News | July 8, 1999

Optical Fiber Improves Position In Desktop LAN Applications

Source: Corning Cable Systems

Performance, Reliability, and Safety Considerations
Higher Reliability, Performance, and Flexibility with Fiber
The Truth About Leveraging Copper
Installation And Maintenance Considerations
High Speed Copper Links Complex to Test
Attenuation: An Accurate Test of Fiber Cable
Standardization Issues
Understanding The Cost Of Fiber
ATM and Beyond
References
Annex A
Annex B

Introduction
Optical fiber, long the transmission medium of choice for LAN backbones, is improving its position in horizontal cabling applications. While copper is still the most popular horizontal cabling medium, an increasing number of MIS managers are choosing fiber for their desktop cable and an even larger number plan to install fiber in the next three to five years. Driving the decision to install fiber is the nation's growing commitment and dependence on high-speed LANs to support business requirements. In less than a decade, we have witnessed data rates nearly double every year as companies add more users to their networks and take advantage of more bandwidth intensive applications.

Network planners specify optical fiber cable for a number of reasons, but the most common are:

  • Fiber's error-free transmission over longer distances. This results in fewer outages, less downtime and greater reliability. With longer link distances, network designers also have more flexibility in planning their networks, and are able to take advantage of new architectures.
  • Fiber's ability to support higher data rates. Fiber's high bandwidth gives it the headroom network designers need to take advantage of existing and emerging high-speed network interfaces and protocols such as FDDI (Fiber Distributed Data Interface), ATM (Asynchronous Transfer Mode), 100BASE-FX, 100VG-AnyLAN, ESCON®) (Enterprise Systems Connection), Fibre Channel and other foreseeable gigabit LAN applications, including the next generations of switched Ethernet, FDDI and Fibre Channel.
  • Fiber's ease of handling, installing, and testing. Fiber is now a proven and established medium that can be easily tested for performance at all data rates. Typically fiber requires only attenuation tests, since bandwidth is unaffected by installation practices.
  • Fiber's long term economic benefits. Over the lifetime of the network, optical fiber is typically a more economically viable choice than copper. For example, fiber's superior reliability reduces operating costs by minimizing network outages. Similarly, fiber's higher bandwidth can produce considerable savings by eliminating the need to pull new cable when the network is upgraded to support higher bandwidth applications. Also, fiber's high bandwidth and long distance capability allow all hub electronics to be centrally located, rather than distributed in closets throughout a building. Centralization reduces the cost of cabling and electronics, and reduces administration and maintenance efforts by making moves, adds and changes easier.

Performance, Reliability, and Safety Considerations
Optical fiber provides users with higher reliability, superior performance and greater flexibility than copper-based systems. The construction of optical fiber makes it essentially immune to many of the factors that adversely impact copper, factors that often become more pronounced at higher data rates, thereby increasing network cost and complexity.

Fiber is immune to EMI/RFI signals. Because optical fiber carries light rather than electricity, it cannot be affected by electromagnetic interference from power (sub-kilohertz), radio (kilohertz to megahertz), or microwave (gigahertz) sources. Optical frequencies are more than 1,000 times higher than microwaves, so are completely out of the range of these signals. However, these interfering signals can couple into copper cables and make it difficult for the receiver to differentiate between the interference and valid data. This type of problem, which typically happens sporadically, is difficult to troubleshoot and repair. Copper-based systems must be carefully designed and installed to minimize this interference.

Fiber is immune to cross talk. Cross talk occurs when unwanted signals are coupled between conductors. Because the structure of optical fiber nearly completely constrains the light energy to the core of the fiber, signals cannot couple between fibers in a cable, thus eliminating cross talk. However, cross talk can occur in copper cabling systems because electromagnetic fields are generated around each signal carrying conductor. Cross talk can be described as EMI within the cable itself. Cross talk most frequently occurs near the source of the transmitted signal (near-end cross talk, or NEXT), but can also occur at the opposite end of the link (far-end cross talk, or FEXT) on short distance links. FEXT is difficult to distinguish from NEXT and can cause confusion in testing. Of critical concern, however, is that the cross talk performance of a copper cable or connector depends upon operating frequency. Therefore, cabling with sufficient performance at low speeds does not ensure proper performance at higher speeds.

Fiber complies with all FCC, European, and Pacific Rim safety, emissions and susceptibility standards. Copper cables can radiate and receive electromagnetic energy over a broad range of frequencies. The FCC and other regulatory bodies limit the levels of radiated and conducted emissions permitted by electrical devices, primarily to reduce interference with radio and television reception. In some countries, copper-based systems must also be tested for susceptibility to external electromagnetic interference. However, passing these tests does not eliminate interference; but merely reduces its probability. If other parameters are held constant, the band of potential interference frequencies broadens as bit rates increase. To combat interference problems, copper-based systems must rely on more complex and costly transmission schemes, including various categories of cabling and increasingly complex coding electronics. By contrast, radiated emissions and susceptibility to external interference are almost entirely eliminated simply by the inherent design of optical cables.

Fiber has very low attenuation, enabling much longer link distances than copper. Attenuation is proportional to cable length and impacts the receiver's ability to distinguish between valid signals and noise. As a signal travels down copper cable, the impedance of the cable conductors reduces signal amplitude and quality. In copper cable, attenuation also depends on operating frequency; the higher the frequency of transmission, the higher the attenuation. Frequency-dependent attenuation limits the bandwidth of copper cables. With optical fiber, attenuation depends on the wavelength of the lightwave carrier rather than the transmission rate. This allows fiber systems to have the same attenuation at megabit rates as gigabit rates, which provides benefits for system upgrades.

Fiber's high bandwidth reduces signal jitter. Signal attenuation and jitter do affect fiber LANs, but to a much less extent than they do copper LANs. Therefore, fiber systems support much longer distances than copper. For example, most copper applications are limited to 100 meters (m) with data running up to only 155 megabits per second (Mbps). On the other hand, 62.5/125 micron multimode fiber can cost-effectively provide multi-gigabit data rates for distances of 300 m.

Fiber systems are more immune to impedance mismatches. When copper components (cable and connectors) of different impedance are interconnected, the resulting mismatch causes signal reflection, which increases the noise level on the cable. Reflections can occur at connection points in an optical system, too. However, nearly all present multimode fiber systems are completely unaffected by reflections. Reflections are of concern in some singlemode fiber systems, especially those carrying CATV signals. Fortunately, these reflections can be reduced by using the proper connection hardware.

Fiber provides greater reliability and safety.Unlike copper facilities, all-dielectric fiber cabling systems do not conduct lightning strikes. They also do not conduct the currents that can result from differences in ground potential or from improper grounding of the shield used in shielded or screened twisted pair.

Higher Reliability, Performance, and Flexibility with Fiber
Fiber's immunity to the factors detailed above translates into important benefits to the network.

Because of inherent immunity to external interference, deploying fiber networks can reduce error rates dramatically, even at gigabit data rates. Its near error-free transmission also improves network performance by reducing the need to retransmit entire packets of data. This significantly increases the overall throughput of the network segment.

Fiber's high transmission integrity can also improve network performance by reducing the need for cumbersome error detection and correction protocols. Systems designers in the telecommunications markets already use this benefit to their advantage and replace X.25 with streamlined protocols like ATM and frame relay that use minimal error checking.

In addition, fiber's increased transmission distance and greater capacity give network planners greater flexibility in how they configure a network. With fiber, network planners can either distribute electronics into telecommunications closets or collect electronics into one centralized cross-connect or equipment room. The use of centralized electronics greatly simplifies the management of LAN networks, provides more efficient use of ports on electronic hubs, increases network security, and allows simple implementation of various network applications throughout the building. These factors often reduce the initial system cost and definitely reduce ongoing support costs throughout the life of the network.

In September, 1995 the TR41.8.1 committee of TIA approved publication of Telecommunications Systems Bulletin (TSB) 72, Centralized Optical Fiber Cabling Guidelines. This bulletin provides information supplementing the ANSI/TIA/EIA-568-A, Commercial Building Telecommunications Cabling Standard regarding the design and installation of centralized optical fiber cabling. TSB72 lays the groundwork for future modifications to ANSI/TIA/EIA-568-A when it is reissued as ANSI/TIA/EIA-568-B in a few years.

The Truth About Leveraging Copper
Despite the superior transmission characteristics of optical fiber cable, many users and vendors believe that installing copper will help them minimize the initial cost of their network -- and they hope to build off their existing copper infrastructure to support new protocols and the ever higher bandwidth applications. Unfortunately, many find that leveraging the existing copper is not that easy.

In fact, upgrading existing copper networks to support 100 Mbps transmission often requires expensive cable plant replacement and redesign. Obtaining higher performance also demands more stringent installation, routing and testing procedures. All of which impact the potential benefits of using copper.

None of the unshielded twisted pair (UTP) cable installed prior to 1992 is categorized and much is below category 3 performance. However, many of the 100-Mbps applications require category 5 UTP cable and connectors. Consequently, new category 5 cable and connecting hardware must be installed to support the data rates used in high-bandwidth LANs like FDDI and 100BASE-TX. Other protocols like 100BASE-T4 and 100VG-AnyLAN can use category 3, but that means that most cable installed before 1992 still must be replaced.

Outlets and patch panels also must be upgraded to support category 5 cable facilities. These new hardware components cost as much as 60% more than their predecessors. Moreover, specification for this hardware has only been available since 1994. As a result, category 5 cable that was installed prior to 1994 (with category 3 or 4 connecting hardware) may have to be retrofitted with new category 5 outlets, patch panels and jumper cables before it can be a complete category 5 system.

Installation And Maintenance Considerations
One of the myths that has kept users from installing fiber is that it is more difficult to handle and install than copper. While this may have been true a decade ago, today, several factors have changed that situation. First, optical fiber cable is now an established medium, used extensively for building backbones and risers. Most installers have had considerable experience installing 62.5 micron multimode fiber in backbones -- experience that is easily transferred to installing the same type of fiber in the horizontal.

At the same time, copper cabling has changed considerably. Installers who have had experience with low-speed copper networks find that high-speed copper networks require more stringent and time-consuming installation procedures. Greater care must be taken to ensure proper termination and cable dress. Also, because copper cable is susceptible to EMI/ RFI-induced errors, extreme care must be exercised when installing it near electrical machined heat sources, high voltage transmission systems and other EMI/RFI sources. By contrast, fiber's immunity to EMI/RFI, ground-plane effects, lightning, impedance differences, and wide temperature changes greatly diminishes routing concerns.

The TIA/EIA specifications for category 5 UTP place stringent requirements on the way that copper cable is pulled and terminated. In particular, the ANSI/TIA/EIA-568-A Commercial Building Telecommunication Cabling Standard specifies that the force used to pull UTP cable must not exceed 25 pounds. Moreover, to preserve the performance advantages that are afforded by tight pair twisting, particularly at the points of signal transmission, specifications require that pairs not be untwisted by more than 1/2 inch at termination points. Adding to the difficulty, Type 1A shielded twisted pair (STP) has no specifications covering cable pulling or termination practices.

By contrast, fiber has more robust pulling specifications than category 5 UTP or STP. Although bare fiber requires some handling precautions, optical fiber actually has a greater tensile strength than copper or steel fibers of the same diameter. Further, when fibers are incorporated in a cable, the result is an extremely durable assembly. The strength of fiber cables can be attributed in part to the use of aramid yarns, which bear almost all the tensile load placed on the cable with little to no load transmitted to the fibers. The strength of the fiber cable assembly enables it to withstand more than 100 pounds of pulling force without affecting performance as required by the ANSI/TIA/EIA-472CAAA specification referenced by ANSI/TIA/EIA-568-A. This tensile strength reduces concern of damage during installation. In comparison, the conductors themselves are the major strength elements of a copper cable, so pulling forces are transmitted directly to the critical components of the cable.

Optical fiber cable's small diameter and light weight also makes it easy to install and maintain. Compared to four-pair category 5 UTP cable, two-fiber cable is 25% to 40% lighter and occupies as much as 15% less space. Optical fiber also has no requirement to maintain tight twists at termination points, a requirement critical to category 5 cabling systems.

The physical similarities between the various categories of UTP are also a cause for concern. While category 5 systems require multi-page documents detailing new, stringent installation requirements, it is still UTP. It looks the same as the low grade UTP that contractors have been installing without concern for pulling tension and termination untwisting for many years. If new rigorous techniques are not used when installing category 5, then the required link performance may never be achieved.

If the cable plant is installed successfully, the burden then shifts to the maintenance technicians to employ the same category 5 rigor when maintaining the cable plant. Empirical testing has demonstrated that replacing a category 5 patch cord with a category 3 patch cord degrades the link performance. In fact, ANSI/TIA/EIA-568-A classifies a link made up of different category components as having the performance level of the least performing component in the link. In a survey reported in the May 1995 Business Communications Review, 29% of network managers reported moving one quarter of their workstations in the past year. With this volume of network cabling changes, much care must be taken to preserve category 5 performance.

The specifications for optical fiber, on the other hand, were standardized several years ago so that 62.5 micron multimode fiber is now used in virtually all intrabuilding applications. Using the same installation and termination practices, 62.5 micron multimode fiber can be extended into the horizontal for a totally uniform cabling infrastructure.

High Speed Copper Links Complex to Test
Testing requirements also become more complex when high speed copper is used for data networks. To verify category 5 link performance, tests for attenuation, cable length and NEXT must be conducted. Attenuation and NEXT tests must be performed across the entire frequency range of 1 to 100 megahertz (MHz), with measurements made at more than 400 frequencies within that range. The entire frequency range must be tested because copper based systems operate at different frequencies, and the link's performance at one frequency does not ensure proper performance at other frequencies. Further, NEXT measurements must be made in both directions, potentially doubling the testing effort.

Recent studies indicate that just using category 5 components does not ensure that the user receives category 5 link performance. Independent studies indicate that not all category 5 components are interchangeable and that installation techniques can have a dramatic effect on the link performance.

Only as of September, 1995 has the industry developed field testing specifications for category 5 UTP links in the form of TSB-67. Several vendors now are marketing new testers that claim to meet its level 11 accuracy requirements. However, a great deal of effort remains to develop a truly repeatable, accurate test procedure that will ensure system performance. This is apparent within the TSB in Section 6 and Annex A with the discussion on accuracy. Worse yet, STP has no link performance field test procedure at all.

In addition to attenuation, cable length and NEXT, copper link impairments such as Longitudinal Conversion Loss (LCL), FEXT, and "power-sum" NEXT are just now surfacing. LCL is related to link imbalance, and converts signals into noise and noise into signals. FEXT can become a problem for short distance links and can be confused with NEXT causing links to fail certification tests. Power-sum NEXT is an issue for multi-pair cables where crosstalk disturbances can combine from all pairs within the cable. Hand-held testers are not capable of measuring LCL or of measuring power-sum NEXT for multi-pair cables. Additionally, most hand-held testers cannot distinguish FEXT from NEXT.

To further compound the issue, a category 5 system will provide adequate performance for lowspeed data applications even if not installed and maintained properly. However, in these cases category 5 link performance will not be available when high-speed data applications are deployed. Therefore, users of copper cabling systems are entrusting their ability to support higher speeds on a potential cabling time-bomb.

Because of the lack of standardized field-test procedures, the lack of interoperability between category 5 components, and the dramatic effects of installation on performance, Stephen Saunders in Data Communications (June 1994) reported, "Cabling consultants estimate that 20 percent of all category 5 cabling runs now in place may be unsuitable for carrying high-speed traffic."

Attenuation: An Accurate Test of Fiber Cable
Fiber facilities, by contrast, are extremely easy to test accurately. Fiber cable facilities are not affected by NEXT and their operating performance is not affected by frequency; therefore, they can be tested by simply measuring the attenuation of the optical fiber link. Because optical bandwidth can not be adversely affected by installation practices, the user needs only to install cable with the minimum bandwidth required by ANSI/TIA/EIA-568-A to support applications up to 2.5 gigabit-per-second (Gbps) ATM for 300 meters.

Field test procedures for end-to-end fiber attenuation were first standardized in 1990 as EIA/ TIA526-14 and since then have proven to be very reliable, accurate and repeatable. ANSI/TIA/ EIA 568-A refers to this standard for link certification. It requires the use of an inexpensive handheld power meter and light source to test horizontal links in one direction at one wavelength, either 850 nm or 1300 nm.

As important as the availability of accurate and reliable hand-held test equipment and standardized procedures is the fact that optical fiber systems are simple, predictable, and have plenty of operating headroom. There are no interoperability problems with ANSI/TIA/EIA568-A compliant components from various manufacturers. Most significant is the ability to predict and ensure system performance at higher speeds based on performance at today's lower speeds. Because attenuation does not change with data rate and because bandwidth can not be adversely affected by installation, a horizontal link installed to support Ethernet today will support multi-gigabit LANs tomorrow.

Standardization Issues
One of the most confusing aspects of working with copper is its lack of uniformity. Copper cable comes in myriad UTP and STP types. Because each has its own transmission characteristics, selecting the appropriate cable for a particular LAN application can be a tricky proposition.

Fiber, on the other hand, provides a uniform solution. A single fiber cable specification encompasses the full spectrum of fiber-based LAN applications, including Ethernet, Token Ring, FDDI, 1 00BASE-FX, ATM, Fibre Channel and ESCON®. Each of these specifies 62.5/ 125 micron multimode fiber with the same optical performance parameters specified in ANSI/TIA/EIA-568-A.

Understanding The Cost Of Fiber
Historically, the strongest argument against fiber, even for applications that require new cabling, has been higher initial cost. However, more stringent UTP requirements, coupled with advances in fiber technology and higher fiber production volumes, have narrowed the gap considerably. Moreover, when ongoing system-level cost factors such as network operation, management, and plant upgrades are considered, fiber is less expensive over the long haul. To understand a network's total/installed lifetime costs, you should consider:

  1. Cabling component costs, including the cable, wall outlets, patch panels, patch cords and connectors;
  2. Labor costs to install and test the cabling system;
  3. Electronics costs, such as hubs, concentrators and network interface cards;
  4. Activation costs, including the labor and software needed to activate the network;
  5. Downtime costs, including the impact of network outages on productivity;
  6. Maintenance costs;
  7. Network management costs, including the recurring cost of network changes;
  8. Recabling costs associated with upgrading to higher bandwidth networks in the future.

Fiber systems have distinct advantages that reduce overall life-cycle costs. These savings can far exceed any remaining initial cost premium.

The following subsections examine each of these costs in some detail.

Cabling System Component Costs.The cost of fiber components is still marginally higher than that of category 5 UTP. However, more stringent category 5 requirements have caused the cost of copper-based systems to increase. In addition, lower costs of fiber components (because of higher fiber component production volumes), has caused the cost of copper and fiber cabling components to converge.

Installation and Testing Costs. The total labor costs to terminate and test optical fiber cable and category 5 UTP cable are now comparable. Many new field-mountable fiber optic connector technologies make terminating fiber optic cables easier and faster than before, so that the termination time for fiber optic connectors is now similar to that of category 5 copper. While category 5 systems may still hold an advantage in termination labor costs, the testing costs for category 5 UTP are significantly higher than for fiber.

Fiber's small size, light weight and longer transmission distance also provide savings by simplifying initial installation and reducing cable space requirements. Relative to copper, fiber typically requires less space in cable trays, conduits and telecommunication closets. And when recabling in existing space, fiber's immunity to interference provides the unique option of using the power conduit to run new data cable as long as NEC (National Electrical Code) guidelines are observed.

Open office cabling is another application where optical fiber has significant advantages over copper. More users are electing to install "zone" cabling in open office environments, specifically where open systems furniture is used. Because optical fiber has no cross talk, a multi-fiber cable can serve 6, 8 or even more users with no concern over interference between incompatible signals. Additionally, there is no concern about separation from power cables within the furniture or wiring columns. The increased patch cord length required with multi-user outlets also presents no problems for optical fiber.

In comparison to this favorable fiber scenario, installing power-sum 25-pair category 5 cable can be more expensive, more difficult to terminate, and riskier because the cable has no history or field-proven effectiveness. Many 25-pair category 5 cables have non-standard or vendor specific conductor color codes. This can lead to confusion during installation and maintenance. In order to meet the requirements for power-sum performance, some vendors resort to unusual cable constructions with pair-bundle separators or non-uniform pair bundling. These dissimilar and unfamiliar cable constructions require special training, unique to each vendor's product, to install and maintain. In addition, routing copper cables in proximity to power conductors within modular furniture raceways, and the need for longer patch cords with multi-user outlets, pose problems for copper cabling systems.

To address the need within the industry for assistance with these issues, TIA TR-41.8, the committee responsible for ANSI/TIA/EIA-568-A, is presently considering a proposal for TSB publication regarding open offfice cabling called PN 3398, Additional Horizontal Cabling Practices for Open Offices.

Electronics Costs. Until recently, higher electronics costs, particularly for transceivers, has been a stumbling block for deploying fiber cable. These premiums have been reflected in the cost of system-level electronics such as hubs, concentrators and network cards. Fiber-based Ethernet and Token Ring PC adapter cards, hubs, and concentrators, have historically cost up to twice as much as their copper-based counterparts. Fiber-based FDDI hubs, concentrators, and network adapter cards, similarly, have carried a premium of about $500 per attachment. However, this scenario is changing. Already lower cost electronics are available, and new architectures are developing that allow users to install fewer electronics, thereby lowering the overall cost of their system.

As data rates rise, the increasing complexity of copper-based electronics will cause copper and fiber solutions to converge. Les Baxter from Bell Laboratories predicts in Cabling, Installation and Maintenance that, "For the next generation of high-speed LAN equipment (operating in the 100 to 155 Mbps range), we expect the price of fiber-based equipment will be approaching parity with copper-based equipment. At higher speeds, such as 622 Mbps, fiber-based equipment will probably have a cost advantage."

Another factor changing the cost equation is the development of lower-cost transceivers for high data rate inbuilding cabling distances. In the past, the optical fiber industry has been focused on developing solutions for campus backbone applications with distances up to 2 kilometers. With the introduction of the low cost Compact Disc (CD) laser-based light sources in 1991, and high-speed Light Emitting Diodes (LEDs) and Vertical Cavity Surface Emitting Lasers (VCSELs) in 1995, the cost differential between fiber and copper-based implementations for hubs, concentrators, and network adapter cards operating at any speed is being eliminated. These lasers and LEDs hold the promise of lower costs for high-bandwidth LANs such as FDDI and ATM. The cost-lowering effect of these new optoelectronic technologies is already evident. For example, FDDI concentrators and adapter cards based on new light sources have been publicly announced at prices that are at parity to their UTP counterparts. Also, by leveraging off earlier FDDI developments, 155 Mbps ATM electronics are available today with fiber and copper counterparts priced near parity.

The TIA FOLS is actively petitioning the major application standards groups to consider these low-cost components for in-building cabling needs. A copy of the proposal is in Annex B.

Activation Costs. The costs associated with activating a network include the costs of installing network interface cards (NlCs) into PCs; installing and configuring PC software; making attachments to LAN hubs, routers, switches and gateways; and installing and configuring network administration software. All of these tasks are required for both fiber and copper networks, therefore the costs for either cabling alternative are the same.

Productivity Costs. While network managers are justifiably concerned with first-order costs such as cabling, components and installation, there are a number of system-level factors that can ultimately make fiber less expensive over the long haul. Consider, for example, fiber's superior reliability. Fewer data errors occur with fiber because it is immune to the effects of EMI, RFI, crosstalk and impedance mismatches. Because fiber supports near error-free transmission, network outages are minimized, resulting in minimal work disruptions for all people and machines connected to the network. The cost savings that result from increased productivity alone can more than offset the incremental cost associated with installing fiber.

Copper-based data networks, by contrast (according to a 1992 Communications Week user survey), average 2.3 network outages per month, at an average cost of $19,175 per month. LAN wiring accounted for 17% of these outages, which have historically been caused by EMI, RFI, cross talk, impedance mismatches and excessive transmission distances.

Maintenance Costs. Fiber networks cost less to maintain than copper networks because they experience fewer network problems, and require less time and effort in troubleshooting and correcting cabling problems. These advantages have been enjoyed for years by the telephone industry and more recently by cable television providers.

For example, troubleshooting problems is easier in an optically-based premises networks because the only parameter you need to examine is attenuation -- and that is done with only a hand-held power meter and light source. An Optical Time Domain Reflectometer (OTDR) may also be used to help locate the source of attenuation problems.

Fiber networks are a testament to the adage that simpler is better. Standard fiber outlets are made from simple two-fiber appearances; one fiber for transmitting and the other for receiving.

In contrast, standard copper outlets use 4-pair wiring. In general, one copper pair is used for transmitting and another for receiving, but the pairs used vary by application. This increased complexity and non-uniformity can result in confusion during troubleshooting. Because of the multi-pair nature of copper wiring, problems can result from incorrect wire mapping with crossed pairs, reversed pairs, and split pairs. These problems can manifest themselves as open or short circuits, and lead to erroneous assumptions and wasted effort.

If an improper wire map is not the cause of a copper problem, troubleshooting diagnostics must continue with tests for excessive length, attenuation and cross talk. This list of potential problems and diagnostic tests increases the maintenance costs of copper networks relative to fiber networks .

Administration Costs. Fiber's ability to transmit over longer distances facilitates the use of cabling alternatives that reduce cost and improve network manageability. In many cases, using fiber cable can eliminate the need to place active components in telecommunication closets within 90 meters of work areas. Instead, network managers can use centralized electronics to reduce the costs associated with under-populated hubs and underutilized boards and switches. As added benefits, locating hubs in the same room with bridges, routers, gateways and servers simplifies network troubleshooting, reconfiguration and security. This reduces network downtime and staffing requirements.

To address these capabilities TIA approved publication of TSB-72, Centralized Optical Fiber Cabling Guidelines, providing supplemental information to ANSI/TIA/EIA-568-A regarding the design and installation of centralized optical fiber cabling.

Recabling Costs. Fiber's greatest cost advantage lies in its ability to support higher data rate networks without recabling, thereby extending the life of the network infrastructure. The inability of the installed base of precategorized copper facilities to handle new LAN implementations has led to the development of increasingly stringent cabling requirements resulting in the present category 3, 4 and 5 specifications. Further complicating the issue is the fact that now category 6 specifications are in debate. While the copper standards are still evolving, 62.5 micron multimode fiber already provides more than enough headroom to handle not only existing LANs, but those that are on the drawing board for the future.

ATM and Beyond
Even as copper proponents struggle to achieve reliable FCC-compliant 100-Mbps implementations, many network managers are expressing interest in alternative upgrade paths. One reason is the growing popularity of multimedia communications--the marriage of video, audio, image, and animation along with data, text, and graphics. Transport and switching of multimedia information imposes new constraints on networks including control of latency, latency variation, channel bandwidth, bandwidth allocation, burst errors and packet loss.

Historically, the data communications industry has developed LANs using shared bandwidth for time-insensitive data traffic, which allows re-transmission if errors are detected. Latency sensitive multimedia traffic does not allow re-transmission of frames or cells. This is an issue for multimedia traffic transport over UTP desktop links where burst errors can become significant. If you must add capability at the Application Level to conceal burst errors from corrupting the presentation below a satisfactory user level, this additional cost must be weighed against the relative cost of physical link technology. Copper proponents are hoping to push the copper technology envelope with yet more complex coding, cable/apparatus design, and installation schemes. Meanwhile, fiber is already a step ahead, providing the high data rates and high reliability needed for not only ATM, but for the higher-performance LANs of tomorrow.

Annex A provides further discussion on these issues.

References

  1. Copper and Fiber Duke it Out on the Desktop, LAN Times, October 18,1993, p.47.
  2. The Cabling Cost Curve Turns Toward Fiber, Data Communications, Nov.,1993, p. 55.
  3. Survey: Outages, Communications Week user survey, January 13,1992, p. 28.
  4. Bad Vibrations Beset Category 5 UTP Users, Data Communications, June 1994, p.49.
  5. Fiber Versus - No, Fiber And Copper, Cabling Installation and Maintenance, Oct. '94, p. 49.
  6. Active Component Manufacturers Lower the Cost of Fiber to the Desktop, Lightwave, February 1994, p.58.

    Note: This document was prepared by the Fiber Optics LAN Section and does not necessarily reflect the views of all TIA members.

    Annex A
    Extending UTP Data Rates Proves Complex and CostlyAs copper proponents try to extend the use of UTP cable into yet higher data rate applications, they face fundamental technology hurdles. EMI/RFI emissions and susceptibility to both signal degradation and external interference, which can cause burst errors, increase as the transmission frequency increases. This, in turn, makes it increasingly difficult for copper networks to meet FCC emissions requirements and ANSI bit error rate requirements. For example, to meet the bit error rate objective of the FDDI Twisted Pair Physical Medium Dependent (TP-PMD) specification, while still meeting European and FCC emissions requirements, copper proponents have adopted complex multi-level signal coding techniques. Multi-level coding reduces the spectrum (range of frequencies) needed to achieve a given data rate by pushing more of the information into lower frequencies. This frequency reduction cuts down on EMI/RFI emissions and enables high-data-rate signals to be less susceptible to degradation. However, multi-level codes reduce signal-to-noise ratios and require more sophisticated receivers.

    For comparison, Ethernet and Token Ring are based on two-level Manchester line coding. Manchester has a fundamental frequency of 1 Hertz per bit. Thus,10-Mbps Ethernet and 16Mbps Token Ring have fundamental frequencies of 10 and 16 MHz respectively, with spectra extending below and above the fundamental frequency. FDDI combines 4B/5B source code with simple two-level NRZI line code to deliver 100 Mbps with a 62.5 MHz fundamental frequency on optical fiber.

    To increase operating margins and comply with emissions requirements, UTP system designers have replaced NRZI with MLT-3 line coding to reduce the spectrum needed for the 100-Mbps data rate of FDDI TP-PMD. By using three-level line coding, MLT-3 is able to compress more of the spectrum required to transmit 100 Mbps below 30 MHz, the lowest frequency at which radiated emissions become an issue. UTP system designers have also added scrambling to MLT-3 coding. Scrambling smoothes the spectral peaks caused by repetition in the transmitted bit patterns. On the other hand, the ATM Forum has allowed the use of NRZ coding at 155 Mbps on category 5 UTP. Since NRZ was ruled out for FDDI TP-PMD at 100 Mbps, due to emissions and performance issues, the use of NRZ at even higher bit rates raises considerable doubt on its ability to meet both the bit error rate objectives and emissions requirements.

    Of further concern is that both NRZ and MLT-3 line code spectra extend well beyond the 100 MHz specifications of category 5 cable. The consequence is that jitter and inter-symbol interference increase as the upper end of the spectra are attenuated or filtered out by the cabling system. Therefore, high speed data transmission on UTP is a delicate balance between radiated emissions requirements and bit error rate performance.

    The use of multi-level coding and scrambling have two main drawbacks. First, compared to the non-scrambled two-level coding used in low-speed LANs, the added complexity requires more expensive coding and decoding electronics. Second, the addition of each new signal level requires increasing the signal amplitude in corresponding increments to maintain the same signal-to-noise ratio (SNR). However, increasing the signal amplitude also increases EMI/RFI emissions and NEXT, so a practical system will actually have a lower SNR, making it more difficult for the receiver to distinguish between data signals and noise. This, in turn, requires the use of more sensitive, and therefore, more expensive receivers. Even with MLT-3, signal attenuation limits the maximum distances over which data can be reliably conveyed at high data rates to 100 meters.

    Other high-speed networking groups are considering even more complex coding schemes, such as CAP-64, which rely on costly digital signal processors (DSPs). Some groups are considering 4-pair transmission instead of the conventional 2-pair transmission, which will not work on 2-pair or 3-pair cabling. Finally, with the hope of reducing emissions, other groups are considering common-mode termination methods to augment the traditional differential termination method. The issue with common-mode termination is that copper components and links are tested using differential test methods and performance is uncertain with common-mode termination schemes. Another method of increasing data rate capabilities while curtailing EMI/RFI emissions, is through the use of more sophisticated cable designs. In plenum-rated category 5 UTP, for example, the

    Ethylene and monoCloroTriFluoroEthylene (ECTFE) conductor insulation used in category 3 UTP is replaced with polyolefin or FEP (Fluorinated Ethylene Propylene) which reduces signal loss. Simultaneously, the twist ratio is increased from two to six twists per foot (used in category 3) to one to two twists per inch. Each pair must be given a different twist rate in order to minimize periodic signal coupling between the pairs in the cable. These techniques, combined with the additional manufacturing control, testing, and development costs prove expensive and result in as much as a 300% premium for category 5 UTP relative to category 3 UTP. Even with these cable enhancements, adhering to the category 5 UTP requirements does not eliminate EMI/RFIinduced errors. It only reduces the probability of experiencing those errors.

    Annex B
    TIA FOLS Premises Fiber Technology RecommendationsTIA FOLS members have observed national and international standards organizations propose and adopt a wide range of technical optical fiber solutions. For building cabling, the lowest cost solution for optical fiber is rarely included. Instead, proposed applications for optical fiber typically address the interbuilding backbone network, characterized by distances up to 2 km, with virtually no effort or attention focused on the requirements for shorter horizontal fiber-to-the-desk links. Consequently, the applied cost of fiber optic solutions is not always appropriately considered and optical fiber is often judged to be sub-optimal for desktop applications.

    The FOLS believes that in order to be truly commercializable, all new and existing optical fiber applications standards should include cost-effective solutions for distances of 300 meters and less. The distance of 300 meters will provide users with a migration path based on current and proposed source technology that supports applications up to and including 2.5 Gbps.

    TIA chartered the Fiber Optics LAN Section (FOLS) to:

    • Position fiber as the optimum choice for higher speed LANs to the desktop;
    • Help network planners and users understand the price/performance benefits of optical fiber;
    • Correct misconceptions about the viability and cost-effectiveness of optical fiber for implementing easily upgradable high-speed LANs;
    • Encourage standards organizations to develop cost-effective optical fiber Building Cabling solutions optimized for distances up to 300 meters; and
    • Identify network architectures that allow users to take advantage of the benefits of fiber while at the same time simplifying their network design and management.

    TIA FOLS' mission is to clarify these issues and communicate to end-users accurate information regarding the technical and economic benefits of implementing optical fiber technologies.

    TIA FOLS asserts the following for Building Cabling applications:

    • Building Cabling is defined as premises cabling with runs up to 300 meters in length;
    • Greater than 95% of combined horizontal and backbone intrabuilding cabling distances are less than 300 meters;(1)
    • Optical fiber Building Cabling is 62.5/125 micron multimode with at least 160 MHz·km bandwidth in the 850 nm window and 500 MHz·km bandwidth in the 1300 nm window;(2)
    • Users should employ the lowest cost device technology for Building Cabling applications, which may differ from device technology for Campus Cabling applications;
    • Whenever a connector is used, it will be the Duplex SC as defined in EIA-568-A;
    • Electronics can be located in a centralized cross-connect (main equipment room) with passive cross-connects, interconnects, splices, or pull-through cables in the telecommunications closets;
    • Emerging technologies, like vertical cavity surface emitting LASER (VCSEL) devices or fast LEDs, should be considered for applications that exceed the speed or distance capabilities of conventional LEDs.

    The Premises Fiber Technology Recommendations matrix that follows indicates the FOLS' view of the most cost-effective solutions for a wide range of standard LAN applications. Bold underlined entries indicate where modifications to existing or draft standards are needed to support the most cost effective approach. The TIA/FOLS recommends that the responsible standards bodies develop specifications to support the entries in this table in order to promote the use of optical fiber for Building Cabling by providing the most cost effective technology possible.

    (1) In a study commissioned by AT&T to determine intrabuilding cabling distances, it was discovered that 95% of the lines from desktop to Main Room were shorter than 244 meters (800 feet). To include 98% of the sampled lines, distances up to 365 meters (1200 feet) would need to be considered. The study included 45 small businesses with fewer than 200 PBX lines and 34 larger businesses -- 10,000 lines in total.

    (2) Virtually all data communications standards efforts have recommended the use of 62.5/125 micron multimode fiber or single-mode fiber for their respective applications (although a few carry informative annexes detailing the allowance for 50/125 micron multimode fiber).

    See