370 (2)

R. M. Devens, Our First Century or the One Hundred Great and Memorable Events in the History of Our Country During the One Hundred Years of Its Existence, C.A. Nichols & Co., 1876

Information is of considerably increased value if it is shared with others. In this information age of high technology, we understand this principle well and we routinely share great quantities of information across vast distances. The conveyance, or transmission, of information across a distance necessarily involves some form of transmission medium that supports the propagation of the signal. The selection of an appropriate physical transmission medium is critical to the successful conveyance of the information. While the medium is not the message, at least not in the telecommunications domain, it is critical to message communications, particularly if the communication mode is an interactive one.

This chapter addresses all transmission media commonly used in traditional voice, data, video, and image networks, whether analog or digital in nature. Those media fall into two distinct categories, the first of which includes all wired media, also referred to as conducted, guided, bounded, or wireline media. The second category includes all traditional wireless media, also known as radiated, unguided, free space or unbounded.

Wireless transmission systems do not make use of a physical conductor to guide or bind the signal. Therefore, they also are known as unguided or unbounded systems. Rather than relying on electrical energy, such systems generally make use of radio waves; hence the term radiated often is applied to wireless transmission. Finally, wireless systems employ electromagnetic energy in the form of radio or light waves that are transmitted and received across space. Therefore, wireless systems often are referred to as airwave systems, although spacewave is a more accurate term, as the air in the space between transmitter and receiver actually serves to weaken the signal. Microwave, satellite, cellular, and a great number of special-purpose radio systems are wireless in nature. Free Space Optics (FSO) systems are wireless systems using infrared (IR) light signals.

Each specific transmission system has certain unique properties that manifest in advantages and limitations that point to appropriate applications. The application to be supported clearly must be of primary consideration in designing a network and in selecting the most appropriate transmission medium, assuming options are available.

Public Switched Telephone Networks (PSTNs) provide raw, voice-grade bandwidth in channels of 4 kHz, with 3 kHz (300–3300 Hz) used for voice transmission and the balance used for signaling and control purposes and for guardbands for signal separation when multiple analog voice channels are multiplexed. This range of frequencies (i.e., level of bandwidth) is sufficient to support voice communications of reasonable, if not perfect, fidelity. In an electrified telecommunications cable system, the carrier frequency, or range of carrier frequencies, depends on the specific nature of the medium and the requirements of the applications supported. Twisted-pair systems, for example, can support bandwidth of 10–10 6 Hz, and coaxial cable can support signals of up to 10 6–108 Hz. Band-limiting filters commonly used in transmission systems allow only the specified passband range of frequencies to pass on and stop all others [1–3].

The Institute of Electrical and Electronics Engineers (IEEE) defines frequency as the number of complete cycles of sinusoidal variation per unit time, with the unit of time generally being that of 1 s. Plotting y = sinx, where x is expressed in radians, yields a sine wave as illustrated in Figure 2.1. [Note: From the Latin radius, a radian is a unit of plane angular measurement equivalent to the angle between two radii that enclose a section of a circle's circumference (arc) equal in length to the length of a radius. There are 2 π radians in a circle.] A complete sine wave entails a cycle as measured from a point of zero amplitude to a point of maximum positive amplitude (+A) through zero to a point of maximum negative amplitude (−A) and back to a point of zero amplitude. In an electrical network, by way of example, + A could be in the form of positive voltage (e.g., +6 V) and −A in the form of negative voltage (e.g., −6V). Alternatively, +A could be in the form a relatively high level of positive voltage (e.g., +3 V) and-A in the form of a relatively low level of positive voltage (e.g., +1.5V). The wavelength, or length of the sine wave, can be measured from peak to peak or trough to trough or between the points that cross zero amplitude in the same direction. Wavelength is expressed as the Greek letter λ (lambda)

It is worth noting at this point that signals in electrical and radio networks are described in terms of hertz (Hz). Once the frequency of the electromagnetic signals exceeds the Extremely High Frequency (EHF) of level of 300 GHz and crosses into the infrared light (IR) range of the optical spectrum, hertz no longer has relevance as either a bandwidth measurement or a signal descriptor, as the numbers are just too large and difficult to express. Rather, wavelength is used in the optical domain to describe the nature of the signal. In support of this logic, consider that the upper range of an analog voice channel is 4 kHz. Recalling that the velocity of propagation of all electromagnetic energy in a vacuum is roughly that of the speed of light, or 300,000 km/s, consider that at a frequency of 4 kHz (4000 cycles per second) each cycle is 75 km in length. Once the signal crosses into the optical spectrum, the scale changes in consideration of the preservation of human sanity. Consider that an IR optical signal in a fiber-optic cable at a commonly used wavelength of 1550 nm (0.000001550 m) has a nominal frequency of 193 THz (193,548,387,096,774 Hz). In consideration of the fact that adjacent signals can be spaced at intervals of 200 GHz, or 1.6 nm (at 1550 nm), it is fairly obvious that a lot of time and effort are saved by talking about wavelength rather than frequency (Hz) at this level:

2.2 TRANSMISSION MEDIA SELECTION CRITERIA

The selection of the most effective transmission system for a given application must be made in the context of a number of key design considerations. Such considerations include general transmission characteristics such as bandwidth and error performance, both of which affect throughput. Additionally, you must consider the allowable distance between devices as well as issues of propagation delay, security, mechanical strength, physical dimensions, and speed of deployment. Finally, and perhaps most importantly, consider local availability and cost, including cost of acquisition, deployment, Operation and Maintenance (O&M), and upgrade or replacement.

The basic transmission characteristics of a given medium are of primary importance. Those characteristics include bandwidth, error performance, and distance between network elements. These three dimensions of a transmission system, in combination, determine the effective throughput, or the amount of information you can put through, that is, send over, the system.

Bandwidth, in this context, refers to the raw amount of bandwidth, or capacity, the medium supports. Error performance refers to the number or percentage of errors introduced in the process of transmission. Distance refers to the minimum and maximum spatial separation between devices over a single link, rather than over a complete, end-to-end circuit. Clearly, the attractiveness of any given transmission system increases to the extent that you realize greater available bandwidth, fewer errors, and a greater maximum distance between various network elements such as amplifiers and repeaters.

· A four-wire ISDN BRI (Integrated Services Digital Network, Basic Rate Interface) circuit, for example, typically runs over a physical two-wire local loop that supports three channels delivering aggregate bandwidth of 144 kbps in both directions at a frequency of approximately 40 kHz. The local loop circuit can span a distance of up to 18,000 ft between the carrier Central Office (CO) and the customer premises. ISDN BRI offers excellent levels of error performance at these distances, with no requirement for repeaters. The twisted-pair local loops must be of excellent quality to support that level of performance.

· A T1 circuit traditionally runs over a four-wire twisted-pair local loop, with each pair providing bandwidth of 1.544 Mbps in one direction at a frequency of approximately 0.75 MHz. T1 error performance is excellent through the placement of regenerative repeaters approximately every 6000 ft to overcome the effects of signal attenuation at the substantially higher carrier frequency. Again, the twisted-pair local loops must be of excellent quality.

· A LAN can support transmission rates of 10 Mbps (16 MHz), 100 Mbps (100 MHz), and even 1 Gbps (600 MHz) between attached devices such as work-stations, hubs, and switches and with excellent error performance. The twisted-pair cables that support these levels of performance are of various data grades comprising multiple conductors across which the signals are split. The device separation is limited to a hundred meters in some cases and several hundred meters in others.

Also note that radio systems are naturally limited in terms of bandwidth. As the laws of physics tell us that there is only so much raw bandwidth in the frequency spectrum, various international, regional, and national regulatory bodies apportion frequency ranges on the basis of application and geography. The regulators also limit the power levels at which such systems operate to ensure that the signals do not exceed a certain geographical range or assigned radio cell and therefore do not affect licensed users in adjacent geographic areas. As a result, the licensing of frequency bands is on a site-specific or path-specific basis. Given the limited availability of radio spectrum, it is especially important that it be used with maximum effect. Digital systems are preferable for a number of reasons, including the fact that they support data compression and, thereby, are more efficient in their use of the precious resource of radio spectrum.

Further, consider that radio systems are highly sensitive to the quality of the atmosphere between the transmitter and the receiver. Dust, smoke, haze, and humidity have decidedly negative effects on signal performance. Precipitation (e.g., rain, sleet, snow, and hail) can cause substantial degradation in performance; this phe-nomenon is known as rain fade. These issues of atmospheric quality are more significant for terrestrial radio systems than for nonterrestrial satellite systems, as satellite systems largely transmit through the vacuum of space.

Father Time must be astounded at the extent to which the telephone has challenged him in a domain over which he has long held undisputed sway. The voice travels over telephone wires and cables at between 10,000 and 180,000 miles per second. By radio its speed is the same as that of light.

Telephone Almanac, American Telephone & Telegraph Company, 1937

Propagation delay refers to the length of time required for a signal to travel from transmitter to receiver across a transmission system. Factors impacting propagation delay include the distance between transmitter and receiver and the density of the medium.

The speed of the electromagnetic signal depends on the density of the medium through which it travels. All electromagnetic energy travels at roughly the speed of light, which is approximately 300,000 km/s (actually 299,792,458 m/s), or 186,000 miles per second (actually 186,282.397 miles per second), in a vacuum. As a vacuum is void of all physical matter, there is nothing to impede the signal as it travels from point to point. A signal travels at a slower rate through the atmosphere of earth, as oxygen, carbon dioxide, water molecules, smoke, dust, and other physical matter are present, acting together to increase the density of the medium and impede the progress of the signal. An electrical signal propagates (i.e., moves or transmits forward) more slowly as it travels through a copper wire. (Note: On loaded loops, i.e., where loading coils are installed, the signal can propagate as slowly as 10,000–20,000 miles per second.) Similarly, an optical signal slows as it travels through the relatively dense glass comprising an optical fiber. Table 2.2 provides comparative data on signal velocity in various media.

Geosynchronous Earth-Orbiting (GEO) satellite systems illustrate propagation delay very effectively. As the radio signals must travel approximately 22,300 miles (36,000 km) up to the satellite and the same distance on the return leg, the resulting round-trip delay is approximately 0.25 s. (Even at 186,000 miles per second, it takes a while to travel 44,600 miles.) Note that the signal mostly travels through the vacuum of space, contending with only a few miles of atmosphere—and the increased delay and signal distortion it induces—on the uplinks and downlinks. Considering the amount of time required for signal processing on board the satellite, as well as at the Earth stations, the total delay for a round-trip transmission is about 0.32 s. Therefore, the delay between signal origination and receipt of response is approximately 0.64 s, assuming an immediate response. Hence, satellite communications is considered ineffective for highly interactive real-time voice, data, and video applications.

Security, in the context of transmission systems, addresses the protection of data from interception as it transverses the network. Clearly, increasing amounts of sensitive data are being transmitted across public networks, well outside the range of physical protection on the user's premises. Therefore, security is of greater concern than ever before, and that level of concern will only heighten as commercial enterprises and nations increasingly seek to gain and protect competitive advantages and as they apply even more sophisticated means to doing so. In hearings before the U.S. Senate in May 1996, a statement revealed that 120 nations either had or were in the process of developing sophisticated computer espionage capabilities. That number undoubtedly has increased during the intervening years. Through its Echelon system, the U.S. National Security Agency (NSA) reportedly eavesdrops on approximately three billion conversations a day in defense of national security. Echelon apparently can tap any electromag netic transmission system, including fiber optics, anywhere on the globe. (Note: While the NSA's activities make most U.S. citizens feel safer from terrorism and other threats to national security, it certainly makes many feel that their privacy is violated.)

Note that airwave systems (e.g., microwave and satellite) are inherently insecure, as access to the signal is easily accomplished and virtually undetectable through an antenna properly tuned and in proximity to the signal path. It is much more difficult to physically tap a wireline circuit. Also note that digital systems offer much greater security potential than analog systems by virtue of the fact that application software can quite effectively encrypt, or encode, the data to conceal its true meaning.

Mechanical strength applies especially to wired systems. Installers must physically manipulate twisted-pair, coaxial, and fiber-opticcables while deploying and reconfiguring them. Clearly, each type of wire, fiber, and cable has certain physical limits to the maximum severity of the bend it can tolerate (bend radius) without cracking or breaking and the amount of bending and twisting (flex strength) it can tolerate. There also are limits to the amount of weight or longitudinal stress a cable or wire can support (tensile strength) without suffering deformation or breaking (break strength). Strength members improve the tensile strength of OutSide Plant (OSP) aerial cables and inside riser cables. Strength members can be either metallic or nonmetallic in nature, depending on issues such as the weight of the cable and the need for lightening protection. Aramid fiber such as Dupont's Kevlar is used in fiber-optic riser cables not only for improved strength but also for increased protection of the fragile glass fiber from physical damage.

The issue of mechanical strength also applies to airwave systems, as the reflective dishes, antennas, and other devices used in microwave, satellite, and infrared tech-nologies must be mounted securely to deal with the stresses of wind and other forces of nature. Additionally, the towers, walls, and roofs on which they typically are mounted must be properly constructed and braced in order to withstand such forces and must flex as appropriate.

The physical dimensions of a transmission system must be considered as well. Certainly, you must consider the sheer weight of a cable system. The bulk (diameter) of the cable is important, as conduit and raceway space often is at a premium. The physical dimensions of airwave systems are no less important, as the size and weight of a reflective microwave or satellite dish and mounting system (e.g., bracket and tower) may require support, particularly in locations that experience high winds.

Speed can be of the essence at times. Wired connections take some time, even under the best of circumstances and even using the most pliable wires and quickest connectors. Radio antennas may take some time to install, but once they are in place, the time required to configure and reconfigure the connections between them can take little time, if any. If fact, the antennas can even establish and maintain connections while in motion. Portability and even mobility are key wireless advantages. It is said that time is money, which leads us to consider cost.

Ultimately, financial considerations rule, and media selection is no exception. Cost considerations include acquisition, deployment, operation and maintenance (O&M), and upgrade or replacement. Without getting involved in a lengthy discussion of each cost issue at this point, it certainly is particularly worth pausing for a moment to compare the deployment costs of wireline and wireless media.

Wired transmission systems require securing legal rights-of-way and digging trenches, boring tunnels, planting poles, placing conduits and manholes, pulling and splicing cables, placing amplifiers or repeaters, and so on. Such costs, clearly, are not trivial. Wireless systems, on the other hand, require securing rights-of-way, erecting towers, mounting antennas, securing spectrum licenses, and so on. While it is difficult to make hard-and-fast generalizations, the deployment of wired systems certainly involves a set of cost issues that can be problematic. Further, wired systems tend to be more susceptible to the forces of man (e.g., cable-seeking backhoes, posthole diggers, and trains) and nature (e.g., earthquakes and floods). Whether caused by man or nature, catastrophic failures add repair costs to the equation.

2.3 TWISTED PAIR: INTRODUCTION TO TELEPHONE WIRE

Telephone Almanac, American Telephone & Telegraph Company, 1938

Metallic wires were used almost exclusively in telecommunications networks for the first 80 years, certainly until the development of microwave and satellite radio communications systems. Initially, uninsulated galvanized iron and then steel telegraph wires were used, although it soon became clear that copper was a much better choice for a number of reasons, including its malleability and electromagnetic energy conducting properties. The early metallic electrical circuits were one-wire, supporting two-way communications with each telephone connected to ground in order to complete the circuit. In 1881, John J. Carty, a young American Bell technician and one of the original telephone operators, suggested the use of a second wire to complete the circuit and, thereby, to avoid the emanation of electrical noise from the ground [7]. The first long-line copper wire telephone circuits were strung between New York and Chicago. Consisting of uninsulated hard-drawn copper conductors about as thick as a pencil, the two-wire circuit weighed 870,000 lb, filled a 22-car freight train, and cost $ 130,000 for the copper alone [8]. In certain contemporary applications, copper-covered steel, copper alloy, nickel-and/or gold-plated copper, and even aluminum metallic conductors are employed. The most common form of copper wire used in communications is that of the Unshielded Twisted Pair (UTP), which has no shield, or outer conductor, to protect the signal from outside sources of electromagnetic interference.

A twisted pair (Figure 2.3) involves two copper conductors, generally solid core, although stranded wire is used occasionally in applications that require additional flex strength. Each conductor is separately insulated by a dielectric (nonconductor of direct electric current) material such as polyethylene, PolyVinyl Chloride (PVC), flouropolymer resin, Teflon, or some other low-smoke, fire-retardant sub-stance. The insulation separates the conductors so that the electrical circuit is not shorted and protects the conductors from physical damage. Twisted pair is known as a balanced medium as both conductors serve for signal transmission and reception and as each conductor carries a similar electrical signal with identical direct and return current paths. At any given point in the cable, the signals are equal in voltage to ground but opposite in polarity, which has the effect of reducing radiated energy and, therefore, reducing attenuation, which increases signal strength over a distance [1, 3].

In applications that involve multiple pairs, the lay length can become an issue. For example, 10/100Base-T Ethernet LANs make use of two pairs of a four-pair Category 5 (Cat 5) UTP cable. The 1000Base-T standard for Gigabit Ethernet (GbE) can run over four pairs in the same Cat 5 cable if the cable is of the proper type and is installed properly. 1000Base-T splits the gigabit signal into four signals, each running over a single pair at 250 Mbps. Each of the four pairs in a Cat 5 cable has a slightly different twist ratio, or twist pitch, in order minimize crosstalk, which can be caused by the unwanted coupling of signals between pairs. (Note: You undoubtedly have experienced crosstalk in voice communications as extraneous conversations intruded on yours. Aggravating in voice applications, crosstalk renders data communications difficult, if not impossible.) The difference in twist ratios results in a slightly different lay length, that is, physical length if the cable were to be untwisted and laid flat, for each pair. (Note: More twists per foot make for a longer physical path per pair foot.) This causes built-in propagation delay skew simply because it takes more time for a signal to travel a longer physical path (Figure 2.4). As too much delay skew will cause transmission errors because of timing differences between the signals spread across the various pairs, some cable manufacturers use foamed insulation, rather than solid insulation, on the conductors.

Gauge is a measure of the diameter of the conductor. The greater is the diameter of the wire, the less the resistance, the stronger the signal over a given distance, and the better the performance of the medium. Thicker wires also offer the advantage of greater break strength.

American Wire Gauge (AWG), originally known as Brown and Sharp (B&S) Gauge, is the standard measurement of gauge in the United States for all metals other than iron and steel. The gauge numbers are retrogressive; in other words, the larger the number, the thinner the conductor. The AWG number indicates the approximate number of wires that, laid side by side, span 1 in. (Historically, the AWG number indicated the number of times during the manufacturing process that the copper wire was drawn through the wire machine, with each draw involving a die of slightly smaller diameter in order to reduce the diameter of the wire a bit more. The contemporary process involves many fewer draws.) As an example, a 24-gauge (AWG) wire with a diameter of 0.0201 in. (0.511 mm) has a weight of 1.22 lb/kft (1.82 kg/km), maximum break strength of 12.69 lb (5.756 kg), and DC resistance ohms of 25.7/kft (84.2/km). Twisted pairs commonly employed in telco networks vary from 19 to 28 gauge, with the most common being 24 gauge. England uses Imperial Standard Wire Gauge, also known as British Standard Gauge, which also is retrogressive. Many other countries use Metric Gauge, which expresses the gauge as 10 times the diameter of the wire in millimeters. Still other countries (e.g., Namibia) simply express wire gauge in terms of kilograms per kilometer of bare wire. Table 2.3 provides an abbreviated comparison of the various UTP categories [9].

Category 3 was developed in support of 10Base-T Ethernet LANs, which run at 10 Mbps. Its excellent performance characteristics, coupled with its relatively low cost, have led to its application in OSP applications as well. Telephone companies in the United States commonly use Cat 3 cable for local loops in support of ISDN, T1, and even POTS. Cat 3 UTP has three to four twists per foot.

Category 6 cabling specifications include UTP, STP and Screened Twisted Pair (ScTP) rated at 250 MHz over distances up to 220 m, although some manufacturers boast of performance up to 400 MHz. Cat 6 is intended to support GbE, spreading the signals over each of four pairs. Applications include 100Base-T, ATM, and 1000Base-T for GbE applications.

Category 7 specifies STP with a combination foil and braided screen construction. Cat 7 supports signaling rates up to 600 MHz, although the usable spectrum can be up to 750 MHz. Cat 7 is intended to support 10GbE (10-Gigabit Ethernet), spreading the signals over each of four pairs. Due to its STP construction, Cat 7 is relatively expensive to manufacture. It also is bulky and requires larger conduits, ducts, and cable trays.

Twisted pair generally is protected by a sheath or jacket made of polyethylene, PVC, Teflon, or some other insulating material that protects the conductors from physical damage and eases the installation process. A cable generally comprises multiple pairs, which is not only cost effective but also provides a measure of redundancy, as some wire pairs invariably suffer damage during the installation process and others may fail over time due to rodent damage or other unpleasant circumstances.

Larger numbers of pairs are bundled into larger cables to serve departments, quadrants of a building, or floors of a high-rise office building. Such cables may contain 25, 50, 100, 250, or 500 or more pairs. As appropriate, the cables and pairs interconnect at cross-connect points. (Note: As the cables get farther from the switch or router, for example, and closer to the end user, the pair counts tend to get smaller.) In large cables, pairs combine into binder groups of 25 pairs for ease of connectivity management. Each binder group is wrapped (bound) with some sort of plastic tape to separate it from other groups. Each pair within a binder group is uniquely color coded for further ease of connectivity management. (Note: This is oversimplified, of course. Cable and wire management is a notoriously difficult busi-ness and one which is largely lacking in standardization.)

Inside wires and cables are of several types, according to the various national and regional standards such as the National Electrical Code (NEC) in the United States, with the construction of the cables depending on their application. Plenum cables, for example, are intended for use in plenums, or air-handling spaces, such as those between walls, under floor structures, and above drop (false) ceilings. While plenums are convenient places to run cables, they also are conducive to the spreading of fires within buildings. Therefore, the NEC specifies that the insulation on plenum cables must be fire retardant, low smoke, and low toxicity. Riser cables, intended for use between floors of a building, also must be fire retardant. Note that these categories apply equally to all cables, including fiber-optic cables.

While twisted-pair cables of up to 3600 pairs are still used in outside plant (OSP) applications, fiber-optic cables largely have replaced them in contemporary networks. Categories of OSP cables include the following:

· Overhead cables hang from poles.

· Direct burial cables lie directly in trenches dug in the ground.

· Indirect burial cables lie in ducts or conduits placed in trenches dug in the ground.

· Submarine cables are underwater, perhaps miles deep.

The effective capacity of twisted-pair cable depends on several factors, including the gauge of the conductor, the frequency of the signal, the nature of the dielectric insulating material, the length of the circuit, and the spacing of the amplifiers/repeat-ers. One must also recognize that a high-bandwidth (high-frequency) signal will not only attenuate relatively quickly but also may cause interference with other transmissions on other pairs in proximity. This issue of crosstalk is particularly sensitive in data applications.

While a voice-grade channel is guaranteed at 4 kHz, standard copper circuits can support much greater bandwidth. A single twisted pair in a typical telephone installation may provide up to 250 kHz, or 1–4 Mbps compressed, assuming amplifier or repeater spacing every 2–3 km [1]. Additional examples follow:

· T1 connections provide bandwidth of 1.544 Mbps at approximately 0.75 MHz. T1s are routinely provisioned over specially conditioned, four-wire twisted-pair cable, with repeaters spaced at approximately 6000 ft.

· Category 5e copper, in a LAN environment, provides bandwidth of 100+ Mbps at a signaling rate of 100+ MHz over twisted-pair cable at distances of up to 100+m.

· Asymmetric Digital Subscriber Loop (ADSL) simultaneously supports as much as 6.144 Mbps in the downstream direction, a bidirectional channel supporting as much as 608 kbps, and an analog voice channel over a single conditioned physical two-wire twisted-pair local loop at distances up to 2 miles. Standards-based ADSL specifications also define considerably higher speeds, although over shorter distances.

Signal quality is always important, especially relative to data transmission. Twisted-pair cable is especially susceptible to the impacts of outside interference, as the lightly insulated wires act as antennas and, thereby, absorb such errant signals. Potential sources of ElectroMagnetic Interference (EMI) and Radio Frequency Interference (RFI) include electric motors, radio transmissions, and fluorescent light boxes. As the carrier frequency of a transmission increases, the error performance of copper degrades significantly, with signal attenuation increasing approximately as the square root of frequency. Further, a high-frequency transmission radiates a strong electromagnetic field, which is absorbed by adjacent pairs in a multipair cable and which affects their error performance.

UTP is perhaps the most distance limited of all the media options. As distance between network elements increases, attenuation (signal loss) increases, and error performance degrades at a given frequency. Even low-speed (voice-grade) analog voice transmissions require amplifiers spaced at least every 2–4 miles or so. As a result, local loops generally are 10,000–18,000 ft in length. As bandwidth increases, the carrier frequency increases, attenuation becomes more of an issue, and amplifiers/repeaters must be spaced more closely. T1 transmission, for example, requires repeaters spaced at intervals of approximately 5000–6000 ft. (Remember that there are 5280 ft in a mile and about 3281 ft in a kilometer.)

UTP is an inherently insecure transmission medium. While it is relatively simple to place a physical tap on a UTP circuit, it also is fairly simple to detect the presence of a tap. Through the use of an antenna or inductive coil, you can easily intercept the signal without the placement of a physical tap, as so much of the signal travels in an electromagnetic field around the conductor. With the proper wiretap technology in place, proximity is not as much of an issue as one might think. As is the case with any transmission over any medium, encryption is the best protection.

The acquisition, deployment, and rearrangement costs of UTP are very low, at least in inside wire applications involving only a few pairs (e.g., between a terminal and a switch or hub). In high-capacity, long-distance applications (e.g., interoffice trunking), however, the relative cost is very high due to the requirements for trenching or boring, placement of conduits or poles, and frequent splicing of large, multipair cables that tend to be relatively short in length. (Note: There is only so much cable of a given size and weight that is manageable. The bigger and heavier the cable, the shorter the length of cable a construction crew can put on a truck and pull through a conduit or hang from a pole.) Additionally, there are finite limits to the capacity and other performance characteristics of UTP, regardless of the inventiveness of technologists—hence the popularity of alternatives such as microwave and fiber-optic cable.

Copper has applications other than telecommunications, of course. As the price of scrap copper approximately doubled in 2004 to reach historic highs, UTP cables became attractive targets for theft all over the world. During the first two months of 2006, thieves cut down aerial telephone cables in Kent, Washington (United States), putting hundreds of customers out of service. The Tucson, Arizona (United States), city council is considering an ordinance that would require junk dealers to give police the identities of people who sell scrap metal. While teaching a seminar on telecommunications in Windhoek, Namibia, in early 2006, I read that Telecom Namibia is offering cash rewards for citizens who report thefts of copper telephone cables, with the size of the rewards amounting to thousands of dollars, depending on the amount of cabling stolen. I know that the problem also exists in South Africa, where I teach seminars several times a year.

Unfortunately, UTP also has application in arts and crafts. Several years ago I noticed a gaily colored basket in a hotel gift shop in Johannesburg. The colors were familiar and the heft was considerable. The tag on the basket explained that Zulu warriors in times past whiled away their spare time by weaving these mbenge from native grasses dyed with the juices of native herbs. Actually bowls rather than baskets, the mbenge were covers for clay ukhamba (Zulu sorghum beer pots), serving to keep insects and dust out of the beer while still allowing the beer to ferment and breathe. As the Zulus increasingly moved to the cities in search of work, many men became night watchmen, in keeping with the warrior tradition. To while away the long hours while watching over the assets of their employers, the watchmen often wove mbenge from materials at hand, including scrap telephone wire. Well, that certainly explained the familiarity of the colors and heft. (Note: The insulation surrounding UTP cables is color coded in binder groups of 25 pairs to assist in proper splicing. The insulation colors are blue, orange, green, brown, slate, white, red, black, yellow, and violet and are presented in both solid colors and stripes of various color combinations.) It certainly did not explain the origins of the scrap wire, however. While I cringe at the thought of being in possession of stolen property, I just had to buy one, which now adds a dash of color to my living room. I hate to admit it, but I just bought another small telewire basket during that same seminar tour in early 2006. [Note: Conduct a Web search for mbenge and you, too, will find the opportunity to purchase an mbenge made of scrap telephone wire. Mbenge made of more traditional (and reliably legitimate) materials also are available.]

2.4 SHIELDED COPPER

Shielded copper offers the advantage of enhanced performance through reduction of emission of energy from the subject conductors and reduction of interference from ambient sources of electromagnetic energy such as electric motors, radio systems, and adjacent cables and wires. The shield absorbs ambient energy and conducts it to ground through the drain wire, thereby protecting the signal transmitted through the center conductor. The shield also serves to confine the electromagnetic field associated with the transmitted signal within the core conductors, thereby reducing signal loss and maintaining signal strength over a longer distance. This reduction of emissions also provides additional security and minimizes the potential for causing interference in adjacent pairs or cables.

ScTP and STP also have several disadvantages. First, the raw cost of acquisition is greater because manufacturing costs are higher. Second, the cost of deployment is greater because the additional bulk and weight of the shield and extra insulation increase the difficulty of installation. (Note: The insulation used in shielded copper cable systems is significantly thicker than that of UTP of similar gauge and resistance. This is due to the effect of the shield on the electromagnetic fields of the transmitted signal. The closer the shield is to the core conductors, the greater its effect. Hence, the dielectric insulation must be thicker to increase the separation between them.)

Also, the electrical grounding of the shield requires more time and effort during the installation process. As the continuity of the shield must be protected from end to end, either significant flexing or a severe bend radius can compromise the integrity of the shield.

2.5 COAXIAL CABLE

Coaxial cable (Figure 2.6) is a very robust shielded copper cable. The center conductor is much thicker than a twisted-pair conductor (e.g., 20 AWG versus 24 AWG) and is surrounded by an outer shield/conductor that serves to greatly improve signal strength and integrity. A layer of dielectric material, either foam or solid, generally separates the two conductors. The entire cable is then protected by another layer of dielectric material, such as PVC or Teflon. The two conductors share a common axis, hence the term coaxial. Invented by AT&T Bell Telephone Laboratories in 1934, the first coaxial cables were hollow tubes about one-quarter inch in diameter. Down the center of each pipe ran a copper wire held in place by insulating discs. The pipes were in pairs—one for transmission in each direction. The first coaxial system was placed into service in New York City in 1936. Such a cable was used in New York City to televise the 1940 Philadelphia Republican National Convention at which Wendell Wilke was nominated for president of the United States [27]. While Wilke was unsuccessful in his bid for the presidency, the coaxial cable proved to be popular. By the early 1940s, coaxial cable in commercial service could carry 500–600 telephone channels. By the late 1950s, frequency division multiplexers supported some 1800 conversations over each pair of coax tubes. Specific types of coaxial cables often are referred to by RG number (e.g., RG − 6, RG − 8, and RG − 58). The terminology was established by the United States military in the 1930s, with RG referring to Radio Guide, as the Radio Frequency (RF) signal is guided down the center conductor of the cable system. The RG num-bering system does not really have any special significance; rather, each RG number is just a page in a book, so to speak. Each RG number does, however, specify the impedance, the core conductor gauge (AWG), and the Outside Diameter (OD) of the cable.

Regardless of the specifics of the configuration, the center conductor(s) carries the carrier signal. The outer conductor generally is used only for electrical grounding and is maintained at 0 volts. Therefore, coax is described as an electrically unbalanced medium. A balun (balanced/un balanced) connector is used to connect (balanced) twisted-pair and (unbalanced) coax cables.

The gauge of the cable is much thicker than a twisted pair. The heavier gauge increases the available bandwidth and increases the distance of transmission as the thicker wire offers less resistance to high-frequency signals. However, the more complex construction of coax also increases the cost of acquisition. Further, coax is more expensive to install as its weight and mass make it harder to pull. It also is difficult to splice and connect, and it must be properly grounded. Traditional coax, such as that used in the original Ethernet LANs, is quite thick, heavy, bulky, rigid, and altogether difficult to manipulate. The subsequent Ethernet generation, sometimes known as CheaperNet, made use of ThinNet coax of smaller dimensions, but at the expense of performance.

The effective capacity of coax cable depends on several factors, including the gauge of the center conductor, the nature of the dielectric insulation, the length of the circuit, and the spacing of amplifiers and other intermediate devices. Because the bandwidth available over coax is very significant in comparison to twisted pair, it often was used in high-capacity applications, such as data and image transmission. Note the following examples of coax standards for classic Ethernet LANs; 10Base5 involves a much more substantial cable, with a thicker center conductor than 10Base2:

· 10Base2: 10Mbps; Baseband (single channel); 200m (180m, rounded up) maximum link length

The 10GBase-CX4 standard (February 2004) for 10GbE yields a specification for twinax assemblies operating over distances up to 50 ft. Each twinax cable will transmit at a rate of 2.5 Gbps in simplex mode (i.e., in one direction) at a frequency of 3.125GHz.

Coax is easily capable of supporting data rates of 100 Mbps over much longer distances and with better error performance than twisted pair. In Community Antenna TeleVision (CATV) and other applications, coax routinely supports transmission of multiple channels at an aggregate rate of 500–750 MHz. The CATV trunk cables commonly are made of very thick copper-clad aluminum, and the drops to the premises are much thinner solid-core copper.

Coax offers excellent error performance due to the outer shielding. As a result, coax was used extensively in classic data applications, for example, mainframe to Front-End Processor (FEP) to cluster controller to terminal. But the end-to-end integrity of the shield is crucial; either physical damage or poor splicing will result in awful error performance. The shield also must be grounded properly.

Coax does not have the same distance limitations as either UTP or STP, because the thicker center conductor offers less resistance to the signal. Further, the outer shield helps confine the signal to the inner conductor, which maintains signal strength. The outer shield also serves to make the signal riding over the center conductor much less sensitive to ambient noise, which is especially important as the signal weakens over long distances. Nonetheless, amplifiers or other intermediate devices must be placed at appropriate intervals to extend high-frequency transmissions over significant distances.

Coax is inherently quite secure as it is relatively difficult to place physical taps on coax without detection. (Note: The original Ethernet taps were designed to be easy, but LANs operate in secure office environments.) Little energy is radiated through the outer shield, so radio antennas and inductive coils are of little use in gaining access to the raw signal.

The acquisition, deployment, and rearrangement costs of coax are very high compared with UTP due to increased bulk and weight as well as the requirement for grounding the outer shield. In certain high-capacity data applications, however, its positive performance characteristics can outweigh that cost.

Coax is used extensively in CATV networks due to its overall performance characteristics. Coax-based CATV networks commonly provide bandwidth of either 330 MHz in support of up to 40 channels or 750 MHz in support of up to 116 channels. Such analog, one-way downstream networks support each analog TV signal over a 6-MHz channel, with the channels frequency division multiplexed at the head end, or point of signal origin. The signals are demultiplexed at the set-top box, or converter. Much coax remains in place in CATV applications, although it is being replaced with fiber optics in the backbone, with the existing coax used for the last leg of the connection to the premise. Such a hybrid network will support one-way entertainment TV, two-way Internet access, two-way voice, and other applications.

2.6 MICROWAVE RADIO

The medium through which electrical and magnetic forces act is called "ether." With the theories as to its composition or construction we are not here concerned. It is, however, by virtue of this medium which fills all space that electromagnetic disturbances … are made manifest at a distance. Of the periodic disturbances thus transmitted light and heat are two classes. The periodic disturbance made use of in wireless telegraphy form a third class.

John Mills, Radio Communication, McGraw-Hill Book Company, 1917

Microwave radio, a form of radio transmission that uses ultrahigh frequencies, developed out of experiments with radar (radio detecting and ranging) during the period preceding World War II. The first primitive systems, used in military applications in the European and Pacific theaters of war, could handle up to 2400 voice conversations over five channels. Developed by Harold T. Friis and his associates at AT&T Bell Telephone Laboratories, the first public demonstration was conducted between the West Street laboratory and Neshanic, New Jersey, in October 1945. Construction began on the first experimental microwave telephone network in 1947 [7].

The requirement to so tightly focus the signal clearly limits microwave to application as a point-to-point, rather than a broadcast, transmission system. Additionally, microwave is a Line-Of-Sight (LOS) technology as such high-frequency radio waves will not pass through solid objects of any significance (e.g., buildings, mountains, or airplanes). Actually, line-of-sight is not quite enough, as the signal naturally disperses (i.e., spreads out) in a conical pattern. As a result, portions of the signal reflect off of bodies of water, buildings, and other solid objects and can interfere with the primary signal through a phenomenon known as multipath fading. The impact of multipath fading is that multiple copies of the signal reach the receiving antenna at different levels of strength at slightly different times and slightly out of phase, thereby confusing the receiver and distorting the signal much like the ghosting effect that can be so aggravating at times to broadcast television viewers. So, additional clearance is required in the form of a Fresnel ellipse, an elliptical zone that surrounds the direct microwave path. In consideration of LOS and Fresnel zone clearance, antenna positioning and tower height are important considerations in microwave path selection and network design. Clearly, so to speak, antennas atop tall towers positioned on the roofs of tall buildings and the peaks of high mountains tend to provide optimum signal paths. Figure 2.8 illustrates a multihop microwave configuration with consideration given to Fresnel zone clearance.

At the lowest microwave frequencies, attenuation is low enough that the horizon becomes a major consideration, as the curvature of the earth limits LOS. In this scenario, it is necessary to consider the difference between optical LOS and radio LOS. True optical LOS is a straight line between the two antennas. Radio LOS can be somewhat longer as the density gradient in the atmosphere acts like a lens and tends to bend radio beams back toward the earth, as illustrated in Figure 2.9 [11].

Microwave clearly is distance limited, especially at the higher frequencies (see Table 2.4). As a point-to-point LOS radio system, design considerations include topography, antenna height, climate, and curvature of the earth. Distance limitations can be mitigated through larger antennas and antenna arrays incorporating spatial diversity and frequency diversity to increase the quality of the collected signal.

As is the case with all radio systems, microwave is inherently insecure. A radio antenna tuned to the proper frequency range and positioned in proximity to the microwave path can easily capture the raw signal. Security is imposed through signal encryption (scrambling).

The acquisition, deployment, and rearrangement costs of microwave can be high. But these costs often compare very favorably with those of cabled systems, which require extensive right-of-way procurement processes, trenching and pole placement, conduit systems, splicing, and so on. Additionally, microwave is not affected by backhoe fade, as are cabled systems.

Microwave historically has been used extensively for long-haul voice and data communications. Competing long-distance carriers, first in the United States, found microwave a most attractive alternative to cabled systems, due to the relatively high speed and low cost of deployment. Where technically and economically feasible, however, fiber-optic technology currently is used in most long-haul applications. Contemporary microwave applications include long-haul carrier networks, private networks, carrier bypass, disaster recovery, interconnection of cellular radio switches, and WLL. Microwave certainly is an excellent alternative to cabled systems where terrain is challenging. In nations where regulatory authorities have liberalized telecommunications, emerging competitors find microwave to be an excellent means for deploying competing networks quickly and at low cost, particularly in WLL applications.

2.7 SATELLITE RADIO

Energy … must obviously have some medium for its action—or, if we like to call it so, for its conveyance…. Wave-motion … of any kind requires a medium in which to propagate the wave…. The Ether, as this universal medium is called, must be unimaginably rare and subtle…. We can neither see it, nor feel it, nor weigh it; its necessary properties seem in their contradiction to mock all our faculties; yet we must suppose it, or something like it, or all our physical theories fail us, and our known facts fall to the ground as a confused mass.

The Wonders of The Universe, The Werner Company, 1899

Satellite radio is a microwave transmission system utilizing nonterrestrial relay stations positioned in space, where there is no atmosphere and certainly no ether. The concept initially was offered in an article entitled "Extra-Terrestrial Relays: Can Rocket Stations Give World-wide Radio Coverage" published in Wireless World in February 1945 by Arthur C. Clarke, then a physicist at the British Interplanetary Society. Clarke is better known as the author of the famous short story, The Sentinel, which served as the basis for the movie 2001: A Space Odyssey, which he cowrote with Stanley Kubrick. The launch of the Earlybird I satellite in 1965 proved the effectiveness of the concept of satellite communications. AT&T Bell Telephone Laboratories subsequently launched Telstar I, the first active commercial communications satellite, in 1962. Since that time, satellites have proved invaluable in extending the reach of voice, data, and video communications around the globe and into the most remote regions of the world. Exotic applications such as the Global Positioning System (GPS) would have been unthinkable without the benefit of satellites [12]. The approximately 200 GEO satellites in operation clearly speak to the popularity of communications satellites. (Note: There are approximately 2200 active satellites of various kinds in orbit at any given time.) In a subsequent article entitled "A Short Pre-History of Comsats, Or: How I Lost a Billion Dollars in My Spare Time," Clarke lamented his failure to patent the concept [13].

A GEO can see roughly one-third of the earth's surface from its vantage point; it cannot effectively transmit to such a wide area. Therefore, the downlink transmission is focused on a particular footprint, or area of coverage, which might be as wide as an entire continent. Spot beams, even more tightly focused downlinks, serve spe-cific applications over smaller regions. The popularity of satellite communications has placed great demands on the limited number of GEO slots, which are spaced at intervals of approximately 1.5 °–2.0 °. Competition for orbital slots, coupled with the development of new applications (e.g., mobile voice and data) and the technologies to address them, has given rise to new generations of microwave satellite platforms known as Middle-Earth Orbiting (MEO) and Low-Earth Orbiting (LEO) systems. MEO systems operate at altitudes of 10,062–20,940 km and LEO systems operate at altitudes of 644–2415 km. At these lower altitudes, the satellites circle the earth rather than remaining in a fixed position above and relative to it. Further, the satel-lites operate along multiple paths, most of which are nonequatorial. Also at these lower altitudes, the satellites have much more restricted views of the earth's surface and, therefore, considerably smaller footprints, as illustrated in Figure 2.11.

Satellite radio signals theoretically can propagate infinite distances in the vacuum of space with no signal loss if the radio beam can be perfectly focused. (Note: Practically speaking, that is not possible.) Such high-frequency signals, however, can suffer considerably from attenuation in the few miles of atmosphere on the uplink and downlink segments, with the higher frequency bands suffering the most. The uplink and downlink signals run at fairly widely separated frequencies in order to avoid the potential for self-interference between incoming and outgoing signals. As is suggested in Table 2.6, the higher of the two frequencies is used for the uplink as the increased signal attenuation at the higher frequency can be overcome through the application of higher radiated signal power and as power is much more readily available and much less expensive on the surface of the earth than on a satellite that depends on solar power. At a much lower power level, the lower downlink frequency can better penetrate the earth's atmosphere and electromagnetic field, which can act to bend the incoming signal much as light bends when entering a pool of water.

In order to maximize the strength of such a high-frequency signal as well as to direct the uplink transmission to a specific satellite, the uplink radio beams are tightly focused. As is the case with terrestrial microwave, the transmit antenna is centered in a concave, reflective dish that serves to focus the radio beam with maximum effect on the receiving satellite antenna. Table 2.6 provides select uplink and downlink frequency bands and dish sizes. Although the higher frequencies are more susceptible to signal fading through the atmosphere, they offer the advantages of lesser interference and smaller dishes.

Although a GEO system can see roughly one-third of the earth's surface from its vantage point, it cannot effectively transmit to such a wide area. Without shape or focus, the signal would be so weak, particularly at the fringes of the coverage area, that it would be unusable. Therefore, the downlink transmission is focused on a particular footprint, or area of coverage, which might be as wide as an entire continent. Spot beams, even more tightly focused downlinks, serve specific applications over smaller regions. Spot beams are heavily used in Ka-band satellites, as the downlink frequency is so high that a tightly shaped beam is required to overcome the effects of atmospheric attenuation. Spot beams offer the additional advantage of frequency reuse, as a given frequency can be used again in a nonadjacent foot-print, much as a frequency can be reused in nonadjacent cells of a cellular telephony network.

More recently developed satellites can serve a mesh network requirement, whereby each terrestrial site can communicate directly with any other site through the satellite relay station. Previously, all such communications were required to travel through a centralized site, known as a head end. Such a mesh network, of course, imposes an additional level of difficulty on the network in terms of managing the flow and direction of traffic.

The space segment antennas are mounted on a satellite, of course. The satellite can support multiple transmit/receive antennas, depending on the various frequencies that it employs to support various applications and on whether it covers an entire footprint or divides the footprint into smaller areas of coverage through the use of more tightly focused spot beams such as those required in the Ka-band. A satellite repeater, or transponder (transmit and res pond), accepts the weak incoming signals, boosts them, shifts them from the uplink to the downlink frequencies, and transmits them to the earth stations in what is known as a bent-pipe network configuration. Contemporary satellites commonly support as many as 28–46 transponders. Current generations of broadband satellites are replacing the relatively dumb bent-pipe transponder approach with onboard processing capability in the form of circuit switching and even fast packet switching and statistical time division multiplexing. These satellites variously support I P, Frame Relay, and ATM traffic.

Very Small Aperture Terminals (VSATs) are terrestrial dishes of very small diameter, or aperture. Note: Very small is relative, as typical VSATs are only as small as 0.9, 1.2, 1.8, and 2.4m (approximately 3–8 ft) in diameter, with the specific dish size being sensitive to the placement of the antenna within the satellite footprint. The smallest dishes work well in the center of the footprint, where the signal is strongest. As antenna placements creep farther from the center and closer to the fringes of the footprint contour, larger dishes are required to collect more signal and thereby improve the quality of reception.

Satellites can support multiple transponders and, therefore, substantial bandwidth. Contemporary GEO systems commonly support aggregate bandwidth in the range of 1 GHz per beam, yielding raw bandwidth of 1.344 Gbps, through 28 transponders, each with a channel capacity of 36 MHz [16]. A few GEO systems are based on transponders with channel capacities of 72 MHz, yielding data rates up to 155 Mbps. The level of bandwidth, the number of beams and frequency bands, the number of transponders, and the footprint all influence the size and power requirements of a given satellite.

As in the case of other transmission systems, the higher frequency bands offer greater raw bandwidth. In this sense, the C-band is the most limited, while the Ka-band is the most attractive of the commercial satellite frequency bands. As a point of reference, Intelsat I could accommodate only 240 voice circuits, while Intelsat VI supported 120,000 voice circuits and three TV channels, with total bandwidth of 3.46GHz [17].

Satellite transmission is susceptible to environmental interference, particularly at frequencies above 20 GHz. Sunspots and other types of electromagnetic interference can have considerable impact on microwave transmission in general and satellite transmission in particular. Error performance also is sensitive to the proximity of the earth stations to the equator, as those nearest the poles and farthest from the equator must deal with more signal absorption and environmental interference as the signal must travel diagonally through more atmosphere. Error performance also is sensitive to the physical location of the receiving antenna within the footprint, as the signal is strongest in the center of the footprint and weakest at the edges. C-band satellite transmission also must deal with competition from terrestrial microwave through highly directional antennas and spot beams. As a result of these several factors, satellite transmission requires rather extensive error detection and correction capabilities [18]. Throughput typically is far less than raw bandwidth, as error control requires either retransmissions or forward error correction, with the latter approach involving the embedding of redundant data in the transmitted data stream.

By virtue of their high orbital altitude, GEO satellites impose rather significant propagation delay on the signal and, therefore, doubly affect response time (see Figure 2.13). Given the fact that the radio signals must travel approximately 22,300 miles or more up to the satellite and the same distance on the return leg, the signal propagation delay is about 250 ms (0.25 s). Considering the amount of time required for processing on board the satellite, as well as at the earth stations, the total delay for a one-way transmission is about 320 ms (0.32 s). Therefore, the delay between signal origination (transmission) and receipt of response is about 640 ms (0.64 s), assuming an immediate response requiring only a single satellite hop. Note that the exact level of propagation delay is sensitive to the proximity of the earth stations to the equator.

As a result of such severe propagation delay, highly interactive voice, data, and video applications are not effectively supported via two-way GEO satellite communications. Although satellites commonly support interactive videoconferencing, it generally is accomplished in broadcast lecture mode, with the video signal sent from the central location to the satellite, which broadcasts it to multiple receive-only earth stations. As illustrated in Figure 2.13, the participants at the distant locations interact with the lecturer via landline connections. This hybrid approach mitigates the impact of propagation delay, which would create an intolerable situation if the conference made use of satellite links in both directions. This approach also reduces the overall cost, as receive-only earth stations are much less expensive than transmit/receive stations. Interactive Internet access also is offered via satellite by DBS providers through implementation of this same hybrid approach, known as telco return.

As is the case with all microwave and other radio systems, satellite transmission is inherently insecure. Satellite transmission is especially vulnerable to interception, as the signal is broadcast over the entire area of the footprint. Therefore, the unauthorized user must know only the satellite location and the associated frequency range in order to gain access to the raw signal. Only through the use of an encryption or scrambling mechanism can any level of security be implemented.

National, regional, and international bodies more or less carefully regulate the space segment of satellite communications. Additionally, local zoning ordinances and health and safety regulations may affect the placement of terrestrial antennas. While satellite network technology is widely available in North America and other countries, it is not widely available to end users worldwide. Many countries in Asia, Europe, and Africa have rejected open-sky policies, in support of the incumbent carriers. Developing countries, particularly, depend heavily on the imbalance of trade in telecommunications services as a source of hard currency. In fact, many of them intentionally price originating traffic at very high levels in order to suppress outbound traffic and, therefore, to suppress the amount of hard currency outflow. As private satellite transmission bypasses the national network, it has a negative impact on both the incumbent carrier's revenue stream and the trade imbalance (depending on one's perspective) and, therefore, is strongly discouraged.

Although traditional international voice and data services have been supplanted, to a considerable extent, by submarine fiber-optic cable systems, satellite applications are many and are increasing rapidly. Traditional, and still viable, applications include international voice and data, remote voice and data (e.g., island nations, isolated areas, and sparsely populated areas), television and radio broadcast, maritime navigation, videoconferencing, transaction processing, inventory management and control, and paging. More recent and emerging applications include air navigation, GPSs, mobile voice and data (LEOs and MEOs), Advanced Traffic Management Systems (ATMSs), DBS TV, ISDN, and Internet access. Satellite communications recently have proved to be invaluable in disaster recovery, as evidenced by their extensive use in the aftermath of the tsunami that devastated Southeast Asia (2004), Hurricane Katrina along the Gulf Coast of the United States (2005), and the earthquakes in Pakistan (2005). Terrestrial systems, including microwave and cellular systems, were either totally destroyed or knocked out of service for long periods of time as a result of these natural disasters.

2.8 FREE SPACE OPTICS

Alexander Graham Bell originally developed the concept of using light waves for communications. In early 1880, Bell invented and experimented with the photo-phone, a system utilizing mirrors to focus modulated sunlight onto a selenium cell. He was successful in transmitting voice over a distance of 700 ft on sunny days and was granted four patents for the invention. While the technique was clearly impractical, Bell nonetheless felt the invention to be his greatest achievement. During World War II, the Nazi military experimented with similar but more advanced systems in tank warfare applications, but the technology remained impractical [7].

FSO transmitters use laser light sources in the form of either Vertical-Cavity Surface-Emitting Lasers (VCSELs) or laser diodes. VCSELs generally operate at a slower speed but are less expensive and less sensitive to environmental damage than laser diodes and require active heating and cooling to maintain signal intensity. FSO VCSELs generally operate in the range of 780–850 nm, while laser diodes operate in the 1550-nm range. As signal power can be higher and attenuation is lower at 1550 nm, the distances can be greater between transmitters and receivers. Also, the 1550-nm range is considered to be eye-safe, even at the higher power levels. The optimization process of balancing cost and performance has led most manufacturers to focus on systems in the range of 800–850 nm, however.

It was clear that flexible glass fibers offered much more potential as a transmission medium for light. As early as the 1950s, the efforts of the American Optical Corporation resulted in optical fiber cable that could carry light signals a few feet. It was at Standard Telecommunications Laboratories in 1966 that Charles Kao and George Hockham developed the first practical conceptual breakthrough—the purity of the glass was the issue. During the early 1970s, the first practical fiber-optic systems were developed. These systems were made possible by the manufacturing of glass fibers pure enough to permit the transmission of light over long distances with little signal loss. Donald Keck, Bob Maurer, and Peter Schultz of Corning, continued that work with fused silica for both the core and the cladding, adding controlled levels of impurities to the core to make its refractive index slightly higher than the cladding [21]. At roughly the same time, AT&T Bell Telephone Laboratories invented laser diodes, which serve as light sources in high-speed optical transmitters. Since then, fiber-optic development has progressed to the point that virtually all high-speed networks are based on fiber-optic technology.

Generally speaking, the higher the transmission window (i.e., the longer the wavelength and the lower the frequency), the less the signal attenuation and the less the signal degradation but the more expensive the associated electronics [22].

Fiber-optic systems consist of light sources, fibers, and light detectors, as depicted in Figure 2.15. In a simple configuration comprising a single link, one of each is used. In a more complex configuration over longer distances, many links are involved. Much like any other transmission system, optical signals require boosting at various intervals in order to overcome the effects of attenuation. Traditionally, this signal boosting was accomplished by regenerative repeaters, which essentially are back-to-back detectors and sources. Repeaters accept the incoming optical signal, convert it to electrical format, reamplify it, retime it, regenerate it, and reconvert it back to optical format. Various types of optical amplifiers increasingly are now employed over long-haul links.

· Speed is directly related to the cycle time, or rise and fall times, that is, the length of time it takes for the light source to cycle through a rise to its peak and a fall to its trough of signal intensity (power). Fiber-optic systems use Amplitude Modulation (AM), so the faster the light source can cycle, the higher the bit rate. (Note: Most fiber-optic systems are digital, although a few are analog in nature.) Light sources never completely turn off, as that would limit their speed. Lasers are the fastest light sources, followed by VCSELs and LEDs.

· Power refers to the optical power, or light intensity. A higher power signal can survive more attenuation and, therefore, can survive over a longer distance without requiring amplification. Lasers are the most powerful light sources and surface-emitting LEDs are the least powerful.

· Coupling efficiency refers to the efficiency with which the light source connects to the fiber. The more precisely the light source can inject a tightly focused signal directly into the inner core of a fiber, the stronger the resulting signal and the better the signal performs over a distance. Coupling efficiency is a key advantage of pairing DFB lasers with single-mode fibers, which have an inner core of only 5–10μm.

· Directionality is the extent to which the light beam is lined up with the fiber core. Collimated light beams are lined up in perfect parallel. Divergent light beams spread out as they exit the source. LEDs emit the most divergent light beams, and laser diodes emit the least divergent.

· Spectral width refers to the range of wavelengths, or window, emitted by the light source. LEDs emit signals of the greatest spectral width, which limits their application. Lasers emit very narrowly defined signals that may be only 1 nm wide or less. Therefore, lasers have application in long-haul systems that multiplex numerous wavelengths.

· Coherence describes light signals that are synchronized in phase, with the sine waves rising and falling in unison. Too much coherence is a bad thing, as perfectly coherent light waves can interfere with each other and cause speckling, that is, cause the signal to appear grainy, if they travel slightly different paths, or modes. Coherence also means the source has a narrow range of wavelengths, ideally only one, so that all photons act identically, which is a good thing.

· Cost is always a major consideration. The essence of optimization is balancing cost and performance, and optimization is the goal of any organization geared toward profitability.

Laser diodes generally resemble LEDs in structure, although they are much more difficult and expensive to manufacture. They also are associated with more expensive and complex supporting electronics which require careful control of ambient temperature. But they generally are much faster and, therefore, offer much more bandwidth. Diode lasers offer significant mechanical and optical coupling efficiency. In other words, they can mechanically couple to a very thin Single-Mode Fiber (SMF) and can tightly focus a high-speed optical signal for presentation to its smaller axis, or core. Diode lasers also are capable of generating tightly defined optical signals in very small spectral ranges, or windows. Diode lasers also generate signals at wavelengths longer than 850 nm. In these higher transmission windows (see Table 2.7), the signals attenuate much less and, therefore, can travel much farther without being repeated or amplified. In long-haul carrier-class transmission systems, this tight definition allows the multiplexing of a number of wavelengths through a process known as Wavelength Division Multiplexing (WDM). Dense WDM (DWDM) systems are particularly intense, multiplexing lambdas closely around a centerpoint of 1552.52 nm (193.1 THz). The spacing between carriers is implementation dependent and includes options of 200 GHz (1.6 nm at 1550 nm), 100 GHz, 50 GHz, and even 25 GHz. Issues of modulation noise and channel isolation currently limit implementations to spacing of 100 GHz. In some relatively short haul carrier-class systems, a similar, but less intense, technique known as Coarse WDM (CWDM) is employed. CWDM standards specify 18 wavelengths in the range of 1270–1610 nm, with spacing of 20 nm (2500 GHz at 1550 nm) and targeted at networks with a reach of 50 km or less. All of these WDM techniques essentially are Frequency Division Multiplexing (FDM) at the optical level. As each frequency window is added, the bandwidth of the system increases. A system operating at 2.5 Gbps, for example, enjoys a bandwidth increase of 2.5 Gbps as each DWDM wavelength is added. As diode lasers enjoy a high level of coupling efficiency, they can be used in conjunction with either MMF or SMF. Because diode lasers are the more capable light sources, they generally are matched with the more capable SMF and Avalanche PhotoDiode (APD) technologies.

While plastic fibers are used in some specialized, low-bandwidth, short-haul applications (e.g., automobiles, airplanes, televisions, and stereo equipment), glass predominates. Generally speaking, fiber cables contain a large number of fiber strands because the incremental cost of redundancy is relatively low. Oftentimes, only a few of the fibers are active, with the remaining ones left dark for backup or future use. Major carriers deploy cable systems with as many as 620 fiber strands along a single route, with each fiber supporting as many as eight lambdas (i.e., wavelengths) through DWDM, with each lambda operating at speeds as high as 10Gbps. While current technology can support two-way transmission over a single fiber, two fibers generally are used, with one transmitting in each direction.

The light pulse is intended to travel through the center core of the pure glass fiber. Surrounding the inner core is a layer of even purer glass cladding. The difference in purity or clarity of cladding yields a slightly different Index Of Refraction (IOR) than the glass that comprises the core. The IOR (n) is calculated as

where c is the velocity of light in a vacuum, which is a constant value of 186,000 miles per second, or 300,000km/s, and v is the velocity of light in the subject medium, which is a variable sensitive to the chemical composition of the medium.

This scenario so far assumes that the fiber lies along a perfectly straight line, which is never the case, of course. In fact, fibers snake around corners and excess fiber is coiled in anticipation of subsequent splicing requirements to resolve future cable breaks. So, light rays injected into the very center of the fiber have no chance of remaining centered from end to end. So, some level of modal dispersion is assured and some level of pulse dispersion is a certainty. As a result, transmission speeds are kept relatively low and link lengths are kept relatively short.

The phenomena of modal dispersion and pulse dispersion in a multimode step-index cable are analogous to a convoy of very large white and black automobiles (white representing the presence of light pulses, or 1 bits, and black representing the absence of light pulses, or 0 bits) traveling at a speed of 50 mph (50 Mbps) and tailgating (small separation, or short bit time) each other down a five-lane (62.5-μm) interstate highway (glass fiber). While each of the automobiles begins in the center lane, some of them drift from shoulder to shoulder, thereby traveling a longer distance from point to point. Not only does the convoy lose its shape over a long (100-m) distance, but also collisions occur frequently.

Now consider a system involving a much faster VCSEL operating in the 1300-nm region and coupled to a graded-index MMF with an inner core of 50 μm. The VCSEL emits a pulse that is less divergent than that of the LED and the inner fiber core is considerably smaller in diameter. Some rays of light are injected into the core at angles and enter the cladding, and some find their way into the cladding through the normal course of propagation as modal dispersion occurs at bends in the fiber. Some enter the cladding where the angle of incidence changes due to bends in the cable as it snakes its way around corners. Those errant light rays that stray away from the center of the core do not encounter a core–cladding boundary with a sharp step in IOR. Instead, they encounter very gradual changes in IOR. Rather than either reflecting off the core–cladding boundary or being lost in the cladding, the light rays gradually increase in speed and bend slightly as they enter glass regions of lesser density. The sharper the angle, the greater the increase in speed and the greater the bend. Ultimately, the effect is that the errant light rays gain speed and bend back toward the fiber axis to rejoin the light rays that traveled directly through the center of the core, and the pulse largely regains its shape, as illustrated in Figure 2.18. So, modal dispersion is less of a concern and pulse dispersion is less of an issue even given the faster bit rate and, therefore, shorter bit times associated with VCSELs. The level of attenuation is less as the wavelength is longer, that is, the frequency is lower, at 1300 nm, so the link length can be increased as well.

Continuing with the convoy analogy, pulse dispersion in a multimode graded-index fiber is like small black and white automobiles traveling at 622 mph (622 Mbps) and really tailgating (even smaller separation, or shorter bit time) each other down a four-lane (50-μm) interstate highway (glass fiber). While each of the automobiles begins in the center lane, some of them drift from shoulder to shoulder, thereby traveling a longer distance from point to point. However, the narrower highway is banked (graded) like an oval racetrack, so that the force of gravity (IOR) causes the cars to turn back onto the center portion of the track (core axis) and increase speed. The errant automobiles resume their positions in the convoy, which maintains its shape over a longer distance (2 km), with fewer collisions.

Using the same analogy of a convoy of automobiles traveling down a highway, the SMF strand is a 5-μm-wide single lane. Even at 10,000 mph (10,000 Mbps or 10 Gbps) and over a very long distance (200 km), the convoy maintains its shape, with resulting collisions unlikely. There are several types of SMF: Non-Dispersion-Shifted Fiber (NDSF), Dispersion-Shifted Fiber (DSF) and Zero-Water-Peak Fiber (ZWPF). (Warning: The following discussion of SMF nuances is a bit detailed. If your interest in SMF is casual, you might consider skipping over the next few points. That said, proceed at your own risk!)

Dispersion-shifted fiber shifts the optimal dispersion point by adjusting the interface between the core and the cladding. There are two types of DSF. Zero-Dispersion-Shifted Fiber (ZDSF) shifts the point of zero dispersion by increasing material dispersion to the point that it cancels out chromatic dispersion at 1550 nm, rather than at 1310 nm. Dense wavelength division multiplexers and Erbium-Doped Fiber Amplifiers (EDFAs) both work in this higher window, which can create yet another noise problem in the form of Four-Wave Mixing (FWM), a phenomenon by which wavelengths interact to create additional wavelengths. The EDFAs amplify those signals, and superimpose them on the DWDM channels. Non-Zero-Dispersion-shifted Fiber (NZDF) addresses this issue by shifting the optimal dispersion point slightly above the range in which EDFAs operate. A small but finite amount of dispersion remains, which actually helps by providing a means of separating wavelength channels [21] and [22]. Most contemporary SMF is NZDF DSF. (Try introducing that sentence into normal conversation at the next neighborhood barbecue, and see if you are ever invited back.)

Zero-water-peak fiber is designed to eliminate issues of attenuation caused by water ions that are residuals of chemical reactions in the manufacturing process or humidity in the environment. This water peak causes attenuation of wavelengths and pulse broadening in the general regions of 950, 1380, and 2730 nm. ZWPFs resolve water peak issue in the 1380 nm (1383 nm) region thereby opening the entire spectrum from 1260 to 1625 nm for high-performance optical transmission. ZWPF is the contemporary industry standard for all SMF.

There are exceptions to every generalization, and Plastic Optical Fiber (POF) is the exception to the rule of glass in the domain of fiber-optic transmission systems. There certainly is no debate about the fact that the performance characteristics of Glass Optical Fiber (GOF) are far superior to those of plastic, but glass tends to be expensive both to acquire and to install. Also, glass is highly susceptible to catastrophic failure, as it can be so easily broken. Therefore, optical fiber generally is limited to applications where it can be protected from physical damage; even then, redundant fibers often are deployed, which further increases costs. These issues of cost and fragility generally have worked against optical fiber to the desktop and in favor of UTP and even wireless technologies. Recent developments in POF, however, may change that fact. In favor of POF are its ability to withstand extremes of temperature (−40 to +85°C) and its ability to withstand a bend radius of down to 25 mm with no break or damage. Running at a wavelength of 650 nm, POF can support data rates of up to 400 Mbps over distances up to 100m, which compares very favorably to standardized Cat 5 and Cat 5e UTP, as well as the developing Cat 6 and Cat 7 standards. While its speed rating does not compare favorably with GOF, the ATM Forum has approved POF as a viable medium for use in 155-Mbps horizontal links up to 50m in length, and the IEEE included POF in the 1394b FireWire standard, also for links up to 50m. Outside of what one would consider to be the normal data networking domain, POF enjoys considerable popularity in applications such as automobile and airplane wiring, where distances are short, temperatures can be extreme, and bandwidth requirements are fairly modest [26]. Also, POF is commonly used in contemporary stereo and television equipment.

Optical fibers are organized into cords and cables which can be indoor or outdoor in nature. Inside fiber systems generally involve one or a very few fibers. As is the case with UTP, inside wires and cables are of several types, according to the various national and regional standards such as the National Electrical Code (NEC) in the United States. Plenum cables are intended for use in plenums, or air-handling spaces, such as those between walls, under floor structures, and above drop (false) ceilings. As plenums are conducive to the spreading of fires within buildings, the NEC specifies that the insulation on plenum cables must be fire retardant, low smoke, and low toxicity. Riser cables, intended for use between floors of a building, also must be fire retardant.

In applications such as Fiber-To-The-Neighborhood (FTTN), six acrylate-coated fibers commonly are contained in a loose-tube buffer filled with water-blocking gel. There may be a number of stranded loose-tube buffers, with the loose-tube design allowing the tube, strength member, armoring, cable sheath, and other elements of the cable to expand and contract independently and, thereby, protect the fiber from damage. (Note: Tight buffered fibers are used in indoor applications, where temperature variations are more modest.) In more bandwidth-intensive long-haul applications, 6 or 12 fibers commonly are contained in a flat ribbon. There may be 12 or more such ribbons in a single cable.

Strength members are used to improve the tensile strength of outside plant (OSP) aerial cables and inside riser cables. Strength members can be either metallic or nonmetallic in nature, depending on issues such as the weight of the cable and the need for lightening protection. Aramid fiber such as Dupont's Kevlar are used in fiber-optic riser cables not only for improved strength but also for increased protection of the fragile glass fiber from physical damage.

Detectors consist of several basic types of photodiodes, the most common being PINs and APDs. The light detectors serve to reverse the process accomplished by the light sources, converting optical energy back into elec-trical energy, that is, converting photons to electrons. (Note: A diode is a device that allows a charge carrier, and thereby an electric current, to move in only one direction. A diode is analogous to a one-way valve, or check valve, that allows a liquid to flow in only one direction.)

Positive Intrinsic Negative (PIN) diodes comprise three layers of semiconducting material in the forward current-carrying direction. The first layer is chemically doped, that is, infused, to create a positive (p) electromagnetic region. The second layer is either undoped or lightly doped to retain its intrinsic (i) properties and, therefore, is neither strongly positive nor strongly negative. The third layer is doped to create a negative (n) electromagnetic region. So, the diode structure is Positive, I ntrinsic, and N egative, or PIN. A PIN generates a single electron from each photon received and therefore does not provide a significant gain or increase in signal strength. However, PIN diodes are fairly rugged and inexpensive. PIN diodes generally are matched with LED and VCSEL light sources and multimode fibers.

Much like any other transmission system, optical signals require boosting over a distance in order to overcome the effects of attenuation. Traditionally, this signal boosting was accomplished by regenerative repeaters, although various types of amplifiers increasingly are now employed.

Optical repeaters, or regenerators, are optoelectric devices. On the incoming side of the repeater, a light detector receives the optical signal, converts it into an electrical signal, boosts it and adjusts for noise, retimes it, and converts it into an optical signal that it injects into a fiber. This process actually is Optical–Electrical–Optical (OEO). Optical repeaters essentially are fiber terminals (light detectors/sources) mounted back to back, with an intermediate electrical element connecting them. There may be many repeaters in a long-haul fiber-optic transmission system, although typically far fewer than would be required in a terrestrial system based on other transmission media. In the last few years, technology has developed to the point that the maximum spacing between repeater sites has increased from 600 km or so to as much as 5000 km. However, an OEO regenerator can repeat only a single wavelength. If multiple wavelengths are multiplexed in a WDM system, they must be demultiplexed, each wavelength must be regenerated independently, and they must then be remultiplexed before being sent on their way.

Raman amplification requires no fiber doping and usually is accomplished throughout the length of the transmission fiber itself in a configuration known as distributed amplification, rather than in a discrete amplification, or lumped amplification configuration such as that employed by EDFAs. Raman amplification occurs as a high-energy (i.e., high-frequency, short-wavelength) pump wavelength is sent in the reverse direction (i.e., in the direction opposite the signal transmission) from the output end of the fiber span, where the incoming signal is weakest. The pump wavelength, which generally is in the 1450-nm range (E-band), interacts with atoms in the crystalline lattice of the fiber core. The atoms absorb the photons and quickly release photons with energy equal to the original photon plus or minus the atomic vibration. In other words, a frequency/wavelength shift occurs as the pump wavelength propagates along the fiber in the reverse direction. The energy lost in the pump wavelength is shifted to longer wavelength (within about 100-nm) signals in the forward direction, thereby serving to amplify them. Raman amplifiers offer the advantage of amplifying signals in the broad range extending from 1300 to 1700 nm. Further, they perform better than EDFAs in terms of Signal-to-Noise Ratio (SNR). Raman amplifiers increasingly are used as preamplifiers to enhance the performance of EDFAs in DWDM systems [27–34].

The concept and effect of Raman amplification is explained very effectively in the following analogy: As you are crossing an old wooden bridge at a walk, you feel the normal slow oscillation of the bridge under the weight of each of your own steps. Then a much larger person is running toward you from the other end of the bridge. The vibrations, or oscillations, you create are waveforms, and they move in both directions. The same is true of the waveforms created by the runner. But those created by the runner are stronger (he is heavier) and longer (he has a longer stride). Your waveforms and those of the runner meet in the middle. Some of the energy from the runner's waveforms is transferred to your waveforms, which are reflected right back to you with greater strength.

Fiber-optic systems can either be analog or digital in nature, although digital is much more common. Analog systems simply vary the intensity (amplitude) of the light wave. Digital systems pulse on and off (although never totally off) to represent 1s and 0s via Amplitude Modulation (AM). Frequency Modulation (FM) is possible in optical transmission and has been demonstrated but is not considered practical as the complexity of modulating frequencies in the terahertz range is extreme. Further, FM would render WDM difficult, if not impossible, as the wavelengths are so tightly packed, with spacings of only 200, 100, or even 50 GHz. Phase-Shift Keying (PSK) also is possible but not practical due to the complexities of signal manipulation at such high frequencies. Because digital systems offer significant advantages (e.g., error performance and compression), all long-haul fiber systems used in carrier networks are digital. While CATV providers deployed significant amounts of analog fiber for purposes of cable TV delivery, they increasingly are upgrading those systems to digital technology. Clearly, digital fiber is required in a convergence scenario, where the CATV providers support not only TV but also voice and high-speed data.

As the fiber itself is dielectric, it is not susceptible to EMI/RFI; neither does it emit EMI/RFI. Assuming you have properly powered and grounded the associated repeaters and other devices, ambient interference is not an issue. The optical signal does suffer from attenuation and distortion as it transverses the fiber, although not to nearly the same extent as do other transmission systems. Attenuation is a function of a number of factors, primarily scattering and absorption, and is much less of an issue in fiber optics than in other transmission systems. As attenuation is wavelength dependent, long-haul systems focus on the 1550-nm region. Signal distortion is caused by factors including modal dispersion, polarization-mode dispersion, chromatic dispersion, and simple mechanical splices in the cable system. Error performance, depending on factors such as the compression scheme utilized, ranges between 10 -9 and 10-14, or one errored bit in every 100 trillion [37]. For all practical purposes, fiber is error free, at least in comparison to the alternative transmission systems.

Fiber-optic systems suffer from diurnal wander, as do copper and coax systems. Diurnal (daily cycle) wander is a loss of signal synchronization in digital cable systems caused by temperature variations over the course of 24 h. As the ambient temperature can vary considerably from the heat of the day to the cool of the night, the cable stretches and contracts, with the overall length of the cable changing, if only ever so slightly. As the length of the medium changes, the latency of signal propagation is affected, and the number of digital pulses effectively stored in the medium changes. As a result, the network elements (e.g., repeaters and multiplexers) can get out of synch. Diurnal wander most especially affects cables strung on poles, rather than buried under ground, as the exposure to ambient temperatures is greater and as the weight of the cable magnifies the effect.

Single-mode fiber-optic systems are routinely capable of transmitting unrepeated signals over distances well in excess of 600 miles (1000 km) through the use of Raman amplifiers and EDFAs. As a result, relatively few optical repeaters are required in a long-haul system, thereby both reducing costs and eliminating points of potential failure or performance degradation.

While the acquisition, deployment, and rearrangement costs of fiber are relatively high (approximately 130 percent the cost of Cat 5 copper) in the LAN domain, the immense bandwidth can outweigh that cost in bandwidth-intensive applications such as GbE and 10 GbE. Storage Area Networks (SANs), which are very bandwidth intensive, typically are designed around fiber-optic transmission. In the Metropolitan Area Network (MAN) and Wide Area Network (WAN) domains, fiber optics generally is the only realistic option for long-haul transmission employing wireline systems. At Gbps speeds, a single set of fibers can carry much greater volumes of digital transmissions over longer distances than alternative systems, thereby lowering the transport cost per voice conversation to a small fraction of a penny per minute. The cost of transporting a single bit, therefore, is essentially zero; for that matter, the cost of transporting a multimegabyte file is essentially zero.

While glass fiber certainly does not have either the break strength or flex strength of copper or coax, it does enjoy the same tensile strength as steel of the same diam-eter. In vertical riser cable applications (i.e., between floors), the integrity of the fiber often is protected through the use of aramid fiber (e.g., Dupont Kevlar) strength members; at some point, neither copper nor glass cables can support their own weight. When covered by a protective jacket or armor, fiber can be treated fairly roughly without damage. Note, however, that you must respect limits of bend radius, as the integrity of the data stream can suffer and the glass fiber can break under a severe bend. Additionally, fiber is more resistant to temperature extremes and corrosion than alternative cable systems. However, notably, and in consideration of the huge number of conversations supported over a typical fiber-optic cable in a WAN application, a train derailment, earthquake, or other traumatic event can have consequences of catastrophic proportions. In LAN backbone applications, the impact of a fiber failure also can be considerable, although train derailments are less likely to be the root cause. Note that while plastic optical fiber (POF) is much more forgiving in terms of flex strength and bend radius, it does not perform at nearly the same speed as does glass.

In-house PLC technologies have been used in key telephone and intercom systems for decades, although not particularly successfully. Standards for in-house BPL, a premises networking technology, however, are a relatively recent development, with HomePlug standards being the most prevalent. Loosely based on Ethernet LAN standards, in-house BPL allows any device to connect to the LAN directly through the LV electric lines (110 V at 50–60 Hz or 220 V at 50 Hz). HomePlug 1.0 supports up to 16 nodes sharing bandwidth up to a theoretical maximum of 14 Mbps. Some proprietary systems support raw signaling rates up to 85 Mbps, which comes very close to 100Base-T performance.