Optical Networks: Capacity and Components

Date: May 1, 2002

Sample Chapter is provided courtesy of Prentice Hall Professional.

This chapter introduces the optical network. We begin with a survey of three generations of digital transport networks, followed by a discussion of the extraordinary capacity of optical fiber. The optical network marketplace is examined with a look at current and projected installations. Next, we examine the key nodes (machines) that make up the optical network, then we look inside a node to learn about its components. The chapter concludes with a general explanation of the attributes of optical fiber.

Three Generations of Digital Transport Networks

The focus of this book is on third generation digital transport networks, usually shorted to 3G, or 3rd generation, transport networks. The main characteristics of three generations of digital transport networks are provided in Table 1–1. The information in this table will be helpful as you read the remaining chapters in this book. Most of the terms in the table are self-explanatory, or, if not, are explained in this chapter.

Table 1–1 Three Generations of Digital Transport (Carrier) Networks

Name	Family	Designed for	Mux/SW Schemes at Inception	Principal Media at Inception	Capacity	Typical Payload	Protocol Inter-Working?
T1/E1	First	Voice, Non-BOD, Static	TDM/E/E/E	Copper: (Early1960s)	Mbit/s	Fixed Length	No
SONET/SDH	Second	Voice, Non-BOD, Static	TDM/O/E/O	Copper, Fiber: (Mid–1980s)	Gbit/s	Fixed Length	Somewhat: PPP, IP, ATM
OTN	Third	Voice, Video, Data, Tailored QOS, BOD, Dynamic	WDM/O/O/O	Fiber (Late1990s to Early 2000s)	Tbit/s	Fixed or Variable Lengths	Yes: PPP, IP, ATM, MPLS

The first column in the table is the name (or names) usually associated with the technology. The first generation systems are known as T1 or E1. The second generation systems are called SONET (for the Synchronous Optical Network) or SDH (for the Synchronous Digital Hierarchy). These terms are explained in more detail in later parts of this book. However, the industry has not yet settled on a handle for the third generation digital carrier network, but the term Optical Transport Network (OTN) is widely used. The second column identifies the generation family.

The third column shows what kinds of user payloads the networks are designed to support. Although the first and second generation networks are designed to support voice traffic, they can and do transport data and video images. But they are not "optimized" for data and video traffic. In contrast, the 3G transport network is designed to support voice, video, or data payloads. When used with multiprotocol label switching (MPLS), the resource reservation protocol (RSVP), and DiffServ, as well as some of the new specifications dealing with optical bandwidth on demand, they are also designed to provide tailored quality-of-service (QOS) features for individual customers. The point will be made repeatedly in this book that the 3G transport network no longer consists of fixed, static "pipes" of capacity; it can dynamically change to meet the changing requirements of its users.

The third column also contains the notations of Non-BOD or BOD. The first and second generation systems are not designed to provide bandwidth of demand (BOD). The bandwidth is configured with crafting operations at each node. 3G systems are more dynamic and allow bandwidth to be requested on demand.

The fourth column lists the predominant multiplexing schemes: TDM or WDM. The fourth column also lists the manner in which the networks switch traffic when they were first deployed (at their inception). First generation systems were solely E/E/E operations: (a) they accepted electrical signals (the first E), (b) processed them (the second E), and (c) sent them to another node (the third E). Second generation systems are O/E/O operations: (a) they accept optical signals (the first O), (b) convert them to electrical signals for processing (the E), and (c) convert the electrical signals back to optical signals for transmission (the second O). Third generation systems are intended to be all optical (O/O/O), in that they process optical payloads, and do not need to convert the bits to electrical images for processing. Today, all three generations are mainly O/E/O oriented.

The fifth column lists the principal media used by the technologies at their inception, as well as the time that these networks were first introduced into the industry. All three generations now use a combination of copper, fiber, and wireless media.

The sixth column lists the typical capacity of the generation. It is evident that each succeeding family has increased its transport capacity by orders of magnitude.

The seventh column goes hand-in-hand with the third column ("Designed For"). The first and second generation networks were designed for fixed-length voice traffic, based on the 64 kbit/s payload, with a 125-μsec clocking increment. The third generation network supports this signal, but also supports variable-length payloads, an important capability for carrying data traffic. As well, the first and second generation networks can carry variable-length traffic, but they are not very efficient in how they go about transporting variable-length data traffic.

The eighth column explains whether any of the generations were designed to interwork with and directly support other protocols. T1/E1 was not so designed; again, 1st generation transport systems were set up to support voice traffic. Any efforts to devise methods of carrying other payloads were an afterthought and in vendor-specific procedures. With the advent of 2nd generation systems with SONET/SDH, efforts were made by the standards groups to define procedures for carrying certain kinds of data traffic, and many manufacturers adapted these standards into their products.

3rd generation transport networks are geared toward supporting many kinds of payloads, and specifically the Internet, ATM, and MPLS protocol suites. As we shall see as we proceed though this book, extensive research has resulted in many specifications defining how MPLS contributes to the operations of the third generation digital (optical) transport network.

All Features Are Not Yet Available

Not all the features and attributes cited in Table 1–1 are available in 3G transport networks. In fact, third generation transport networks are just now appearing in the marketplace, and some capabilities that are touted for them are still in the lab. Nonetheless, many people think full-featured 3rd generation transport networks will be in the marketplace by around 2004. Certainly, pieces are emerging, such as bandwidth on demand, and of course, WDM and terabit networks. Other parts of 3G transport networks have yet to be implemented. For example, O/O/O operations are far from reaching commercial deployment on a mass scale.

Optical Fiber Capacity

To gain an appreciation of the transmission capacity of optical systems operating today, consider the facts in Table 1–2. Prior to the advent of optical fiber systems, a high-capacity network was capable of operating (sending and receiving traffic) at several million bits per second (Mbit/s). These electrical/electromagnetic transmissions take place over some form of metallic medium such as copper wire or coaxial cable, or over wireless systems such as microwave. In contrast, optical fiber systems transmit light signals through a glass or plastic medium. These systems are many orders of magnitude "faster" than their predecessors, with the capability of operating in the terabits-per-second (Tbit/s) range.

Table 1–2 Magnitudes and Meanings

Magnitude	Term	Initial	Meaning
1 000 000 000 000 000 000=1018	exa	E	Quintillion
1 000 000 000 000 000=1015	peta	P	Quadrillion
1 000 000 000 000=1012	tera	T	Trillion
1 000 000 000=109	giga	G	Billion
1 000 000=106	mega	M	Million
1 000=103	kilo	k	Thousand
100=102	hecto	h	Hundred
10=101	deka	da	Ten

As depicted in Figure 1–1, a terabit fiber carries 1012 bits per second. At this rate, the fiber can transport just over 35 million data connections at 28.8 kbit/s, or about 17 million digital voice channels, or just under 500,000 compressed TV channels (or combinations of these channels). Even the seasoned telecommunications professional pauses when thinking about the extraordinary capacity of optical fiber.

Figure 1–1 Capacity of one fiber with a 1 Tbit/s rate.

A logical question for a newcomer to optical networks is, why are they of much greater capacity than, say, a network built on copper wire, or coaxial cable? The answer is that optical signals used in optical networks operate in a very high position and range of the frequency spectrum, many orders of magnitude higher than electromagnetic signals. Thus, the use of the higher frequencies permits the sending of many more user payloads (voice, video, and data) onto the fiber medium.

Figure 1–2 shows the progress made in the transmission capacity of optical fiber technology since 1982 [CHRA99]. The top line represents experimental systems, and the bottom line represents commercial systems. The commercial results have lagged behind the experimental results by about six years. The dramatic growth in the experimental capacity was due to improved laboratory techniques and the progress made in dispersion management, a subject discussed later in this book. As the figure shows, the transmission capacity of optical fiber has been growing at an extraordinary rate since the inception of the technology.

Figure 1–2 Transmission capacity as a function of year [CHRA99].

A Brief Introduction to WDM and TDM

We would like to keep this introductory chapter free from technical detail as much as possible. However, it is necessary to introduce two terms before proceeding further: (a) wave division multiplexing (WDM), and (b) time division multiplexing (TDM). Later chapters will embellish on this introduction. Refer to Figure 1–3 during this discussion.

Figure 1–3 WDM and TDM.

WDM is based on a well-known concept called frequency division multiplexing or FDM. With this technology, the bandwidth of a channel (its frequency domain) is divided into multiple channels, and each channel occupies a part of the larger frequency spectrum. In WDM networks, each channel is called a wavelength. This name is used because each channel operates at a different frequency and at a different optical wavelength (and the higher the frequency, the shorter the signal's wavelength). A common shorthand notation for wavelength is the Greek symbol lambda, shown as λ.

The wavelengths on the fiber are separated by unused spectrum. This practice keeps the wavelengths separated from each other and helps prevent their interfering with each other. This idea is called channel spacing, or simply spacing. It is similar to the idea of guardbands used in electrical systems. In Figure 1–3, the small gaps between each channel represent the spacing.

Time division multiplexing (TDM) provides a user the full channel capacity but divides the channel usage into time slots. Each user is given a slot and the slots are rotated among the users. A pure TDM system cyclically scans the input signals (incoming traffic) from the multiple incoming data sources (communications links, for example). Bits, bytes, or blocks of data are separated and interleaved together into slots on a single high-speed communications line.

Combining WDM and TDM

Most optical networks (or, for that matter, most networks in general) use a combination of WDM and TDM by time-division multiplexing fixed slots onto a specific wavelength, as shown in Figure 1–4. This concept is quite valuable because it allows multiple users to share one WDM wavelength's capacity. With some exceptions, the capacity of one wavelength exceeds an individual user's traffic capacity needs.

Figure 1–4 Combining WDM and TDM.

These introductory definitions should be sufficient for us to use them in this chapter. In later chapters, TDM and WDM are examined in considerable detail.

The Optical Marketplace

The optical technology is a high-growth market. As Figure 1–5 shows, it is expected to more than double between 2000 and 2003. Most of the growth will be for terrestrial WDM and optical networks, and submarine cable systems. The growth of conventional TDM systems will continue (shown in Figure 1–5 as SONET/SDH, and explained in Chapter 5), but at a lesser rate than the other areas. Optical systems are being incorporated into CATV networks, but this growth will be modest because the coaxial cable plant to the homes cannot be re-wired in a cost-effective manner. A residence does not need the bandwidth of fiber (at least not for the foreseeable future).

Figure 1–5 Worldwide market for optical components.

Figure 1–6 shows another study comparing the projected transmission capacity and the demand through 2004 [STRI01]. The figures pertain to the national backbone in the United States and not to the access loops. This study holds that the building-out of optical networks discussed earlier will provide excess capacity for the early part of this decade, and if one examines the gap between demand and capacity, it is reasonable to expect that capacity will exceed demand well beyond 2004.

Figure 1–6 Demand vs. capacity [STRI01].

There are those in the industry who disagree. They state that the upcoming applications will require huge amounts of bandwidth, and that this supposed excess capacity will be consumed by these applications. There is no question that some applications do indeed require a lot of bandwidth. A prime example is interactive high-quality Web traffic, exhibiting the integration of high-resolution, real-time voice, video, and data.

The Local Loop Bottleneck Must Be Solved

It is my view that the upcoming applications, and their demand for large chunks of bandwidth, are not going to be realized to any large degree until bandwidth is available on the access line (the local loop) to the end user. It does little good to download a Web response to a user at a terabit rate within the network when the vast majority of access lines are restricted to V.90 speeds (56 kbit/s). Certainly, some businesses can afford to purchase large bandwidths from the business to the terabit backbones. But many businesses cannot afford to purchase this bandwidth, nor can the majority of residential users. In addition, broadband access loops are not available to most residences anyway.

The situation in the United States is interesting, and quite frustrating to many customers, because many of them are limited to very low capacity links to the Internet. What is the incentive for the local access providers (the telephone local exchange carriers (LECs), CATV operators, and wireless providers) to expand their local access plant to the megabit or terabit rate? After all, with some minor exceptions (and in spite of the 1984 and 1996 legislative efforts), these companies have a lock on their market. Some people believe that these companies do not have a lot of incentive to invest in the upgrading of their plants.

Maybe so, but the local access providers will expand their plant if they think there is sufficient demand to enable them to make money on their investment. So, beyond the issues of government-sponsored monopolies, is there really that much demand for the deployment of high-capacity systems into the mass marketplace?1 Most Internet users use the Internet for email or simple text-oriented Web retrievals, and many have been conditioned to the slow response time in their interactions with their networks.

The present situation can be illustrated with a diagram shown in Figure 1–7.2 The circle in this figure illustrates the relationships of: (a) user applications' requirements for bandwidth (labeled "Applications" in the figure), (b) the capacity of the user or network computers (labeled "CPU" in the figure), and (c) the capacity of the communications media to support traffic (labeled "Bandwidth" in the figure).

1By mass marketplace, I mean deployment into residences on a large scale, well beyond the 15–20% penetration rate for the current efforts of telephone company and the cable company.

2I call this illustration the "eternal circle," because it shows a seemingly never-ending dependency-relationship between the three components.

Historically, the bottleneck in this circle has varied. At times, it has been the lack of CPU (and memory) capacity in the user's computer. At other times, it has been the lack of capacity in the network to support the capacity requirements of the users' applications.

Figure 1–7 shows the relationships with two-way arrows, which suggests that these three operations are interrelated and dependent upon each other. But which comes first? Does the application's requirement for more capacity lead to faster CPUs and/or the expansion of network bandwidth? Or does the introduction of more bandwidth encourage the development of faster computers and more powerful applications? There are no clear answers to these questions. Sometimes one pushes the other, and at other times the opposite occurs.

Figure 1–7 The eternal circle.

However, at this time in the telecommunications industry, we can state the following:

Within the optical network (the backbone or core network) the bottleneck is the "CPU," because its electrical-based architecture (in switches, routers, and bridges) cannot handle a large number of connected optical fiber WDM links that operate in the terabit range. Thus, the creation of all-optical photonic switches (PXCs) is a high priority in the industry.
At the edge of the network, and to the end user, the bottleneck is the "bandwidth," but not because of the optical fiber. The bottleneck is due to the continued use of the telephone-based copper plant, and the mobile phone links (and the very slow process of getting it upgraded).
The "applications" part of the Eternal Circle is a question mark to some people. If the network operators finally provide the bandwidth all the way to the mass market (the residence), will sophisticated three-dimensional, voice/video/data applications be developed to take advantage of the increased capacity? I believe the answer is a resounding yes, assuming the network operators can keep the price affordable to most households.

Expansion of Network Capacity

One of the more interesting changes occurring in the long-distance carrier industry in the United States is the extraordinary growth of bandwidth capacity. This growth is occurring due to the maturation of the WDM technology, and its wide-scale deployment. It is also occurring due to the aggressive deployment of fiber networks by the "non-traditional" carriers; that is, those carriers who have come into the industry in the last few years.

Figure 1–8 shows the growth of long-distance capacity since 1996, and projections through 2001. The shaded bars show total mileage, and the white bars show total capacity, in terabits per second.

Figure 1–8 Long-distance growth in the United States.

Some people question if this bandwidth will be used. Others see it as an opportunity to discount excess capacity, at the expense of the traditional carriers, who are enjoying healthy profits from their long-distance revenues. It will be interesting to see how the scenarios develop over the next few years. Some marketing forecasts state that this situation will lead to a "fire sale" of DS1 and DS3 lines.

Wireless Optical Systems

Another entry into the optical technology is a multichannel optical wireless system from [BELL99]. Figure 1–9 shows a system that operates at 10 Gbit/s using four wavelengths. It can transmit over 2.7 miles of free space. Each wavelength operates at 2.5 Gbit/s. It requires a line-of-sight topology. Other vendors are offering this system, including Nortel Networks.

Figure 1–9 Wireless optical systems.

The systems uses WDM with custom-built telescopes, and standard optical transmitters and receivers. Light signals are sent from a transmitting telescope to a receiving telescope and are focused onto the core of an optical fiber using coupling optics within the second telescope.

The system is attractive in situations where the deployment of fiber cable is not feasible, for example, across restricted-access terrain, or bodies of water. It can be deployed much more quickly than fiber cable systems. It can also offer a cost-effective solution to line-of-sight channels in conference and convention centers.

In most countries, optical wireless requires no governmental licensing or frequency allocation schemes.

Key Optical Nodes

We now leave the subject of the optical network marketplace, and focus our attention on the nodes (machines) that comprise the network. Figure 1–10 shows the key nodes in an optical network. The topology is a ring, but the topology can be set up either as a ring, a point-to-point, multipoint, or meshed system. In most large networks, the ring is a dual ring, operating with two or more optical fibers. The structure of the dual ring topology permits the network to recover automatically from failures on the optical links and in the link/node interfaces. This is known as a self-healing ring and is explained in later chapters.

Figure 1–10 Key nodes in the optical network.

End-user devices operating on LANs and digital transport systems (such as DS1, E1, etc.) are attached to the network through a service adapter. This service adapter is also called an access node, a terminal, or a terminal multiplexer. This node is responsible for supporting the end-user interface by sending and receiving traffic from LANs, DS1, DS3, E1, ATM nodes, etc. It is really a concentrator at the sending site because it consolidates multiple user traffic into a payload envelope for transport onto the optical network. It performs a complementary, yet opposite, service at the receiving site.

The user signals, such as T1, E1, ATM cells, etc., are called tributaries. The tributaries are converted (mapped) into a standard format called the synchronous transport signal (STS), which is the basic building block of the optical multiplexing hierarchy. The STS signal is an electrical signal. The notation STS-n means that the service adapter can multiplex the STS signal into higher integer multiples of the base rate, The STS signals are converted into optical signals by the terminal adapter and are then called OC (optical carrier) signals.

The terminal/service adapter can be implemented as the end-user interface machine, or as an add-drop multiplexer (ADM). The ADM implementation multiplexes various STS input streams onto optical fiber channels. OC-n streams are demultiplexed as well as multiplexed with the ADM.

The term add-drop means that the machine can add or drop payload onto one of the fiber links. Remaining traffic that is not dropped passes straight through the multiplexer without additional processing.

The cross-connect (CS) machine usually acts as a hub in the optical network. It can not only add and drop payload, but it can also operate with different carrier rates, such as DS1, OC-n, E1, etc. The cross-connect can make two-way cross-connections between the payload and can consolidate and separate different types of payloads. For example, the cross-connect can consolidate multiple low bit-rate tributaries into higher bit-rate tributaries, and vice versa. This operation is known as grooming.

Key Terms for the Cross-connect

The convention in this book is to use three terms to describe the optical cross-connect. There is a spate of terms to describe a cross-connect. I counted six terms in one paper alone. To make sure there is no ambiguity about the optical cross-connect in this book, the following terms are used:

Optical/Electrical cross-connect (OXC): Receives optical signals, converts them to electrical signals, makes routing/switching and/or ADM decisions, then converts the electrical signals back to optical signals for transmission. These operations are also noted as O/E/O. This technique is also called an opaque operation.
Photonic cross-connect (PXC): Performs the functions of the OXC, but performs all operations on optical signals. These operations are also noted as O/O/O, and are also called transparent operations.
Cross-connect (XC): A more generic term, used when it is not necessary to distinguish between the OXC or the PXC.
Switch: Some recent literature distinguishes between a cross-connect and a switch. This literature states that cross-connect is an outdated term! Well, the term switch has also been around for quite a while. Anyway, the book uses the terms cross-connect and switch synonymously.

Other terms and different definitions of optical nodes are used by various vendors, network operators, and standards groups. In some cases, they are the same as those just cited; in other cases, they are different. Where appropriate, I will distinguish and explain these other terms.

Other Key Terms

Other terms need to be defined and clarified in order for readers to understand the other chapters in this book. For the first few times I use these terms, I will refer you back to these definitions, or repeat them. Unfortunately, the industry is not consistent in the use of some of these terms; some varying interpretations are explained below.

Fiber link set: This term refers to all the fibers (if there are more than one) connecting two adjacent XCs or other fiber nodes. The link set may consist of scores of individual optical fibers and hundreds of wavelengths.
Edge, ingress, egress nodes: These terms refer to the placement of the XCs at the boundaries of the network. The term edge encompasses both ingress and egress. Ingress obviously means the XC sending traffic into the network, and egress is the node sending traffic out of the network.
Interior, transit, or core nodes: These three terms refer to an XC that is located inside the optical network and communicates with other XCs for internal network operations or with the edge nodes for communications (perhaps) outside the network.
Optical switched path (OSP): The optical path between two adjacent optical nodes. The OSP is one logical channel of a fiber link set.
Lightpath and trail: The term lightpath defines an end-to-end optical path through one or more optical nodes or networks to the end users. This term is also used in some literature to identify the optical path between two adjacent nodes, so it must be interpreted in the context of its use. Also, some literature uses the terms lightpath and trail synonymously.
Label switched path (LSP): The end-to-end MPLS path across one or more MPLS nodes (and perhaps optical as well) networks to the end users.

Another Look at the Optical Node

Figure 1–11 shows a more detailed view of the optical network node and its components [NORT99b]. This example shows the light signal transmitted from the left side to the right side of the figure. The events for the operation are explained below, with references to the chapters in this book that provide more detailed explanations.

Figure 1–11 Optical components in more detail [NORT99b].

First, laser devices generate light pulses tuned to specific and precise wavelengths, such as 1533 or 1557 nanometers (nm). Lasers are explained in Chapter 3.
Next, the optical modulators accept the electrical signal (an incoming bit stream), and convert it to an optical signal. In addition to the conversion, the modulator uses the incoming bit stream to make decisions about turning the light stream on and off to represent the digital 1s and 0s of the incoming stream. Chapters 3 and 4 provide more information on this process.
The multiplexer (MUX) combines different TDM slots or WDM wavelengths together. Chapters 5, 6, and 7 explain multiplexing in considerable detail.
The signal is passed to an optical post-amplifier (Post-Amp). This amplifier boosts the strength of the power of the signal before it is sent onto the fiber. See Chapters 3 and 7 for more information on amplifiers.
On the fiber, a dispersion compensation unit (not shown in Figure 1–11) corrects the dispersion of the signal as it travels through the fiber. As explained in more detail in Chapters 3 and 7, dispersion is the spreading of the light pulses as they travel down the fiber, which can cause interaction (and distortion) between adjacent pulses.
As the signal travels down the fiber, it loses its strength. Therefore, the signal power is periodically boosted with an amplifier (Line Amp) to compensate for these losses, again as explained in Chapters 3 and 7.
There may be an XC on the link to switch the signals to the correct destination. The manner in which the signals are relayed through a cross-connect is one of keen interest in the industry and is examined in Chapters 8, 9, 10, 12, and 14.
At the final receiver, optical pre-amplifiers (Pre-Amp) boost the strength of the signal once again (Chapters 3 and 7).
A demultiplexer separates the multiple wavelengths (Chapters 5, 6, and 7).
Optical photodetectors convert the optical wavelengths into an electronic bit stream (Chapter 3).

Evolution of Optical Systems

To set the stage for subsequent chapters, Figure 1–12 shows the evolution of optical systems since the late 1980s/early 1990s to the present time [GILE99].

Figure 1–12 Evolution of optical systems [GILE99].

The early systems had a single channel point-to-point topology, as shown in Figure 1–12 (a). The short fiber spans did not use optical amplifier repeaters, so the span lengths rarely exceeded 40 km. The restriction of length was due partially to the limit of laser transmitter power of 1 mW. Optical systems at this time used a single fiber to increase the overall transmission capacity of the point-to-point link.

The discovery of erbium-doped fiber amplifiers (repeaters) was a major milestone in the ability to extend the fiber link. The optical amplifier allowed the optical transmission to extend to wide areas (see Figure 1–12 (b)).

Now that a link could span thousand of kilometers, it became feasible to deploy add-drop multiplexers (ADMs) to allow the connection of points along the link, shown in Figure 1–12 (c). (Of course, ADMs in optical networks have been around for over ten years, but wavelength ADMs (WADMs) are more recent.) As of this writing, the fixed-wavelength ADM (of selected channels) is the state-of-the-art implementation for WADMs. But WADMs that are dynamically reconfigurable are the next phase for WADMs.

As shown in Figure 1–12 (d), cross-connects (XCs) permit a more powerful grooming of traffic between optical networks (in this example, optical rings). This configuration has been deployed for over a decade in TDM optical systems; wavelength XCs, noted as PXCs in this book, are the next stage of the evolution.

Key Attributes of Optical Fiber

Finally, as a prelude to the remainder of the book, we conclude this chapter by reviewing the major attributes of optical fiber.

The advantages of fiber optics (compared to copper cable) include superior transmission quality and efficiency. Since the optical signal has none of the characteristics association with electrical signals, optical fiber does not suffer from common electromagnetic effects such as experiencing interference from other electrical components, such as power lines, electrical machines, and other optical links.

Because it does not emanate energy outside the fiber, the optical signal is more secure than copper and wireless media, which are easy to monitor and glean information from the residual energy emanating from these media.

Glass fiber is very small and of light weight, a significant attribute for network operators who must install communications links in buildings, ducts, and other areas that have very limited space for the communications links.

We learned earlier that fiber has a very wide bandwidth which allows for the transport of very large payloads, some in the terabits-per-second range.

Since the fiber is comprised of glass with a very small diameter, it is fragile and is somewhat difficult to connect and splice. Also, because glass is not a conductor of electrical current, it cannot carry power to the regenerators (which are used to strengthen signals on long spans). This situation is changing, as passive optical networks are deployed (a subject for Chapter 8).

Summary

Optical fiber and optical-based networks have revolutionized the world of telecommunications, principally because of their extraordinary transmission capacity. They have replaced almost all the older media, such as copper, in the large backbone networks in the world, such as the telephone networks, and the Internet. Unfortunately, due to the dominance of copper wire in the telephone local loop plant, they have not been installed (to any significant extent) in residences. Nonetheless, optical fiber technology will continue to grow, and with the advent of WDM and powerless amplifiers, their presence will become commonplace in all high-speed networks.