlinear Effects in WDM Systems
`Nonlinear effects are proportional to total optical power density and to the distance the
`light travels through the fiber at high power levels. Long-haul W DM systems are particu-
`larly vulnerable because the more optical channels they carry, the higher the total optical
`power. A single-channel system normally has no trouble transmitting 3 mW, but if a W DM
`system tries to transmit 80 channels at that power level, the total power reaches 240 mW,
`which can produce nonlinear effects and interactions among the transmitted channels.
`Nonlinear effects are relatively weak in glass fibers, but their total impact on the signal
`is proportional to the distance the signal travels in the fiber. In practice, they are not sig-
`nificant in metro systems that run tens of kilometers, but can pose problems in long-haul
`systems running thousands of kilometers.
`Fiber attenuation complicates the picture by reducing the optical power as a signal trav-
`els through the fiber, causing nonlinear effects to decline with distance from the light source.
`As a result, nonlinear effects accumulate over a maximum effective length, after an amplifier
`or transmitter which depends on the fiber attenuation. For a typical single-mode fiber with
`0.22 dB/km attenuation at 1550 nm, this is about 20 km per span between optical source
`and receiver. The value is smaller for fibers with higher attenuation. In long-haul systems,
`the maximum effective length is multiplied by the number of spans between amplifiers.
`Four-wave mixing poses particular problems in DW DM systems. As you learned in
`Chapter 5, it occurs when signals at three input frequencies combine to generate a mixed
`signal at a fourth frequency:
`
`V 1 + v 2 ~ v 3 = v 4
`
`The three input signals need not all be at different frequencies; two of them could be on
`the same optical channel. The equal spacing of W DM channels means that the new fre-
`quency is likely to fall on another optical channel, producing noise and crosstalk.
`Four-wave mixing increases if the three input waves remain in phase as they pass through
`the fiber. That occurs when there is no chromatic dispersion to spread them out along the
`fiber. If the fiber has some minimum chromatic dispersion in the transmission window—
`typically at least 1 ps/nm-km— the optical channels do not remain in phase over long dis-
`tances, and the four-wave mixing signal is reduced, as shown in Figure 22.9. This level of
`dispersion is too low to limit transmission bandwidth on the individual channels.
`Dispersion compensation does not enhance four-wave mixing because it uses lengths of fibers
`with positive and negative dispersion that combine to produce low dispersion over the entire
`fiber span. In such systems, the local dispersion is nonzero, so the signal pulses do not stay in
`phase over long distances. In short, near-zero cumulative dispersion over a fiber span is fine, but
`zero local dispersion in a single fiber can keep signals in phase and enhance four-wave mixing.
`The dependence of nonlinear effects on power density makes them sensitive to the
`choice of fiber. Fibers with small effective areas concentrate signal power, making them
`more vulnerable to four-wave mixing and other nonlinearities. Typically, the effective area
`is large in step-index single-mode fibers, somewhat smaller in nonzero dispersion-shifted
`fibers, and smallest in dispersion-compensating fibers. Other trade-offs occur between ef-
`fective area and reduced dispersion slope; increasing effective area tends to increase disper-
`sion slope, and conversely, reducing the dispersion slope tends to decrease the effective area.
`
`High total power
`makes WDM
`systems vulnerable
`to nonlinear
`effects.
`
`Attenuation limits
`nonlinear effects
`to a maximum
`effective length.
`
`Fiber type affects
`nonlinear effects.
`
`
`
`Chapter 22
`
`FIGURE 22.9
`Four-wave mixing
`is high at the
`zero-dispersion
`point, where
`signals stay in
`phase over long
`distances. Some
`local dispersion
`causes the
`wavelengths to
`drift out o f phase,
`reducing four-
`wave mixing.
`
`Zero local dispersion
`keeps channels in phase.
`
`Moderate local dispersion
`causes channels to drift out of phase.
`
`Initial
`
`A lo n g Fiber
`
`al
`
`A lo n g F ib e r
`
`Causes large four-wave mixing.
`1>1
`v2
`v3
`
`Reduces four-wave mixing.
`
`V) + v 2 - v 3 = v 4
`
`Designers can take advantage of these differences in properties when they select fibers
`for dispersion compensation. Nonlinear effects can be reduced by placing a large-effective-
`area fiber close to transmitters or amplifiers, where the optical power is highest. Fibers more
`vulnerable to nonlinear effects can be placed in parts of the fiber span where power is lower.
`
`Optical Amplification and WDM Design
`
`The role of optical amplifiers is more complex in W DM systems than in single-channel sys-
`tems. In single-channel systems, the variation of amplifier gain with wavelength limits the
`usable transmission band. In W DM systems, this variation adds requirements to balance
`gain across the transmission spectrum. The available optical power must be shared among
`the transmitted channels, and gain can be saturated when total input power reaches a high
`level, although power per channel remains modest. Amplified spontaneous emission is a
`concern because it adds to background noise across the gain spectrum.
`These considerations apply to all types of optical amplifiers, but this section will con-
`centrate on erbium-doped fiber amplifiers because they are the most common type in use.
`Amplifier Power Levels
`Amplifier output is Erbium-fiber amplifiers have a maximum output power, typically 17 to 24 dBm for C-band
`amplifiers. The limit comes from energy transfer, both in exciting the erbium atoms and in
`divided among
`stimulating emission, which causes the saturation effects described in Chapter 12. This peak
`many W DM
`output is concentrated on a single wavelength in a single-channel system, but in a W DM
`channels.
`
`
`
`Optical Networking System Design
`
`system it is divided among all populated optical channels, limiting the power per channel. As
`a result, the power per channel in a WDM system decreases with the number of channels. For
`example, an amplifier operated at its maximum output of 80 mW could deliver 20 mW on
`each of 4 optical channels, 10 mW on each of 8 channels, or 1 mW on each of 80 channels.
`Multichannel operation can limit gain on individual channels to much lower values than
`are possible with a single channel. If an amplifier saturates at a power level of 15 dBm
`(30 mW), it can amplify a single -1 5 dBm input channel by 30 dB. However, if the input
`were 30 input channels each at -1 5 dBm (30 |xW), the total output power still would be
`limited to 30 mW, or 1 mW per channel, a gain of only 15 dB. This has important system
`consequences because it means the fiber span between amplifiers can include only 15 dB
`of loss— equivalent to 75 km of fiber with 0.2 dB/km attenuation. Adding optical channels
`without upgrading the amplifiers can downgrade transmission capability. Depending on
`operational details, doubling the number of optical channels would reduce the power level
`on each channel by 3 dB, equivalent to 15 km of 0.2 dB/km fiber.
`Amplifier power also affects maximum data rate per optical channel. As you learned in
`Chapter 11, pulse detection requires a minimum number of photons, so average powers must
`be increased to deliver those photons in shorter pulses. Thus dividing power among more opti-
`cal channels, which can transmit more data in parallel, reduces the power available to transmit
`signals at a higher data rate per channel. In short, when operating near system margins, increas-
`ing the number of optical channels trades off directly with increasing the data rate per channel.
`Gain Flatness and Channel Equalization
`Erbium amplifier gain typically is flat to within 1 to 3 dB across its operating range. This
`sounds good until you start cascading amplifiers. That variation is acceptable for a single
`amplifier, but not in a cascade. A difference of 2 dB between channels becomes 10 dB af-
`ter five amplifiers, which could lead to the loss of weaker channels. Gain-equalizing filters
`can compensate for this by reducing the power on the strongest optical channels, as shown
`in Figure 22.10. The filter’s extra attenuation offsets the stronger gain at certain wave-
`lengths, making amplifier gain flat across the spectrum. Some imperfections inevitably re-
`main, but careful engineering can limit channel-to-channel power differences to no more
`than a few decibels over a chain of more than 100 amplifiers.
`Another approach to gain equalization is by adding another amplifier to add power to the sys-
`tem rather than subtract it. Raman amplifiers are an attractive choice because their gain is higher
`at longer wavelengths than at shorter ones; they complement the spectrum of erbium amplifiers,
`which peak at shorter wavelengths. Hybrid amplifiers combine a Raman amplifier with an
`erbium-doped fiber amplifier to give more uniform gain across a range of wavelengths. The
`Raman amplification stage may be in the transmission fiber, but it requires a strong pump beam.
`
`9
`Filters can
`equalize amplifier
`9a 'n on optical
`channels,
`
`Switching and Optical Networking
`
`A central concept of optical networking is managing signals by the optical channel, some-
`times called a lam bda after the Greek letter X used as an abbreviation for wavelength. The
`idea is to transmit signals in many bit streams at separate wavelengths, such as 40 channels
`
`Optical
`networking
`manages signals
`as optical
`channels.
`
`
`
`Chapter 22
`
`Gain before Equalization
`
`Transmittance of Gain-Flattening Filter
`
`Gain after Equalization
`
`ro
`
`o
`
`-*
`
`in
`
`-»
`
`o
`
`tn
`
`d
`
`Gain (dB)
`
`10
`
`5
`
`mS
`
`
`
`1 5 3 0
`
`1 5 5 0
`154 0
`Wavelength (nm)
`
`1560
`
`1 5 3 0
`
`1 5 5 0
`1 5 4 0
`Wavelength (nm)
`
`156 0
`
`1 5 3 0
`
`1 5 5 0
`1 5 4 0
`Wavelength (nm)
`
`1 5 6 0
`
`Transmission Characteristics
`before Equalization
`
`" ! """"!!!" !!!"!
`
`J 3 -
`Transmission Characteristics
`after Equalization
`
`FIGURE 22.10
`Effect o f gain-flattening filter on an optical amplifier. (Courtesy o f Furukawa Ltd.)
`
`at 2.5 Gbit/s rather than a single 100-Gbit/s data stream. Multiple channels offer more
`granularity because each data stream can be detected and processed as an optical channel
`without disturbing any of the others.
`Figure 22.11 shows how a simple optical network switch in Chicago can distribute eight
`optical channels from Omaha. First the input signal is demultiplexed to its eight compo-
`nent channels. Two wavelengths go to both Detroit and Indianapolis. One wavelength each
`goes to Minneapolis, Milwaukee, Chicago suburbs, and St. Louis. In this example, the in-
`put signals are organized in Omaha, and the Chicago switch directs them on their way
`without any further processing. The same fiber could carry many other wavelengths going
`to other destinations, but you couldn’t make sense of a drawing wirh 40 optical channels.
`This example looks simple because we’ve omitted crucial details. The optical switch is just
`a box; we don’t know how it processes the light. The figure doesn’t show the conversion of
`signals from one wavelength to another, although this may sometimes be necessary. To un-
`derstand how optical networks are designed, let’s take a closer look at some key concepts.
`Transparent, All-Optical, and Opaque Systems
`Both optical switches and networks can be classed into three categories that overlap to some
`extent.
`
`9 Transparent systems transmit optical signals without changing their format, as if the
`light were shining directly through them. They can be amplified, but their wavelength
`remains unchanged, and the optical signal is never converted into electronic form.
`
`
`
`M in n e a p o lis
`
`M ilw a u k e e
`
`Optical Networking System Design
`
`FIGURE 22.11
`Granularity o f
`optical channels.
`
`9 All-optical systems transmit and manipulate only optical signals, which are never
`converted to electronic form. However, the optical signals may be converted to
`different wavelengths. All transparent networks are all-optical, but some all-optical
`systems are not transparent.
`• Opaque systems convert the input optical signal into electronic form for switching
`or processing, then convert that signal back into optical form. The output may be at
`the same wavelength or a different one.
`
`A network does not have to be entirely of one type. An optical network may include
`“islands” of transparency that transmit signals purely optically, separated by opaque opto-
`electronic components such as receiver-transmitter pairs or electronic switches.
`Present systems use both all-optical and electro-optical switches. The big switches at net-
`work nodes that direct signals among many input and output ports are electronic, con-
`verting input optical signals into electronic form for switching, then converting them back
`into optical form for transmission. Electronic switches take advantage of highly-developed
`optical techniques for processing signals and directing them among large numbers of pos-
`sible output ports. All-optical switches generally are simpler devices used to transfer opti-
`cal signals between fibers without additional processing. One family redirects all signals in
`the fiber to another fiber; they are used for protection switching. Another selects one or
`more wavelengths in a W DM signal and redirects only those wavelengths; they can be used
`as add/drop switches. The signal-distribution switch shown in Figure 22.11 could be im-
`plemented with this type of optical switch.
`The advantages of transparent and all-optical networks have been much praised, but in
`practice some electronic components are much more mature than their optical counterparts.
`For example, wavelength conversion is simplest to implement by converting optical signals
`into electronic form and using the electronic signal to drive a laser transmitter at a different
`wavelength. As you learned in Chapter 12, it’s also possible to convert wavelengths by
`purely optical means, but that technology is still in development.
`
`An optical
`network may have
`islands of
`transparency
`separated by
`opaque elements.
`
`
`
`Chapter 22
`
`Wavelength
`conversion is
`needed to
`manage optical
`channels.
`
`The optical
`network
`terminates at
`electronic
`transmission
`centers.
`
`Wavelength Conversion and Routing
`Wavelength conversion is an important management tool in optical networks. From the net-
`work standpoint, the important part of a signal is the information it carries, not the wavelength
`of the carrier signal. The wavelength is like a lane on the highway, a channel for carrying in-
`formation. When signals switch from one fiber to another, the wavelength used on the first
`fiber may not be available on the second, so the signal may have to be converted to a different
`wavelength. Ideally, the output wavelength should be tunable, so the signal can be switched to
`any desired wavelength. This ideal isn’t easy to implement, but developers are working on it.
`Another application for wavelength conversion is in conjunction with a device called a
`wavelength router, which directs input signals to different ports depending on their wavelength.
`A wavelength router can be viewed as a fixed wavelength-division demultiplexer, which always
`directs input signals out a particular port according to their wavelength. In the example of
`Figure 22.11, an input signal at 1542.14 nm might go to Milwaukee, one at 1542.94 nm
`might go to Minneapolis, and signals at 1543.73 and 1544.53 nm would go to Detroit. By
`changing the input wavelength, you can change the output port to which the signal is directed.
`
`System Interfaces and Regeneration
`The optical network interfaces with electronic signals at input and output ends. The input
`electronics typically organize the input electronic signals in some way, such as time-division
`multiplexing to combine signals for high-speed transmission, or regrouping signals from
`other inputs. Then the signals are converted to optical form, and optical interfaces combine
`signals at different wavelengths at the input of W DM systems.
`Interfaces generally are at transmission centers, nodes, or hubs, where many trans-
`mission lines come together. Typically local signals feed into these nodes, which com-
`bine them with signals arriving from other points and transmit them on outgoing lines.
`Big electronic switches regenerate optical signals at major network nodes spaced several
`hundred kilometers apart.
`Optical regeneration has been demonstrated in the laboratory, but has yet to find prac-
`tical applications. System requirements for optical regeneration are not yet clear.
`
`Design Examples
`
`Designing W DM systems is a complex task that requires considering performance at many
`wavelengths and the interactions of the multiple signals passing through the system. To
`demonstrate how it works, we will concentrate on very simple examples that illustrate a few
`of the considerations.
`
`Amplifier Gain and Power
`Suppose we need to transmit 10 optical channels between two switching nodes 200 km apart.
`We have a single fiber, and space for an optical amplifier at the midpoint. The attenuation
`of each span is 25 dB, and the transmitter has 3-mW output on each of the 10 channels
`
`
`
`Optical Networking System Design
`
`after the multiplexer. The company warehouse has optical amplifiers with performance
`similar to that shown in Figure 12.5. Small-signal gain is 30 dB and peak output is 30 mW.
`The demultiplexer has 5 dB loss per channel, and we haven’t picked the receivers yet.
`At first glance it looks like there’s plenty of margin for operation on a single channel.
`Single-channel transmission
`
`Transmitter signal
`Attenuation
`Amplifier input
`Amplifier gain
`
`Amplifier output
`
`dBm
`5
`-2 5 dB
`-2 0 dBm
`27 dB
`
`7
`
`dBm
`
`Remember, however, that with nine other channels the total amplifier input is -1 0 dBm,
`which reduces the gain.
`
`Multichannel transmission (per channel)
`Transmitter signal
`5 dBm
`Attenuation
`-2 5 dB
`Amplifier input
`-2 0 dBm
`Amplifier gain
`22 dB
`Amplifier output
`2 dBm
`Attenuation (span)
`-2 5 dB
`Attenuation (demux)
`-5 dB
`
`Receiver power
`
`-2 8 dBm
`
`Receiver sensitivity o f-3 5 dBm is required to achieve a 7-dB system margin.
`Adding Optical Channels
`Now let’s consider what happens if we add another 10 optical channels to the system
`using the same amplifier and demultiplexer. The amplifier already produces 16 mW
`(10 channels at 1.6 mW each), so assume the extra input power reduces gain 2 dB.
`
`20-channel transmission (per channel)
`
`Transmitter signal
`Attenuation
`Amplifier input
`Amplifier gain
`Amplifier output
`Attenuation (span)
`Attenuation (demux)
`
`Receiver power
`
`5 dBm
`-2 5 dB
`-2 0 dBm
`20 dB
`0 dBm
`-2 5 dB
`- 5 dB
`
`-3 0 dBm
`
`
`
`If we use the same receivers, the system margin is a rather thin 5 dB. This small margin
`is one reason why adding channels to W DM systems may require more extensive upgrades
`of optical amplifiers, transmitters, and receivers.
`Gain Equalization
`Suppose we have a chain of five amplifiers in a W DM system that transmits signals over
`a span of 600 km on a dozen optical channels. For proper performance, the receivers
`require signals between -3 0 dBm and -2 8 dBm. How much gain variation can be tol-
`erated in each amplification stage, assuming that the variation is the same for each
`amplifier?
`The allowable variation is 2 dB divided by 5 amplifiers, or 0.4 dB for each amplifier.
`Gain equalization is needed to keep the variation within these limits.
`Gain equalization normally uses filters to block the additional power. If the amplifiers
`normally have 3 dB of gain variation across the spectrum, the filters will reduce the ampli-
`fier gain by 3 dB, so the gain budget will have to be increased. This is an example of the
`many trade-offs involved in W DM design.
`
`What Have You Learned?
`
`1. W DM is a fundamentally different way to organize signals than time-division
`multiplexing.
`2. Granularity is the subdivision of signals in a network.
`3. Total transmission capacity is the amount of information that can be transmitted
`by a network. It depends on data rate per optical channel, the number of
`channels, and channel spacing. It also depends on the range of wavelengths
`transmitted.
`4. Optical amplifiers are not available for all fiber transmission wavelengths,
`limiting the bands usable for long-distance transmission. Fiber attenuation and
`dispersion limit the bands usable for shorter distances.
`5. Dense-WDM is used for long-distance transmission. Coarse-WDM is used for
`tens of kilometers, and is less expensive to implement.
`6. Dispersion limits fiber transmission distance at high data rates. W DM provides
`higher total transmission capacity on multiple channels with lower data rates.
`7. Spectral efficiency measures how tightly signals can be packed, dividing the total
`data rate by the optical bandwidth used. The best rates achieved without
`elaborate polarization schemes are about 0.8 bit/s/Hz.
`8. Network operators do not populate all available channel slots.
`9. The ITU G694.2 standard established a grid of 18 CW DM slots with center
`wavelengths from 1270 to 1610 nm. This grid is used mainly in metro
`telecommunications networks.
`10. Optical amplifiers have gain bandwidths of tens of nanometers; this limits the
`wavelength ranges available for long-distance DW DM systems.
`
`
`
`Optical Networking System Design
`
`11. Raman gain is offset from the pump wavelength by an amount that depends on
`the material; the value is 13 THz in silica. The Raman wavelength is longer
`(lower in frequency) than the pump wavelength.
`12. Fiber loss varies significantly in the 1250 to 1650 nm; it is 0.35 dB/km at
`1310 nm and under 0.2 dB/km at 1570 nm. Variation in the erbium amplifier
`band is much smaller.
`13. Dispersion slope measures variation of chromatic dispersion with wavelength.
`14. It’s impossible to compensate for chromatic dispersion perfectly across a range of
`wavelengths because the dispersion curves of different fibers don’t cancel each
`other out perfectly. These differences increase over long distances.
`15. Nonlinear effects increase with total power and with the distance the light travels
`at high power. Attenuation reduces nonlinear effects far from the transmitter.
`16. Saturation limits total power available from an optical amplifier. Saturation depends
`on total power, not just power at a single wavelength, so W DM systems can
`saturate.
`17. Opaque systems convert optical signals into electronic form at some point,
`typically for switching or wavelength conversion.
`18. Wavelength conversion is needed to manage optical channels. Wavelength routers
`direct signals according to their wavelength.
`
`What's Next?
`
`In Chapter 23 you will learn about the structure of the global telecommunications network.
`
`Further Reading
`
`Vivek Alwayn, Optical Network Design and Implementation, (Cisco Press, Indianapolis, 2004)
`Rajiv Ramaswami and Kumar N. Sivarajan, Optical Networks: A Practical Perspective, 2nd
`ed. (Morgan Kaufmann, San Francisco, 2002)
`
`Questions to Think About
`
`1. Your design packs 2.5-Gbit/s optical channels only 50 GHz apart, but 10-Gbit/s
`signals require 100 GHz spacing to give adequate performance margins. Which
`spacing transmits more data if you populate every available slot in the erbium-
`amplifier C-band? How much is the difference?
`2. Wavelengths of 1460 to 1625 nm are available in an unamplified metro W DM
`system. How many CW DM optical channels can this network transmit with
`ITU standard 20-nm spacing?
`3. You have two types of fiber available for dispersion compensation, step-index
`single-mode fiber with dispersion of + 1 7 ps/nm-km at 1550 nm, and a reduced-
`slope non-zero dispersion-shifted fiber with - 2 ps/nm-km at the same wavelength.
`
`
`
`How much of each do you need for a 95-km span with zero cumulative
`dispersion at 1550 nm? Which type should you use closer to the transmitter?
`
`4. An erbium-fiber amplifier that has peak output of 20 dBm transmits 40 optical
`
`channels. What is the output power on each channel? What happens to the
`output power if the number of channels is doubled?
`
`5. You need to send 20 optical channels through a series of 50 erbium-doped fiber
`
`amplifiers, and the output signals must differ by no more than 6 dB at the end
`of the amplifier chain. How should you equalize the gain assuming all that
`difference is due to unequal gain across the range of optical channels?
`6. The input to an all-optical switch is a cable containing 48 fibers. Each fiber can
`transmit up to 40 optical channels. How many optical channels can the switch
`direct if it has 128 input and 128 output ports?
`
`Chapter Quiz
`
`1 . One 40-Gbit/s TD M channel is equivalent to how many 2.5-Gbit/s optical
`channels?
`a. 1
`b. 4
`c. 16
`d. 40
`e. 100
`
`2. A W DM system has 40 optical channels spaced 100 GHz apart in the 1550-nm
`
`region. What is the approximate total spectrum used by the system?
`a. 40 GHz
`b.
`100 GHz
`c.
`400 GHz
`d. 4 THz
`e. 193.1 THz
`
`3. How is the spacing required between optical channels related to the data rate
`
`transmitted on each channel?
`a. The spacing increases with the data rate; the required spacing in gigahertz
`is larger than the data rate in gigabits.
`b. The spacing equals the signal speed
`c. The spacing
`in nanometers is equal
`d. The spacing decreases with the data
`the data pulses.
`e. The relationship is impossible to state because data rate is digital and
`frequency spacing is analog.
`
`gigahertz.
`in
`the data rate in gigabits per second.
`to
`rate; it is proportional to the length of
`
`
`
`Optical Networking System Design
`
`Which of the following systems can accommodate the most optical channels?
`a. coarse-WDM in a system that includes several erbium-doped fiber
`amplifiers
`b. dense-WDM in a system that includes several erbium-doped fiber amplifiers
`c. coarse-WDM in a system without optical amplifiers
`d. dense-WDM in a system without optical amplifiers
`e. not enough information to tell
`You need to compensate dispersion for a 60-kilometer length of fiber with
`chromatic dispersion of —3 ps/nm-km. How much fiber with chromatic
`dispersion of + 15 ps/nm-km at the same wavelength (1550 nm) do you need?
`a. 5 km
`b. 12 km
`c. 15 km
`d. 25 km
`e. 60 km
`Your fiber has a dispersion slope of 0.08 ps/nm2-km in the C-band of erbium-
`fiber amplifiers. Assuming that slope is a straight line, how much does the
`dispersion change over the width of the C-band?
`a. 0.08 ps/nm-km
`b. 0.8 ps/nm-km
`c. 2.8 ps/nm-km
`d. 8 ps/nm-km
`e. 28 ps/nm-km
`A span between two fiber amplifiers includes three types of fiber. Which type
`should be closest to the output of the first amplifier to reduce four-wave mixing?
`a. step-index single-mode fiber with effective area 80 p,m2 and dispersion
`+ 15 ps/nm-km
`b. reduced-slope single-mode fiber with effective area 50 |xm2 and dispersion
`-2 .5 ps/nm-km
`c. large-effective-area single-mode fiber with effective area 68 pm2 and
`dispersion -2 0 ps/nm-km
`d. either of the low-dispersion fibers
`e. any of the fibers
`In which fiber is four-wave mixing the largest?
`a. step-index single-mode fiber with effective area 80 p,m2 and dispersion
`+ 15 ps/nm-km
`b. reduced-slope single-mode fiber with effective area 50 pan2 and dispersion
`—2.5 ps/nm-km
`
`
`
`Chapter 22
`
`c. large-effective-area single-mode fiber with effective area 68 pm2 and
`dispersion —2.0 ps/nm-km
`d. zero dispersion-shifted single-mode fiber with effective area 75 |xm2 and
`dispersion 0 ps/nm-km
`e. 50/125 graded-index multimode fiber
`9. An optical amplifier generates 2 mW on each of 40 optical channels. An optical
`switch downstream diverts half of the optical channels to another cable. If
`everything else remains constant, what is the power on the remaining channels?
`a. 2 mW
`b. 3 mW
`c. 4 mW
`d. 5 mW
`e. The amplifier will not work.
`
`10. What is the difference between a transparent and an opaque optical switch?
`
`form;
`
`through in optical
`light signals to pass
`a. A transparent switch allows
`an opaque switch converts them to electronic form.
`b. A transparent switch is made of clear glass; an opaque switch is made of
`metal.
`c. A transparent switch does not change the data rate of the signal; an
`opaque switch changes the data rate.
`d. A transparent switch transmits a signal; an opaque switch blocks the
`signal, turning it off.
`1 1 . You add eight new optical channels to an optical fiber that already was carrying
`eight optical channels. The signals must pass through an optical amplifier
`nearing saturation. All the signals arrive at the amplifier with the same power.
`What happens to the original eight channels?
`a. Output power on all channels after the addition is the same as it was on the
`original channels because the power per channel is well under saturation.
`b. Output power on the original channels is unchanged, but the new
`channels get less power.
`c. Output power per channel drops to a lower level.
`d. Output power on the original channels doubles.
`e. All channels fail to transmit a signal because they overload the amplifier.
`1 2 . Which of the following varies least in the 1550-nm transmission window?
`a. chromatic dispersion
`b. fiber attenuation
`c. erbium-amplifier gain
`d. raman amplifier gain
`
`
`
`23
`
`Global
`Telecommunications
`Applications
`
`About This Chapter
`
`Now that you have learned about fiber-optic hardware, standards, and system design, it’s
`time to look at how fiber-optic systems are used. Changing technology and regulations
`are eroding traditional divisions, but it is still useful to separate telecommunications into
`a few sectors. The largest in scale is the global telecommunications network, the back-
`bone of international telecommunications, including long-distance transmission under
`the oceans and on land. Other sectors are regional or metro networks and distribution
`networks for voice, video, and computer data services. You learned a little about these
`ideas in Chapter 3; now it’s time to take a closer look.
`In this chapter, I will describe the long-distance fiber-optic transmission systems that
`carry data, voice, video, and other signals around the world. They include interconti-
`nental submarine cables as well as national and international systems on land. They are
`the world’s biggest telecommunications “pipelines,” and they are designed to maximize
`both transmission speed and distance. Fiber-optic technology has dominated these systems
`for over a decade, first with single-channel single-mode transmission, and now with
`high-speed DW DM systems.
`These long-distance networks feed into regional or metro networks, which in turn
`connect switching offices and distribution networks, which deliver services to individual
`homes and offices. Later chapters will look at those networks.
`
`
`
`Chapter 23
`
`Telecommunications
`encompasses
`voice, data,
`facsimile, video,
`and other forms of
`communications.
`
`Satellites transmit
`video signals, and
`some voice and
`data services.
`
`Defining Telecommunications
`
`The term telecommunications is deliberately broad. It dates back to the era when communi-
`cation specialists were trying to group telephones and telegraphs under one heading. As the
`telegraph industry faded away, telephony become dominant, but the new word had caught
`on. It was useful because new communication services were emerging. Radio and television
`networks spread, broadcasting voice, music, and pictures. Telex relayed printed messages
`around the world. Facsimile systems transmitted images of documents. Computer data
`communications grew rapidly. Wireless telephones and pagers spread. They all fell under
`the broad heading of telecommunications.
`Different types of telecommunications had different origins, but most are converging
`toward a common network that delivers many different services. The main reason is that it
`costs less to build one versatile network than many specialized ones. Convergence will
`never be complete because some services inherently differ from others. A mobile phone
`small enough to fit into your pocket can display brief text messages, but not complex Web
`pages