Chapter 19 Digital Fiber Modulation and Deep Fiber Architectures
19.1 Introduction
This chapter covers binary (digital) modulation of a fiber-optic cable as well as deep fiber architectures. The two are covered together because binary optical transmission is used in last-mile (to the home and/or business) applications. In addition, binary optical transmission is used extensively in metropolitan loops and in intercity trunks. An advantage of digital modulation is that it can operate with much lower signal-to-noise ratio than can analog modulation, as shown in Figure 12.13.
Figure 19.1 illustrates the difference between digital and analog optical modulation. Figure 19.1 (a) represents digital modulation of the optical transmitter. The data is not modulated onto an RF carrier. Rather, the data directly modulates the laser, turning it on and off. If more than one datastream is provided, they are time division multiplexed (TDM’ed) by switching first to one and then to another until all have been sampled. Since no data can be lost during the multiplexing process, the output data rate must be the sum of all input data rates, usually with additional bits added to synchronize data recovery. The data rate actually carried by the laser is called the wire rate.
Contrast this to the common cable television technique of broadcast optical transmission, as covered in Chapter 12 and shown in Figure 19.1(b). In this system, we often transmit digital signals, which may be TDM’ed but then they are modulated onto RF carriers, normally using either 64- or 256-QAM modulation. These digitally modulated carriers are combined with analog modulated carriers, and the sum of all signals modulates an analog laser, which operates in its linear range, where output power is proportional to input current. The use of multiple carriers to transmit different signals is called frequency division multiplexing, or FDM.
Usually the term analog is applied to the optical transmitter in Figure 19.1(b). It is true that this must be an analog transmitter, since a digital transmitter would never be able to carry the multiple frequencies without creating intolerable intermodulation distortion. However, the signals carried as modulated signals on the analog transmitter may themselves be either analog or digital.
19.2 Difference Between Analog and Digital Modulation
Figure 19.2 illustrates the difference between analog and digital optical modulation of a laser. Compare with Figure 12.13, which covers analog modulation only. The analog signal occupies primarily a range of diode currents in a more nearly linear range of laser operation. Note that the real analog waveform is not a sine wave; rather, it exhibits quite high peaks, as shown in Figure 12.20. Because of this, the carriers really exceed the range shown, with resulting clipping distortion, as described in Section 12.10.3. The levels must be managed so that the clipping distortion is not excessive. The digital signal is truly a rectangular waveform occupying but two levels, with fast transition between the two.
Analog transmitters are characterized by the optical modulation index, or OMI, as explained along with Figure 12.14. By contrast, digital modulation is characterized by the extinction ratio, defined as the ratio between the maximum light output in one binary state to the minimum light output in the other binary state. Extinction ratio is expressed in decibels, as shown in Figure 19.2. The laser is biased so as not to extinguish the lasing process at any time, because restarting the lasing process causes problems in transmission. A high extinction ratio is desirable in order to maximize the ability to discern between the two states. However, over temperature and the life of the laser, the laser must never turn off during either of the two states. A typical minimum extinction ratio is about 9 dB.
19.3 Digital and Analog Transmitters
Figure 19.3 illustrates the block diagram of a digital transmitter in (a) and an externally modulated analog transmitter in (b). Data is supplied to the input of the digital transmitter, usually balanced to minimize signal radiation and noise pickup. The driver amplifier A1 is a limiting amplifier, meaning that its output is not a function of the input amplitude. The drive signal to laser diode D1 is coupled through switch S1 and coupling capacitor C.
Monitor photodiode D2 samples the average light output from D1. The light is converted to a current in D2, and the current to a voltage in R2. The voltage appears on the inverting input of leveling loop amplifier A2, which compares the voltage with a reference. If the two differ, then the output of A2 changes, thus changing the bias in D1 and hence changing its optical output power. The output of the leveling loop amplifier A2 is often supplied to a status-monitoring (also known as element management) system, which monitors the current. The sensitivity of the laser diode decreases as it ages, resulting in more bias current needed to keep the light output constant. As the bias current approaches a specified threshold, it is time to replace the laser. This threshold may be monitored in the element management system, and/or a comparator (not shown) may be added to the transmitter to generate a binary alarm signal.
Switches S1 and S2 are used when the transmitter must be switched off and on, as when it is used for time division multiple access (TDMA) applications. TDMA is explained in Section 4.3.1. When the laser is switched off, both the modulating signal from A1 and the bias from A2 must be removed to make sure the laser is extinguished.
Digital transmitters use F-P lasers for short distances and DFB lasers for longer distances. Because of their simpler circuitry and because the laser does not have to operate linearly, digital laser transmitters are significantly lower in cost than are analog transmitters.
Contrast this to the analog transmitter of Figure 19.3(b). For deep fiber applications, externally modulated transmitters are usually used because of their superior characteristics. Composite video is applied to amplifier A1. In a typical cable television architecture the composite input is the frequency division multiplexed set of signals that define the broadcast service. Attenuator AT1 sets the drive level. After driver amplifier A3, the signals are usually sent to a predistortion circuit, which compensates for distortion in the laser and signal path, by applying an opposite distortion to the signals before they are modulated onto the laser. Finally, signals are coupled to an external modulator, such as the Mach-Zender (M-Z) modulator described in Section 12.6.4. Monitor photodiode D2 serves to keep the output light level constant, as in the digital transmitter.
19.4 Digital and Analog Receivers
Figure 19.4 illustrates digital and analog receivers. In the digital receiver of Figure 19.4(a), detector diode D1 receives the light and converts it to a current. The average current is filtered in L1 and converted to a voltage in resistor R1. Signal output is developed across L1, and AC coupled through capacitor C1 to a transimpedance amplifier (TIA). The TIA has a high input impedance and a controlled output impedance, such as 50 ohms. It is a linear amplifier, so the output waveform is an accurate representation of the input light power. Automatic gain control (AGC) may or may not be used. The output of the TIA is filtered in low-pass filter FL1 to remove out-of-band noise and improve the waveform. Finally, the signal is shaped in limiter A3, which converts the analog levels to logic 1 and logic 0. Typically the output of A3 is balanced, again to keep switching currents out of the ground system. An activity detector detects the loss of either optical signal or digital modulation on the signal and can report this fact to a status-monitoring system.
Contrast the digital receiver just described with the analog receiver shown in Figure 19.4(b). Light is again converted to a current in D1, and the signal portion of the current is impedance transformed in T1 before being applied to preamplifier A1. The gain is adjusted in attenuator R2 under control of either an AGC loop or a feed-forward gain control circuit. A conventional AGC circuit is often used in optical nodes, but for other applications the feed-forward gain control may be used. The feed-forward circuit measures changes in the average optical level received and corrects the gain for the RF level changes produced by changes in optical level. This method is not as accurate as AGC, but it is lower in cost and is satisfactory for applications where level control is not critical. The average received signal level is derived as in the digital receiver, and a threshold detector A4 may be used to generate a loss of input alarm.
19.5 Combining Analog and Digital Transmission on the Same Fiber
It is quite possible to combine analog and digital transmission on the same fiber by using wave division multiplexing (WDM). Figure 19.5 illustrates WDM, in which one transmitter (usually the digital transmitter) operates at 1,310 nm, while the other (usually analog) operates in the 1,550-nm region. Other wavelengths are sometimes used. Two or more wavelengths may be combined using a wave division multiplexer, which is the optical equivalent of a diplex filter. Alternatively, a single mode coupler, analogous to an RF combiner, can be used; but as with RF combiners, the signal will suffer insertion loss. Since receivers are responsive over a wide range of wavelengths, a wave division multiplexer must be used to separate the two wavelengths at the receiving end.
Wavelengths closer to each other, such as multiple wavelengths in the 1550-nm region, may be multiplexed. The ITU has specified standard wavelengths in this band. The technique of multiplexing closely spaced wavelengths is referred to as dense wavelength division multiplexing (DWDM), as discussed in Chapter 13.
19.6 Low-Frequency Content Removal in Digital Transmitters
As with most communications systems, a digital optical system is not able to transmit frequencies down to DC, so there must be some method of low frequency (sometimes called DC) removal. Note that the transmitters and receivers are both AC coupled. There must be a way of ensuring that the data does not contain energy below the cutoff frequency of the system. Besides causing problems with the AC coupling in the transmitter and receiver, content approaching DC implies that you can have a long time between state (1 or 0) changes. Clock signals are recovered by locking a local clock to state changes in the incoming data, so an excessively long time without a state change will allow the recovered clock to wander out of phase, damaging data recovery.
DC removal may be done in a number of ways. One method is to establish rules limiting the number of like symbols (1 or 0) that can be transmitted in a row. Another common method is to use a substitution set of symbols for each byte of data to be transmitted. This technique is used in gigabit Ethernet transmission.1 The substitution method used in gigabit Ethernet and in certain other applications is called 8B/10B encoding. For every 8 bits (one byte), a 10-bit symbol is substituted, hence the name 8B/10B encoding. The substituted 10-bit symbol is chosen to have very close to an equal number of 1s and 0s and three to eight transitions per symbol. The codes satisfy the requirement of no DC component in the signal, and the large number of transitions ensure clock synchronization. Furthermore, since a limited number of the available codes are used, the encoding provides another way to detect transmission errors. The penalty is that, since 10 bits must be transmitted to represent 8 bits, the bandwidth required is increased by 25%. For instance, in a gigabit Ethernet system, the desired data is transmitted at 1 Gb/s, but because of 8B/10B encoding the data rate on the fiber (the so-called wire rate) is 1.25 Gb/s.
Table 19.1 illustrates a few of the many valid code groups used in gigabit Ethernet systems. The two columns labeled “Current RD—and “Current RD+” represent two so-called running disparity sets. The running disparity rules change the transmitted value from one column to the other based on certain rules related to the number of 1s or 0s that have been transmitted in the previous code group. These rules ensure that there is no DC content and that there is not a long string of like binary digits, thus ensuring reliable clock recovery.
19.7 Bidirectional Transmission
Almost all digital networks are bidirectional full duplex links, in which transmission takes place simultaneously in both directions. Full duplex transmission is accomplished differently in different situations. For long-distance transmission using either ATM or Ethernet, it is common to use different fibers for transmission in the two directions. Because of the high volume of transceivers for these applications, it is possible to buy low-cost, very small transmitters, sometimes called GBICs in the Ethernet world, specified for different distances to be covered.
Some systems intended for limited (perhaps up to 10 km) transmission distance transmit in both directions on the same fiber at the same wavelength (usually 1,310 nm). This works for digital transmission because the signal-to-noise ratio required is substantially less than that required for analog transmission and also because the isolation in optical couplers can be quite high (greater than 55 dB). In order to make single fiber bidirectional transmission work, it is necessary to handle digital optical connections the same way you would handle an analog optical connection, in terms of cleaning and the use of angle polished connectors. The handling techniques required are described in Section 12.5.1.
Some systems use a third wavelength to transmit digital signals in one direction. For example, 1310 nm might be used for upstream digital transmission, 1490 nm might be used for downstream digital transmission, with 1550 nm reserved for analog transmission. This plan is used in the FSAN standard, ITU G.983.3 described in Section 19.9.3. Figure 12.6 shows that optical fiber can transmit light at these wavelengths with low attenuation. A special WDM element with three passbands at 1310, 1490, and 1550 nm is used to separate or combine the wavelengths with lower loss than is possible using a wideband coupler.
19.8 Fiber-Deep Architectures
Several architectures are being used to drive fiber very deep into the network, even as far as the home. These architectures are driven by the improved quality of signals delivered over fiber-optic plant as compared to those delivered over coaxial plant, by improved reliability due to fewer devices in the network, and by improved bandwidth. The lack of amplifiers can mean lower maintenance costs due to the lack of a need to sweep and balance amplifiers. At least in FTTH systems shown in Section 19.8.2, the RF signal level never approaches 10−4 W (+ 38.75 dBmV), the level above which composite leakage index (CLI) measurements must be made. Thus you can avoid the cost of making measurements and filing reports.
19.8.1 Fiber-to-the-Curb Architectures
In fiber-to-the-curb (FTTC) architectures, fiber-optic cable brings signals to a termination point, where they are converted to electrical form for distribution to a small number of subscribers. From the termination point, coaxial cable is used for video services, telephone cable is used for telephone service, and Cat5 or Cat6 cable (see Chapter 5) is used for data. Characteristically, no amplification of broadcast signals is provided after the optical termination point in systems that distribute broadcast signals. Not all FTTC systems have included distribution of broadcast signals.
Figure 19.6 illustrates a typical FTTC system. One or more optical fibers extend from the headend to an optical termination point. Besides the normal conversion from optical to RF for the broadcast information (the downstream function of an HFC node), the optical termination point includes a bidirectional digital optical interface and interfaces for telephone and data networks. These are covered more fully next (fiber-to-the-home systems).
19.8.2 Fiber-to-the-Home Systems
Fiber-to-the-home (FTTH) systems have been built in several different configurations. They can be subdivided into several types, but all have in common that they extend to the home the interfaces found in FTTC systems.
Passive Optical Networks (PONs)
Figure 19.7 illustrates a basic FTTH system using a passive optical network (PON). Terminology borrowed from the telephone industry is often used in such systems to define the endpoints. The headend (or hub or central office) side is known as an optical line termination (OLT). For a PON, this would include the broadcast optics and the routers used for digital transmission. The subscriber-side circuitry, which typically goes on the outside of the home, is called an optical network termination (ONT). Usually, the subscriber-side interfaces at the ONT include a broadcast coaxial connection (when broadcast service is supported), one or more Ethernet connections, and one or more analog telephone interfaces.
One point of passive optical splitting is shown in the figure, but in practice two optical splitting points are often used. The primary advantage of a PON over active architectures described next is that there is no active equipment in the field between the hub and the subscriber. No power is required in the field. A disadvantage is the limited distance and a limited number of times the signal can be split, as shown next. Furthermore, more expensive optics may be required for high-speed data.
PONs are differentiated by the underlying layer 2 transport system used (see Chapter 5). Some are based on ATM and are known as APONs. Others are based on Ethernet and are known as EPONs. All deliver data to the home, with Ethernet connections being the dominant subscriber-side connection. In the case of APONs, ATM is transparent to the subscriber.
If the PON carries analog (frequency division multiplex, using both analog and digital modulation) signals for broadcast program delivery, some people add the designation BPON, for broadcast PON. Not all PONs carry analog signals. Some are intended solely as data networks, and some assume that video will be carried only in IP packets (IPTV).
Active FTTH
Figure 19.8 illustrates an active FTTH system. It is differentiated from a PON by the addition of active processing equipment in the network. Power is required at that equipment. In known systems, there is only one active component between the hub and the subscriber. Beyond the processing point, fiber may be home run (a dedicated fiber or fibers exit the active equipment for every home), or a fiber may be shared using optical taps, analogous to RF taps. The active equipment may be installed in a roadside cabinet, a controlled environmental vault (CEV), or may be strand or pedestal mounted, as is an optical node or an amplifier in an HFC network.
The disadvantage of an active fiber system is that power must be provided for the active point, though the system may consume less power per home than do HFC networks. The advantages of having the active device include a much greater reach from the hub to the home and greater degrees of splitting. A further advantage is lower costs due to the ability to employ lower-cost optics at the home. The active system also enables network powering of the home unit if desired. Otherwise, the home unit is powered from the home, using a battery-backed power supply if desired.
19.8.3 Classifying Fiber-Deep Systems
Figure 19.9 is a classification of the different fiber-deep systems. We shall classify them first as either fiber-to-the-curb (FTTC) or fiber-to-the-home (FTTH) systems. FTTC systems are often passive from a neighborhood optical termination point (OLT), but the OLT is active and can be rather large for the number of subscribers served. FTTH systems are more prevalent and may be all-passive or may have some active components, as shown in Figure 19.8.
Data transport on the network may be based on either ATM or Ethernet (see Chapter 5). ATM has its background in the telephone industry and is known for excellent quality of service (QoS) capability. Ethernet has its background in local area networks. Traditionally Ethernet has not incorporated QoS features as are available in ATM, though recent advances have allowed Ethernet to take on QoS features similar to those of ATM.
Telephony transport tends to be circuit-switched-based in ATM-based systems and VoIP-based in Ethernet-based systems. Chapter 6 provides more information on these two options.
Finally, video delivery is usually traditional broadcast. We have used the term FDM for this, since the signals are modulated onto many RF carriers and transported over linear networks. Both analog and digital modulation can be, and are, used. A variation is one system that uses an HFC overlay for video and an FTTH system for data. Where an existing HFC network is in place and provides adequate video service, you may want to start with an overlay for data, emphasizing business applications, and later migrate video delivery onto that system, eventually phasing out the HFC plant at the end of its life.
The alternative video delivery mechanism is to send video via Internet Protocol as part of the data stream (IPTV — Internet Protocol television). Given adequate bandwidth, this works. Some providers use IPTV exclusively, but others see it as an adjunct to the FDM delivery of most video. An interesting concept is to consider sending video on demand (VOD) over IPTV while using FDM for all other video services. Since VOD is by definition a point-to-point service being used by only one subscriber, this may be a better use of bandwidth than tying up broadcast spectrum when it is serving only one subscriber.
19.9 Distance Limitations in Fiber-Deep Systems
Fiber-deep architectures must trade off distance and the number of times the signal can be split. Each time the signal is split, loss is introduced, as described earlier. There are practical limitations on the amount of power that is injected into the fiber, and there are minimum optical signal levels that are required at the receiver to yield good performance. One distance limitation is the tolerable loss in the fiber network (defined as the difference between injected signal and minimum satisfactory received signal level). Sometimes the distance may be limited by other criteria. A budget may be constructed for various cases.
19.9.1 Limitations on the Distance of Analog Transmission
Stimulated Brillouin Scattering (SBS)
SBS limits the maximum power that can be injected into the fiber. When the power level of monochromatic light injected into a long fiber strand is increased, the output power increases proportionately until a threshold is reached. Beyond that level, the received power stays relatively constant, and light scattered back toward the source increases dramatically. The carrier-to-noise and signal-todistortion ratios of the received signals both degrade. This phenomenon, known as stimulated Brillouin scattering (SBS), is described in Section 12.4.6. SBS limits the maximum power that can be injected into the fiber. As the state of the art in optical components has improved, the maximum injection level has increased. The current state of the art is about +16 dBm optical power injection into a long fiber at any one wavelength. This high power is achieved by applying any of several techniques to spread the spectrum of the emitted light. The reason spreading the spectrum works is that SBS is a function of light energy at one wavelength. Transmitting at multiple wavelengths partially mitigates the effect of SBS but introduces dispersion issues.
Shot Noise
A significant limitation in the performance of analog optical receivers is shot noise. Shot noise is generated in the receiver photo diode, caused by the statistical variation in the arriving photon distribution. The subject is covered in Section 12.8. As the modulation of the light source (the OMI) increases, C/NSHOT also improves. Higher diode responsivity also improves the carrier-to-shot-noise ratio. When shot noise is a significant contributor to total optical link noise, the total C/N (carrier-to-noise ratio) will change less as a function of optical receive level than would be the case based on just thermal noise considerations.
Thermal Noise in the Fiber-Optic Receiver
The heart of an optical receiver is the photodiode, which has an output current proportional to the light power impinging on the diode. The photodiode, being a current output device, is often coupled to a transimpedance amplifier (TIA), which amplifies the current and converts it to a standard impedance, usually either 50 or 75 ohms. The C/N of the output signal is dependent on the shot noise current in the diode, as shown earlier, and by thermal noise in the amplifier(s) following the photodiode.
As a practical matter, most optical receivers designed to receive analog carriers are limited to a minimum light input level of about — 5 dBm. For node operation in HFC plants, the minimum level will be somewhat higher. This is because a portion of the noise budget must be reserved for coaxial amplifiers following the node in HFC networks, whereas in most fiber-deep systems there is no amplification following the node.
Optical Amplifier Noise Contribution
Section 12.7 describes the most common type of optical amplifier, an EDFA. Its noise contribution is the final significant contribution to the noise performance of an analog link. The noise added depends on the input signal level and on the amplifier’s noise figure. When several amplifiers are used in cascade, their noise contributions add, just as with RF amplifiers.
Other Noise Contributions
There are a few other noise contributors besides the ones just described, which as a practical matter tend to be of a less limiting nature. These include laser relative intensity noise, which is defined in Section 12.6.
Analog Transmission Distance Limitation
We have established that the maximum optical signal level into the fiber is about +16 dBm, limited by SBS, and that the minimum level at the optical receiver is — 5 dBm. This yields a total loss budget for an analog link of 22 dB. We can convert this to distance, given various splits in the fiber. Optical splices have a loss of 0.1-0.2 dB, and connectors can have loss of upto 0.5 dB. We shall allow 2 dB for splice and connector loss, though more may be necessary in the real world. This gets us to a loss budget of 20 dB. From this we can construct a table of the number of splits and the distance we can go without amplification. Note that this is based on nominal loss of the splitters. Real splitters may have more loss, shortening the distance.
Table 19.2 summarizes the maximum distance you can transport analog signals, given no amplification in the plant and using state-of-the-art components. We have started with the loss budget, subtracted connector loss, subtracted the nominal splitting loss, and allocated the remainder to fiber loss, converting this to distance. This assumes we start with the highest power we can supply to the fiber without encountering problems with SBS. We also assume high OMI in the transmitter (on the order of 3-3.5% per carrier), required to achieve good C/N. This table is just an estimate, based on thermal noise being the limiting factor in length. The table is constructed so that with modern components you can achieve a C/N of 48–49 dB. Some practitioners have gone to as few as a four-way split to make sure they have adequate range, but with splits this small PONs force a very large number of optical cables and transmitters at the hub. Notice that we have included the maximum distance for a 64-way split, tongue-in-cheek. This split has so much loss that you cannot make it work for video at all, given the best components available today.
Of course, optical amplification may be used to achieve longer distances. This is done in active FTTH systems. Adding an amplifier converts a PON into an active system.
19.9.2 Limitations on the Distance of Digital Transmission
Signal-to-Noise Ratio
As a practical matter, digital transmission, where the laser is turned on and off (amplitude shift keying, ASK), is limited by thermal noise, mode partition noise, and crosstalk. Figure 12.12 shows that the signal-to-noise-density ratio required for digital transmission is significantly lower than that required for analog transmission. The receiver includes a low-pass filter that limits the noise band-width. When the bandwidth of the transmission is doubled, the low-pass filter cutoff frequency must also double, meaning that the noise power doubles. Thus, every time the data rate doubles, the received signal power must also double. You can double the received power by halving the number of times the signal is split, by reducing the distance drastically or doubling the transmitted power.
Some PON architectures have reduced the upstream data rate to permit lower-powered transmitters at homes, and some have limited the number of splits for the same reason. The object is to use lower-power, low-cost transmitters at subscribers’ homes.
Mode-Partition Noise
Particularly in the upstream direction but also in the downstream direction, it is important to use low-cost lasers. The lowest-cost lasers in common use today are F-P (Fabry-Perot) lasers, though VCSELs are showing great promise. F-P lasers are multimode lasers, which produce light over a band of wavelengths. The amount of power at each wavelength tends to vary randomly due to the reflective structure in the laser. Without modulation, the total power is reasonably constant, but the wavelength at which that power is emitted changes. This phenomenon is known as mode instability. By itself, mode instability is harmless, but when mode instability exists and the light passes through a fiber-optic cable having dispersion, the result is mode-partition noise (MPN), a limiting factor in transmission distance. Dispersion is discussed in Section 12.4.3, and MPN is discussed in Section 12.6.2. Dispersion is the characteristic of optical fiber by which the velocity of propagation is a function of the wavelength of the light.
As the light output from the laser shifts between wavelengths, the velocity at which the pulses propagate changes due to chromatic dispersion in the fiber. This means that pulses recovered at the receiver will be “smeared” in time. If the smearing becomes excessive, the pulse cannot be recovered, and the bit error rate (BER) increases. A frequent measure of the effect is to determine the amount by which received power must be increased in order to get back to the same BER as would obtain without MPN. A useful relationship is the inequality
where σBLD is an acronym for bit rate, distance (L), dispersion. The laser RMS linewidth isσ. Dispersion versus wavelength — chromatic dispersion — is described in Section 12.4.3. The inequality is dimensionless when compatible units are used for all parameters. This relation ensures that the power penalty is low. In order to achieve longer distance, you can reduce the bit rate (this is going in the wrong direction), or you can decrease the linewidth of the laser (that is, the spectrum over which the laser spreads its energy due to MPN) by choosing a suitable (more expensive) laser. Reducing the dispersion in the fiber is rarely practical. The dispersion issue involves the uncertainty of fiber zero-dispersion wavelength (1,300-1,324 nm) and the laser wavelength (including unit-to-unit variation and variation over temperature). Finally, you could shorten the distance (contrary to the idea of extending the fiber farther into the system).
Chirp
The term chirp comes from the way many birds make their chirps: Over the period of one chirp, the frequency of the pitch changes, going either up or down. In fiber optics, chirp is a change in the wavelength of the light. A directly modulated laser will change its output wavelength with changes in the laser current. This change in wavelength interacts with the chromatic dispersion characteristic of the fiber to produce deleterious effects, noted shortly.
Choice of Lasers
Except for very long distances, digital transmission can use lower-cost Fabry-Perot lasers. Due to their construction, F-P lasers tend to exhibit MPN. The total power from the laser is constant, but the wavelength of that power shifts. So long as the fiber-optic cable used exhibits near-zero dispersion at the wavelength of operation, this does not present a problem. Standard (non-dispersion-shifted) fiber exhibits zero dispersion around 1,310 nm. Without signal splitting, low-cost F-P lasers may be used up to about 10 km (1.25-Gb/s wire rate, or 1-Gb/s data rate in gigabit Ethernet systems). Higher-powered F-P lasers are capable of covering distances up to 40 km; above that, distributed feedback (DFB) lasers can be used.
Crosstalk
Crosstalk can occur in optical systems due to several mechanisms. In wavelength division multiplex (WDM) systems, the same fiber is used to transmit more than one wavelength. Several technologies are available to build components to separate the wavelengths, but all have loss, adding to the loss budget of the link. In addition, the attenuation of the undesired wavelength is limited, so there is a possibility of interference between wavelengths at the receiver. The interference also extracts a penalty in the minimum received power that will provide adequate BER. This is one form of crosstalk that can limit distance.
Another form of crosstalk occurs when the same wavelength is used for both upstream and downstream communications on the same cable. It is possible to separate signals flowing in the two directions, and the isolation between them is typically better than that obtained in RF transmission. However, it is not infinite, so there can be some interference between the near-end transmitted and received signals, again reducing the sensitivity of the receiver. In short-distance transmission, there is frequently adequate margin and cost advantages to favor the use of simultaneous transmission in both directions on one fiber strand. In longer-distance transmission, two fibers are usually used to eliminate this cross talk penalty.
Digital Transmission Distance Limitation
You can purchase digital transceivers specified for different distances, with 10, 40, and 70 km being common for gigabit Ethernet systems. These distances assume no splitting loss (gigabit Ethernet is a point-to-point system). With splitting loss, each two-way split costs nominally 3.5 dB of signal. At 1310 nm, the loss of the fiber is 0.35dB/km, so each split costs 3.5/0.35, or 10km of distance. Add 2 dB for connector loss, as we did in the analog transmission case. This costs 2/0.35, or nearly 6 km of distance.
Table 19.3 shows that splitting the signal reduces the distance you can cover very rapidly. Each two-way split costs 10 km in distance covered, and we are not allowing for worst-case losses in the splitters. Higher split ratios are desirable, because higher split ratios correspond to fewer fibers coming into the hub or headend. Each fiber that returns to the hub or head end must be managed and must be repaired if a cable is cut. Furthermore, each cable returning to the headend must be terminated, either at an EDFA for broadcast fibers or in a router for digital baseband systems.
Some PONs are limited to 10 km or so in order to allow for a reasonable amount of splitting. Splits of up to 64-way are common, though usually at speeds much lower than 1 Gb/s, the speed upon which Table 19.3 is based. As speed is reduced, distances and/or splits can increase. Reducing loss or increasing power by 3 dB will allow a doubling of speed or will permit an additional distance of about 8.5 km. Note from Table 19.2 that if the PON is carrying analog signals, the number of splits is constrained to 32 or fewer.
19.9.3 Standards for Fiber-Deep Systems
There is one complete standard for fiber-deep systems at the time of this writing and one activity that commenced in 2000, aimed at defining a second standard.
Full-Service Access Network (FSAN)
A standard has been developed for APONs, under the auspices of a group known as the Full-Service Access Network (FSAN). Formed in 1995 by seven telecommunications companies, it has grown to a membership of 21 carriers worldwide, including BellSouth, Bell Canada, SBC, Verizon, Qwest, NTT, SingTel, and Telstra. Its specifications have been accepted by the ITU as G.983.
FSAN provides for fiber to the home (FTTH) or business (FTTB) as well as to the curb (FTTC — typically serves fewer than 16 subscribers) and to the cabinet (FTTCab). FTTCab serves more subscribers from one central terminating point.
FSAN specifies a maximum distance from the central office to the fiber termination of 20 km. The maximum optical split is 32. Downstream data transmission is on 1490 nm at a speed of either 155.52 or 622.08 Mb/s. Upstream transmission is at 1310 nm at a speed of 155.52 Mb/s. There is no video provided for in the basic specification, G.983.1. An extension, G.983.3, adds video at a wavelength of 1550 nm. This is sometimes called a BPON (broadcast passive optical network).
802.3ah
The other layer 2 protocol commonly used in PONs is Ethernet. Systems based on Ethernet are known as EPONs. Ethernet is by far the most popular layer 2 protocol used in local area networks such as those found in most offices. Ethernet uses variable-length packets, with the packet size ranging from 64 to 1500 bytes.
As of this writing, there is an industry consortium, under the auspices of the IEEE 802.3 committee, that is defining a standard. The group is known as 802.3ah, or, informally, 802.3 EFM, for Ethernet in the first mile. Several companies developed Ethernet-based systems in advance of a standard.
19.10 Summary
Besides using frequency division multiplexed analog transmission of multiple carriers, it is possible to use digital (binary, or on/off) modulation of lasers, modulating them with a time division multiplexed baseband digital signal. Both forms of optical modulation may be used on the same fiber if different wavelengths are multiplexed. One application of such digital systems is intercity and metro data transmission, but it is becoming more common also to use it, often in conjunction with analog FDM modulation, to deliver signals via fiber directly to homes and businesses. This is known as fiber to the home (FTTH) and fiber to the business (FTTB). Some systems deliver fiber signals to the curb (FTTC). Collectively these are known as fiber-deep systems.
Fiber-deep systems can be differentiated by the layer 2 technology used for data transmission: ATM or Ethernet. They are also differentiated by whether or not they use active elements in the plant. Typically such systems deliver baseband data and modulated video signals to a box on the side of the home, which converts signals back to the more familiar electronic form. The most common data interface is Ethernet, whereas telephone interfaces are predominately analog, and broadcast television is delivered as in conventional cable plant.
1. Institute of Electrical and Electronics Engineers, Standard 802.3, 2000 Edition, section 38.2.4.