Occasionally, a development comes along that revolutionizes an industry. Such is the case with fiber optics. The importance of fiber optics to the world economy cannot be overstated. Today, every major industrialized center is linked to the rest of the world with fiber-optic cable, and the result is high-quality and low-cost bandwidth. Fiber serves at the building level in commercial and industrial centers, freeing them from the restrictions of copper cable and enabling broadband communications that connects entities with high-speed pipelines. Without fiber the Internet would be relegated to its origins of slow-speed connections between text-based computers. Internet, telecommuting, video conferencing, distance learning, and a host of other productivity-enhancing applications ride the fiber backbone. Fiber is the default method of linking LANs and it is a shortsighted organization that installs a campus backbone without providing fiber optics between the buildings.
The conversion to fiber in the public backbone network is complete. The bandwidth restrictions to residences are beginning to disappear as fiber moves into the distribution network. Cable companies serve their customers with hybrid fiber–coax (HFC) and LECs are replacing their copper feeder cables with fiber to neighborhood centers. Eventually, fiber will be brought to the proximity of most households and the capability of end-to-end digital connectivity will be realized, but that day is still in the future.
Fiber optics arrived at an opportune time in telecommunications history, providing unlimited bandwidth and interference-free communications in a world that was rapidly exhausting the microwave spectrum. Fiber provides such high quality that it matters not whether the endpoints of a session are next door or half a world apart. The cable is fabricated from silicon, the most abundant substance on Earth, and in terms of energy consumption the electronics are far more efficient than the technologies they replaced. Best of all, once the cable is in place, it can be expanded to many times its original capacity by upgrading the electronics.
Today carriers routinely deploy fiber with bandwidths that would have been unthinkable a few years ago. OC-192, which is roughly 10 Gbps, is common and OC-768 at 40 Gbps is beginning to appear. DWDM divides the fiber into multiple channels, each on a unique wavelength capable of carrying the bandwidth of the base fiber. Lightwave amplifiers increase the span between regenerator points and optical cross-connects make it possible to route light streams without converting them to electrical signals. This chapter discusses how these hair-thin optical waveguides are manufactured and deployed into the worldwide optical infrastructure.
LIGHTWAVE TECHNOLOGY
The use of light for communication is an idea that has been around for more than a century. Alexander Graham Bell, in the first known lightwave application, received a patent for his “Photophone” in 1880. The Photophone focused a light beam from the sun, modulated it with voice, and radiated its free space to a nearby receiver. The system reportedly worked, but free-space light radiation has several disadvantages that the devices available at that time could not overcome. Like many other ideas this one was ahead of its time. Free-space light communication
is now technically feasible if the application can tolerate occasional outages caused by fog, dust, atmospheric turbulence, and other path disruptions.
Two developments raised lightwave communication from the theoretical to the practical. The first was the laser in 1960. A laser produces an intense beam of highly collimated light, i.e., its rays travel in parallel paths. The pulses from a digital signal trigger the laser on and off at the speed of the modulating signal. The second development was the refinement of glass to the point that it was sufficiently transparent to carry an optical signal. The design target at the time was a maximum loss of 20 dB/km. The decibel is a logarithmic scale. Each 3 dB of loss cuts the signal power in half, so a reduction of 20 dB would mean that after a distance of 1 km only about 1 percent of the original signal would remain. In 1970, Corning Glass Works reached the 20-dB/km threshold and focused the attention of the world on fiber optics as the next communications technology. Today, the technology has advanced to the point that it is possible to purchase fiber with a loss as low as 0.2 dB/km.
With a laser source that is triggered on and off at high speed, the zeros and ones of a digital communication channel can be transmitted through the glass waveguide to a detector that converts the received signal from light pulses to electrical. Figure 17-1 shows the elements of a lightwave communication system. Amplifiers or regenerators are spaced at regular intervals, with the spacing dependent on the transmission loss of the fiber and the system gain at the transmission wavelength. System gain is discussed in a later section. Older systems bring the light signal down to the electrical level for regeneration, but newer systems amplify at the optical level.
A standby channel, which assumes the load when the regular channel fails, protects most lightwave systems. The two directions of transmission are normally protected separately between the digital signal input and output points. If a failure occurs, the protection equipment switches the signal to a new combination of cable, terminal equipment, and repeaters. Fiber has two protection schemes, unidirectional
path-switched ring (UPSR) or bidirectional line-switched ring (BLSR). In a UPSR configuration, the nodes send on two counter-rotating rings, but all equipment receives the signal on the same ring. If the working ring fails, the receiving equipment switches to the other fiber pair. This method provides full path redundancy, but bandwidth reuse between nodes is not possible because the spare fiber must be ready to carry the entire load. A BLSR ring allows traffic to travel between nodes in the shortest route, so bandwidth can be reused between pairs of nodes.
The advantages of lightwave accrue from the protected transmission medium of the glass fiber. The optical fiber attenuates the light signal, however, and as Figure 17-2 shows, the loss is not uniform across the spectrum. Lightwave communication can use three low-loss regions or windows. Loss disturbances labeled OH– result from hydroxyl ion absorption.
The earliest fiber-optic systems used the 850-nm window because suitable lasers were first commercially available at that wavelength. Slower speed LANs also use this window because LEDs, to operate at that wavelength, are economical. As lasers became available at 1300 nm, applications have shifted to this wavelength because of its lower loss. Single-mode fiber (SMF), discussed later, exhibits slightly lower losses in its third window at about 1550 nm. Table 17-1 lists the principal wavelength windows used in fiber transmission.
The first commercial fiber-optic system installed in 1977 operated at 45 Mbps (DS-3) with repeaters required at 4-mile (6.4-km) intervals. Current systems operate throughout the SONET/SDH range, up to and including OC-192, which has a line rate of 9.95 Gbps. If this much bandwidth were populated with 192 DS-3 signals, each of which carries 672 voice channels, an OC-192 system would carry 129,024 voice channels. As mentioned earlier, DWDM systems can multiply that capacity 40 times or more and the industry is beginning to consider the move up to OC-768.
LIGHTGUIDE CABLES
A digital signal is applied to a lightguide by pulsing the light source on and off at the bit rate of the modulating signal and the pulses propagate to the receiver at slightly less than the speed of light. The lightguide has three parts: the inner core, the outer cladding, and a protective coating around the cladding. Both the core and the cladding are of glass composition; the cladding has a greater refractive index so that most of the incoming light waves are contained within the core. Light entering an optical fiber propagates through the core in modes, which are defined as the different possible paths a lightwave can follow.
Two types of optical fiber are manufactured: single mode and multimode. In single-mode fiber, light can take only a single path through a core that measures about 9 microns in diameter, which is about the size of a bacterium. (A micron is one-millionth of a meter.) Multimode fibers (MMF) have cores of 50 to 200 microns in diameter. The original MMF standard had a 62.5-micron core diameter and it is still specified for some applications. More recently, the industry has shifted to 50-micron core because the bandwidth is higher. MMF is used almost exclusively in customer premise applications. SMF is more efficient at long distances for reasons that we discuss below, but the small core diameter requires a high degree of precision in manufacturing, splicing, and terminating the fiber. Despite the greater precision needed, SMF is less expensive than multimode, primarily because of the vast quantities of single mode manufactured.
Lightwaves must enter the fiber at a critical angle known as the angle of acceptance. Any waves entering at a greater angle can escape through the cladding as Figure 17-3 shows. The reflected waves take a longer path to the detector than those that propagate directly. The multipath reflections arriving out of phase with the main signal attenuate the signal, round, and broaden the shoulders of the light pulses. This pulse rounding is known as modal dispersion. It can be corrected only by regenerating the signal. If a light pulse spreads so far that the trailing edge of one pulse merges with the leading edge of the next, bit errors result. The greater the core diameter, the greater the amount of modal dispersion. SMF propagates only one mode of light and therefore does not suffer from modal dispersion.
Both SMF and MMF are subject to another form of dispersion called chromatic dispersion. The term chromatic comes from the multiple light wavelengths that propagate through the core. The amount of dispersion is a function of the quality of the laser. High-quality lasers emit a narrower band of wavelengths, resulting in less dispersion.
A third type of dispersion is known as polarization mode dispersion (PMD). PMD is caused by small variations in the shape of the fiber core. When light travels down a fiber, some polarization modes are at angles to each other. If the core is not perfectly symmetrical, one mode travels faster than the other, resulting in pulse spreading. Both chromatic and polarization dispersion increase as the square of the bit rate. MMF is classified by its refractive index into two general types, step index and graded index. With step index fiber the refractive index is uniform throughout the core diameter. In graded index MMF the refractive index is lower near the cladding than at the core so that lightwaves propagate at slightly lower speeds near the core than near the cladding, which result in lower dispersion. Figure 17-4 shows wave propagation through the three types of fiber.
Loss in fibers is caused by absorption and scattering. Absorption results from impurities in the glass core, imperfections in the core diameter, and the presence of hydroxyl ions or water in the core. The water losses occur most significantly at wavelengths of 1400, 1250, and 950 nm, as Figure 17-2 shows. Scattering results from variations in the density and composition of the glass material. These variations are an inherent by-product of the manufacturing process. SMF has its lowest chromatic dispersion at the 1300-nm wavelength, but minimum loss is at 1550 nm,
which has led to the development of dispersion-shifted fiber. Dispersion-shifted fiber moves the minimum chromatic dispersion wavelength to the 1550-nm window to provide the lowest combination of loss and bandwidth at the same wavelength.
Nearly all carrier applications and some customer premise applications use SMF. LANs often use the 850-nm window because LEDs, to operate at that wavelength, are economical. As lasers became available at 1300 nm, applications shifted to that wavelength because of its lower loss. SMF exhibits slightly lower losses in its third window at about 1550 nm. Although the lasers are more expensive in this window, it is used in DWDM systems because erbium-doped fiber amplifiers (EDFAs) operate in the 1550-nm window and the lower fiber loss helps offset the losses of the DWDM equipment.
Most metropolitan fiber-optic systems use 1300-nm wavelengths because the components are less expensive than in the 1550-nm window. In addition, the distance between termination points is shorter and amplification is generally not needed. In the long-haul network the extra cost of the components is offset by the lower loss of that wavelength. The 1550-nm window is divided into two bands. The most commonly used is C band, which has a wavelength of 1530 to 1565 nm. The L band, using longer wavelengths, is 1565 to 1625 nm. An additional window from 1350 to 1530 nm has not been commercially developed yet, but when components are developed for that band, the potential bandwidth of a fiber is in the order of 50 THz. Table 17-2 lists the ITU-T fiber types and their principal applications.
Manufacturing Processes
Glass fibers are made with a variety of processes: modified chemical vapor deposition (MCVD), outside vapor deposition (OVD), plasma-activated chemical vapor deposition (PCVD), and vapor-axial deposition (VAD). To illustrate, the MCVD process starts with a tube of ultra-pure silica about 6-ft long and 1.5-in diameter. The silica is doped with germanium to create the required refractive index profile. The tube is rotated over a flame of controlled temperature while a chemical vapor is introduced in one end as Figure 17-5 shows. The vapor is a carrier for chemicals that the heat deposits on the interior of the glass. The deposited chemicals form a tube composed of many layers of glass inside the original tube. The OVD process deposits high-purity glass on the outside of a ceramic rod, which is removed after the process is complete. When the deposition process is complete the tube is collapsed under heat into a solid glass rod known as a preform. The perform is placed at the top of a drawing tower where the fiber is heated to the melting point and extruded through a die into a hair-thin glass strand, as Figure 17-6 shows. The fiber is then coated with a protective substance that may be colored for ease in
identifying individual fibers in a cable.
The strands are then tested, segregated by quality, and multiple strands are wound together around a strength member and enclosed in a sheath. Like copper cable, fiber cable sheaths are made of polyethylene and can be enclosed in armor to protect against damage. Fiber-optic cable is suitable for direct burial, pulling through conduit, suspension from an aerial strand, or submersion in water. Two predominate methods are used for cabling: loose tube and tight-buffered. In loose tube construction, the fibers are placed in buffer tubes, which are usually gel-filled to keep moisture out. In tight-buffered design, the fibers are wrapped much like a copper cable. One advantage of loose tube is that it is easy to drop off fibers at branching points without affecting the other fibers in the sheath. It is also easier to identify and administer the fibers. The tight-buffered method takes less space in conduits and raceways. Loose tube is usually preferred for outdoor applications because the fibers are loosely coupled, which allows them freedom of movement during expansion and contraction and during installation.
Fiber comes from the factory with no splices. The size of the reel imposes a practical limit on the cable length, but the objective is to avoid splices by ordering the cable in reel lengths sufficient to span the normal connection points, such as central offices, manholes, and amplifier locations. Where splicing is necessary, it can be done by either adhesion or fusion. In the adhesion process, a technician places fibers in an alignment fixture and joins them with epoxy. The fusion method employs a splicing fixture that precisely aligns the two ends of the fiber under a microscope and fuses them with a short electric pulse. The transmission loss of a mechanical splice is usually higher than a fusion splice, but lower skill and less expensive equipment are needed. After splicing, the loss is measured to check the splice quality. Splices are made with enough slack in the cable that they can be respliced if necessary, until the objective loss is achieved. Agood splice should have 0.1 to 0.3 dB of loss.
Cable Connectors
In midspan, fiber cables are joined by splicing, but at terminal locations they are connectorized for coupling to terminal devices and for ease in rearrangement. The physical structure of connectors is important because of the close tolerances that are required to match the cable to the transmission device. Connectors are made from thermally stable materials and have tightly locking keyed parts and polished mating surfaces.
Connector performance is judged by two criteria: the amount of insertion loss and the amount of reflection attenuation or return loss. Areflection is light that travels down the lightguide, strikes a discontinuity, and reflects toward the source, causing instability or errors. The acceptable amount of loss depends on the application. LAN applications, which usually use multimode cable, are more tolerant of connector performance. The loss of a LAN connector can usually be as much as 1.0 dB and still remain within the loss budget. Asuitable connector for common-carrier applications should have an insertion loss of 0.5 dB or less and a return loss of at least 40 dB.
Couplers are made either free hanging or for bulkhead mounting. Connectors use an epoxy and polishing arrangement for termination. The fiber is stripped and inserted into the connector. The ends are then polished to an optical finish. The durability of the connector is important. The connector holds the ends of the fiber in alignment, and it must remain so even under the strain of pulling or sideways motion. Several different connector styles are available, but the most common is the ST type. The fiber is inserted into a precision pin that has a springloaded mechanism to press the pin against a mating connector. The connector has a strain-relieving device attached to the outer jacket of the fiber-optic cable. ST connectors are available for both MMFs and SMFs.
FIBER-OPTIC TERMINAL EQUIPMENT
A fiber link consists of a fiber pair terminated in a transmitter and receiver. The light transmitter employs either a LED or a solid-state injection laser diode as its output element. Lasers have a greater system gain than LEDs because their output is higher and because the light is more highly collimated, resulting in less dispersion. Lasers can be more tightly coupled into the fiber than LEDs, which illuminate the entire core. The transmitter may either vary its intensity between defined levels or trigger the light on and off in step with the electrical signal. The latter method is prevalent. The primary advantage of a LED transmitter is its lower cost. LANs, which do not need high system gain, usually use LED transmitters.
The light receiver is an avalanche photodiode (APD) or positive–intrinsic– negative (PIN) diode that detects the light pulses and converts them to electrical pulses, which the receiver reshapes into square wave pulses. A lightwave regenerator has back-to-back receiver and transmitter pairs that connect through a pulse reshaping circuit. Fiber-optic systems accept standard digital signals at the input, but each manufacturer develops its own output signal rate. Error checking and zero suppression bits are inserted to maintain synchronization, monitor the bit error rate, and determine when to switch to the protection channel. Because of differences in the line signals, lightwave systems are usually not end-to-end compatible between manufacturers unless they meet SONET/SDH standards, and even then OAM&P systems may not interoperate.
Optical Amplifiers
EDFAs amplify light in the C band, extending from 1530 to 1565 nm. EDFAs are independent of the modulation method, handling both analog and digital signals, whether baseband or DWDM signals. They can handle speeds well in excess of the 10-Gbps signals commonly used today and have been tested at bit rates in excess of 5 Tbps. A typical EDFA offers about 50 dB of gain over bandwidths of about 80 nm and have power outputs as high as +37 dBm or about 5 W. To increase the bandwidth of EDFAs, work is progressing on telluride-based EDFAs, which amplify in the 1565 to 1625-nm L band. L band amplifiers extend the total bandwidth of an EDFA to more than 110 nm. Extending the bandwidth of EDFAs offers significant benefits. In addition to providing more wavelength division multiplexing (WDM) channel capacity, it also enables manufacturers to use broader wavelength spacing, enabling them to use less-expensive components to separate wavelengths. To complicate the process, however, the wider the band the more important it is to provide gain-flattening filters to equalize the loss across the wavelength band.
Although EDFAs are the most common device in use today, they do not cover the 1300-nm window, which has led to the interest in amplifiers that do. One such device is the praseodymium-doped fluoride fiber amplifier (PDFFA). Another technology is the Raman fiber amplifier, which amplifies in both windows. When an intense light is injected into the fiber, the Raman Effect excites the photons to a higher energy state, making the fiber act as an amplifier. The result is to distribute amplification along the length of the fiber instead of at fixed locations.
Raman amplifiers in conjunction with EDFAs will enable the use of a broader band of wavelengths. Manufacturers expect that fiber bandwidth will be approaching the 50-THz theoretical limit by 2010.
Wavelength Division Multiplexing
The capacity of a fiber pair can be multiplied by using WDM. WDM assigns services to different light wavelengths in much the same manner as frequency division multiplexing applies multiple carriers to an analog medium. Different wavelengths or “colors” of light are selected by using light-sensitive filters to combine light wavelengths at the sending end and separate them at the receiving end. Coarse WDM (CWDM) systems apply about 16 wavelengths to a fiber. DWDM systems today use 100-GHz spacing between wavelengths, generally providing 40 channels per fiber in the C band. In future, 50- or 25-GHz spacing will likely be possible with improved filters and interleavers. Coupled with dual band amplifiers, more than 100 wavelengths or lambdas can be transmitted. The vision of optical networking is to manipulate these wavelengths in the optical domain as TDM devices manipulate electrical bandwidth today.
Because the filter introduces loss, DWDM reduces the distance between regenerators and limits the path length by the wavelength with the highest loss. When engineers design lightwave systems, they normally provide enough system gain to compensate for WDM even if it is not used initially.
A variety of filtering techniques are employed in optical networking. They are important not only for DWDM, but also for optical add–drop multiplexers as discussed in the next section. The most common filtering technique is known as fiber gratings. Variations in the refractive index of the fiber core cause discrete wavelengths to be either passed or reflected. The quality of the filters is one of the most important issues in advancing the optical network. Increasing fiber capacity will eventually mean spacing wavelengths more closely. The filter skirts must be steep to avoid interference between channels. The ultimate capacity of a fiber using a combination of high-speed multiplexing over DWDM fiber has hardly been glimpsed.
Wavelength Add–Drop Multiplexers (WADM)
The capacity of a single fiber pair is so great today that it may easily exceed the entire bandwidth needed between the points on a transcontinental network. Some means is needed to drop specific wavelengths in the same manner that a TDM add–drop multiplexer manipulates electrical bandwidth. Such a device allows some wavelengths to pass through while diverting others to an alternate path. The vacated lambda on the other side of the multiplexer can be replaced by an inserted signal.
Three types of WADM are possible. The simplest type is a fixed WADM in which certain channels are filtered out and recombined while the other channels are passed through the multiplexer. A reconfigurable WADM is the next stage of development. This device allows the provisioner to select the channels under manual or software control. The ultimate device is a flexible WADM that can respond to changes in traffic demand. This type of device could sense failures or congestion and reroute traffic accordingly. The application dividing line between a flexible WADM and an optical cross-connect is indistinct, but the technology used is different. WADMs achieve their wavelength manipulation by the use of filters. Optical cross-connects, as we will discuss next, redirect the lightwave by means of an optical switch.
Optical Cross-Connects
It should be clear from our earlier discussions that expanding the network indefinitely with SONET/SDH is not practical. For one thing, digital crossconnects have a finite capacity. When this is exceeded, carriers have no alternative but to deploy multiple DCSs. This is undesirable for several reasons, not the least of which is the cost of multiple optical–electrical–optical (OEO) conversions. As DCSs are interconnected, additional ports are used for the interconnection, reducing the effective capacity of the DCS.
DCS systems are effective when the signal is at the electrical level, but bringing the light signal down to an electrical signal at main branching points is expensive and requires floor space to house the regenerating equipment. Optical networking equipment has the objective of routing signals at the optical level wherever possible. An optical switch or cross-connect (OCX) operates like a DCS, but at the optical instead of the electrical level.
The next generation optical networks require a new architecture based on retaining SONET/SDH functions, while eliminating the equipment layer. The key to making this architecture practical is an economical and scaleable OCX. Switching in the optical domain retains the benefits of SONET/SDH while removing many of its limitations. Optical switches can provide such functions as multiplexing, provisioning, signaling, service restoration, performance monitoring, and fault management. The core of the network is optical, while the edge remains electrical with traditional interfaces as shown in Figure 17-7. The optical core of the network is OC-48 or OC-192 while the edge uses IP or ATM over SONET/SDH bandwidth building blocks.
Optical cross-connects or switches, as they are sometimes known, switch lambdas to form point-to-point connections in the optical domain, transcending the limits of the electrical domain. Optical networking enables carriers to sell wavelengths without giving up the entire capacity of the fiber. This enables carriers to offer services to their customers without being limited to the digital hierarchy. Optical networking offers the ability to route and groom in the optical domain. Carriers want to be able to pick bandwidth out of the optical domain, add to it, drop it, or shift it to another wavelength to groom wavelengths into the total capacity of a fiber.
Ultra Long-Haul Equipment
Undersea cables generally have much longer regenerator spacing than the 400- to 500-km spacing that is typical of long spans. Partly, this is due to the high quality of cable that can be justified to increase spacing; also, it is because undersea fiberoptic systems can be more easily optimized since amplifiers and regenerator locations do not have to be selected to coincide with existing structures. The same is desirable in land-based systems. Ideally, carriers would be able to route lambdas for as much as 4000 km without regeneration. The technology to accomplish this is known as ultra long haul.
Several techniques are under development to achieve these kinds of distances. One is forward error correction to detect and correct bit errors through a complex process of adding overhead bits. Other techniques include new gain-flattening filters for EDFAs allied with Raman amplification.
LIGHTWAVE SYSTEM DESIGN CRITERIA
Fiber-optic system design is a balance between capacity requirements and costs, which include the cable, terminal equipment, regenerators and amplifiers, and construction and engineering. The three primary criteria for evaluating a system are:
- information transfer rate
- system attenuation and losses
- cutoff wavelength
Information Transfer Rate
The information transfer rate of a fiber-optic system depends on the bandwidth, which in turn depends on dispersion rate and the distance between terminal or repeater points. Manufacturers quote bandwidth in graded-index fiber as a product of length and frequency. For example, a fiber specification of 1500 MHz-km could be deployed as a 150-MHz system at 10 km or a 30-MHz system at 50 km. Special purpose fiber-optic systems intended for short-range private data transmission have low bit rates and typically use cables with considerably more bandwidth
than the application requires.
System Attenuation and Losses
System gain in fiber optics is the algebraic difference between transmitter output power and receiver sensitivity. For example, a system with a transmitter output of –5 dBm and a receiver sensitivity of –40 dBm has a system gain of 35 dB. From the system gain, designers compute a loss budget, which is the amount of cable loss that can be tolerated within the available system gain. Besides cable loss, allowances must be made for other elements such as these:
-splice loss including an allowance for future maintenance splices
-connector losses
-temperature variations, which may cause variations in transmitter
output power or receiver sensitivity
-measurement inaccuracies
-current or future WDM
-aging of electronic components
-safety margin
These additional losses typically subtract about 10 to 12 dB from the span between terminal points, which leaves a loss budget of about 25 dB for cable. Cable cost depends on loss, so system designers choose a cable grade to match the loss budget.
Sample Performance Margin Calculation
Table 17-3 shows how to calculate the performance margin of a fiber-optic system. The first step is to calculate the total attenuation, which is the sum of three elements: the loss of the cable, the loss of connectors, and the loss of any splices.
The next step is to calculate system gain, which is the algebraic sum of the transmitter output power minus the receiver sensitivity. For safety’s sake, an operating margin is stated to allow for deterioration of the transmitter with component aging. If the manufacturer does not state a margin, 2.0 dB is typical for LEDs and 3.0 dB for lasers. If the manufacturer states a margin for the receiver, this should also be added. In addition, an allowance is made for future repair splices.
The performance margin is the system gain minus the cable loss, operating margin, and repair splice allowance. If the performance margin is not enough, the designer may have to specify lower loss cable or higher performance transmitter and receiver.
FIBER OPTICS IN THE LOCAL LOOP
Now that the interexchange network is converted to fiber optics, the LECs and CATV companies are turning to the next potential application, the local loop. The rationale behind fiber-in-the-loop (FITL) are broadband applications such as video-on-demand that cannot be served with conventional facilities. Most cable companies have converted their systems to HFC to support such applications as HDTV, Internet access, and IP telephony. If ILECs are to compete in this arena, they must provide fiber local loops.
Local loop fiber will likely assume one of several architectures. One is to replace feeder cable with fiber. Fiber extends from the central office to a serving area interface. The loop from the serving area interface to the customers’ premises remains copper cable, which simplifies the interface problem and resolves the issue of feeding power to the customers’ station. Cable companies use a similar architecture, extending fiber trunk cable from the headend to neighborhood nodes where the optical signal is converted to electrical and applied to coaxial distribution cable. This option is often called fiber-to-the-curb (FTTC).
The second local loop fiber optic option provides fiber direct to the customers’ homes (FTTH). In some applications, two fibers are installed, one for voice and data communication and the other for video; in other plans, a single fiber is installed, using WDM or different transmission windows to separate the directions of transmission.
Passive Optical Network (PON)
A promising method of providing FTTH is the PON. This technique places all the active equipment in the central office. Apassive signal is brought to the residence, either directly to the home or to the curbside. The same fiber could be connected to several residences, with the signals to and from the different premises multiplexed by TDMA.
APON uses passive splitters to break the capacity of the backbone fiber into multiple wavelengths. Today, as many as 32 full-duplex channels can be carried on a single fiber to a terminating point, where the channels are split and routed to the served customers over a short length of fiber. APON requires no power, which makes it practical for bringing fiber to the curb. Just as LECs bring cable to a centralized terminal and serve multiple subscribers from one terminal, a PON splitter can serve multiple subscribers over fiber without bringing individual
fibers all the way back to the central office. The optical path is independent of bit rate, modulation, and protocol.
At the central office, the fiber connects to an optical line terminal (OLT), which is a device that converts the digital TDM signal from the central office into a multiwavelength lightwave signal. The light from the OLT is launched into a fiber, where it travels to a splitter at the distant end. The splitter separates the wavelengths, which are routed to the customer’s premises over a single fiber. At the customer’s end, the PON terminates in an optical network unit (ONU). The ONU separates the signal into its components, which may be voice, Internet access, video, Ethernet, or any other digital service. Conceptually, PON architecture is similar to DSL, except that the components are optical instead of electronic. More details on PON are included in Chapter 32, “Metropolitan Area Networks.”
OPTICAL NETWORKING APPLICATION ISSUES
Lightwave communications systems have applications in both private and public communications systems. The primary LEC and IXC applications ride on SONET/SDH. For noncarrier applications SONET/SDH is also available, as are FDDI, fiber channel, and LAN protocols. The following lists representative fiber applications:
-long-haul transmission systems
-intercontinental and undersea transmission systems
-trunking between local central offices
-metropolitan area backbone systems
-digital loop carrier feeder systems
-local area networks
-cable television backbone transmission systems
-private campus backbones
-interconnection of PBXs with remote switch units
-short-haul data transmission systems through noisy environments
-Fibre Channel for high-speed computer communications
-intelligent transportation systems, such as smart highways with intelligent traffic lights, automated tollbooths, and changeable message signs
-process automation in factories and industrial plants
-diagnostic image transmission in telemedicine applications
The high cost of right-of-way often stands in the way of private fiber-optic systems, but the advantages of this medium make it attractive for private applications. A major impediment to many applications is the common carriers’ refusal to offer dark fiber. Most common carriers offer to lease bandwidth in any increment, but where the application requires dark fiber, many decline to provide it except to other carriers. Nevertheless, as companies are able to obtain right-ofway, they can install fiber for countless applications. The following sections discuss the variety of ways companies can apply fiber optics.
Campus, Intra-, and Interbuilding Backbone
Fiber optics is an excellent medium for a campus or building backbone and most LANs now employ a fiber backbone. Fiber optics not only provides bandwidth, but also offers security and noise immunity that no other medium can match. Any campus or riser cable system should at least consider the potential future need for fiber optics. Either fiber pairs should be installed for future expansion or empty conduit should be installed to support future fiber. Most current applications use MMF, but as speeds increase, single mode is becoming more common, particularly as applications such as Fibre Channel and gigabit and 10G Ethernet gain acceptance. Companies installing fiber today should consider installing both varieties in separate cables. If today’s applications call for MMF, subducts should be placed in conduit to provide a future pathway for SMF.
Fiber to the Desktop
The experts agree about using fiber as a backbone in a building or campus network, but the question of carrying it all the way to the desktop is still controversial. UTP is about the same price as fiber, has enough bandwidth for most applications, and is easier to install and apply. However, fiber has much greater bandwidth, and although the industry is working on higher category UTP, the connectors are not so mature as fiber connectors are. UTP must be installed to
every desktop and the question is whether to install fiber as well. Fiber optics is an ideal transmission medium for LANs, but the terminating equipment is twice as costly for fiber as for UTP. The fiber premium will undoubtedly decrease as the volume increases. In today’s environment, fiber to the desktop can be justified only if the application has a genuine need for high bandwidth, extended range, or if there is an overriding consideration, such as security or need for noise immunity that mandates the use of fiber.
If fiber is not feasible today, should it be installed today and left dark to support a future application? The answer to this question depends on economics. Many buildings are difficult to wire, and placing a composite fiber and UTP cable to desktops may make economic sense because so much of the cost is in installation labor. Other buildings are designed for modular furniture that may be rearranged before the fiber is even used. The question to evaluate here is whether the location of future applications can be foreseen reliably enough to justify the expense of fiber optics.
Evaluation Criteria
Fiber-optic equipment is purchased either as an integrated package of terminal equipment and cable for specialized private applications or as separate components assembled into a system for trunking between switching nodes. For the former applications, which include local area, point-to-point voice, data, and video networks, the evaluation criteria discussed below are not critical. In such systems the main question is whether the total system fits the application. In all fiber-optic systems the questions of reliability, technical support, cost, and compatibility are important. The following criteria should be considered in evaluating a system.
System Gain
In selecting lightwave-terminating equipment, the higher the system gain, the more the gain available to overcome cable and other losses. The cost of a lightwave system relates directly to the amount of system gain. High-output lasers and high-sensitivity diodes are more expensive than devices producing less system gain. The least expensive transmitters use LEDs for output and have less system gain than lasers. When the limits of lightwave range are being approached, obtaining equipment with maximum system gain is important.
Cable Characteristics
Cable is graded according to its loss and bandwidth. The cable grade should be selected to provide the loss and bandwidth needed to support the ultimate circuit requirement. For systems operating at 100 Mbps or more on MMF, bandwidth becomes the limiting factor. The cable composition should be selected with inner strength members sufficient to prevent damage when the cable is pulled through conduits or plowed in the ground. Armoring should be considered where sheath damage hazards exist.
In private applications the core size of multimode cable is an important consideration. EIA/TIA standards specify both 50/125- and 62.5/125-micron cable. (The 62/125 designation means the cable has a core of 62.5 microns and an external diameter of 125 microns.) If the application has not been selected and the cable is being placed for future applications, the safest choice is 50/125-micron cable.
Wavelength
With present technology the most feasible wavelength to choose in MMF is 1300 nm unless the equipment specifications state otherwise. FDDI specifications call for 1300-nm cable with a bandwidth of at least 500 MHz-km. Cable should be purchased with a 1550-nm window if circuit requirements will ultimately justify the use of WDM. For most applications 850 nm should be avoided because of its greater loss.
Light Source
The two choices for light source are laser and LED. A laser has much higher power output than an LED and can operate at higher bit rates. LEDs are lower cost and have a longer life, but they produce a wider beam of light and have a wider spectral width, which means that a broader range of light wavelengths are transmitted compared to a laser.
LEDs are typically used where the distance between the terminals is short; normally 10 km or less, and the bandwidth of the signal is lower than about 150 Mbps. LEDs are generally satisfactory for local networks. In long-haul networks where long repeater spacing and high bandwidth are important, lasers are always used.
Wavelength Division Multiplexing
The question of whether to plan a fiber-optics system with future WDM designed into the transmission plan is a balance between future capacity requirements and costs. WDM can multiply the capacity of a fiber pair many times for little additional cost or can convert a single optical fiber into a full-duplex mode of operation by transmitting in both directions on the same fiber. It accomplishes this by reduced regenerator spacing, however, which is important in long systems but unimportant on systems that do not require an intermediate regenerator. On very
short systems the cost of the WDM equipment may be greater than the cost of extra fibers.
No comments:
Post a Comment