Which technologies will meet speed and data-transmission requirements in the future?
To discuss the future of structured cabling systems, we should spend a moment looking at the past and present practices. Before early 1984, communications wiring systems, as they were then called, were not topics of much discussion. Data-transport speeds were comparatively slow, transmission bandwidth requirements were minimal, and "others" provided the design and installation of the cabling system. The telephone companies took care of the installation and maintenance of the telephone service. When data services became distributed, the cabling system was vendor-specific, and the installation and maintenance of the cable plant was provided by the equipment vendors, or specialized independent contractors.
In January 1984, the courts in the U.S. made several important rulings that changed the way telecommunications was provided and distributed. The user was pretty much left to go at it alone. Outside of traditional suppliers of cabling systems, not much was understood about the cabling requirements for communications transport. The proliferation of media and connector interfaces, a lack of standard transmission specifications, and the introduction of cabling schemes by vendors added to the user's confusion. Bringing order to the confusion and creating generic cabling systems demanded the creation of a standards body whose output would focus on commercial buildings and communications cabling.
Since the introduction of the first cabling standard in 1991 by the Telecommunications Industry Association/Electronic Industry Alliance (TIA/EIA-Arlington, VA), that same group has issued a series of standards and specifications regarding most aspects of the structured cabling systems. These standards have provided guidance related to evolving high-speed information transport systems. The vast majority of past and present digital communications, used in the commercial world, has been transported on unshielded twisted-pair (UTP) cabling systems. UTP became the medium of choice because it was economical, perfectly adequate for the applications, and comparatively easy to install versus other available media types.
With advancements in network speeds, new transmission specifications for UTP cableand connecting hardware have been promoted by manufacturers and eventually endorsed by the standards community. In 1991, the highest rated bandwidth over UTP was 16 MHz; we are now faced with transport speeds demanding transmission bandwidths of 200 MHz and beyond. Manufacturers have risen to the challenge and provided UTP components for today's transmission requirements in excess of 250 MHz. What media and connectors will be available that will be economically feasible at bandwidths of 600 MHz? At what point does UTP become less easy to use and less economical than other media?
As far as we can see into the future, commercial information transfer will consist of both low-speed and high-speed requirements. Applications such as voice, building automation systems, alarms, and security systems will still use low bandwidths. Voice information may change from central office/private branch exchange delivery to intelligent peripheral/voice, but bandwidth requirements won't increase by a large amount. The data packets, with which the voice packets ride, will increase bandwidth requirements. The requirements to transport large amounts of information in shorter and shorter periods of time are changing, and will continue to change. Applications such as graphical data (both schematic and pictorial), scientific modeling, desktop videoconferencing, multi-tiered relational databases, and other data-intensive information will drive up the bandwidth requirements.
If history is any predictor of the future, we will see information transfer speeds increase at least one order of magnitude per decade. We have seen local area network (LAN) speeds, on UTP, increase from 10 Mbits/sec in the mid 1980s to 100 Mbits/sec in the mid 1990s, then to 1 Gbit/sec in the late 1990s. Today, standards are being written for 10 Gbits/sec. Where will we be in 2010 or 2020?
Microsoft's Bill Gates is quoted as saying, "We will have infinite bandwidth in a decade's time." Lawrence Berkeley Laboratory (Berkeley, CA) has projected its throughput needs for 2004 to be 40 Gbits/sec. What will be its media of choice: UTP, coaxial, shielded twisted-pair, optical fiber, or wireless technology?
The two major properties required for any cabling system to be the system of choice are its performance and its relative economics (which include ease of installation). Undoubtedly, fiber and shielded twisted-pair (STP) systems are quite robust and provide greater signal headroom than UTP, however, they lag far behind UTP in customer acceptance for today's applications. But will UTP and other media systems provide the bandwidth for future applications? Will they provide economical solutions? Let's look at the proposed solutions for future requirements.
Currently, standards are being solidified to extend the transmission characterization of UTP to 250 MHz. Many manufacturers are advertising the availability of products that exceed yet-to-be-ratified specifications. Many technical hurdles have to be solved before a Category 6 standard is published, however, it is reasonable to expect that these hurdles - both technical and political - will be resolved. It is generally recognized that UTP has not yet been pushed to its theoretical limits. The question yet remains as to the continued viability of UTP as information speeds increase.
If future technology does not improve the efficiency of band-width utilization, then the cost of the electronics, installation detail, and the testing requirements may diminish UTP's benefits relative to other potential options. It should be noted that currently, there are no plans to develop a standard copper solution for 10-Gigabit Ethernet.
Shielded twisted-pair is currently characterized at frequencies to 300 MHz. The International Organization for Standardization and International Electrotechnical Commission (ISO/IEC-Geneva, Switzerland) is now studying STP for potential publication as a Category 7 standard. The spectral bandwidth will be characterized at frequencies to 600 MHz. Despite being an exception ally good transmission medium, its material and installation costs have restricted its use to special situations and certain countries.
Undoubtedly, STP's bandwidth limitations have not yet been reached, but as is the case with UTP, economics could be the major stumbling block to its adaptation. The installation of STP requires highly trained installers, and it is labor-intensive to install correctly. North American industry experts doubt that end users or installation contractors will be quick to embrace STP.
Screened twisted-pair (ScTP) cable was developed primarily to provide a shielded media that was less expensive and easier to install than shielded twisted-pair (STP). It is a cable design that meets all of the physical and transmission specification requirements of the TIA/EIA-568A Commercial Building Telecommunications Cabling Standard, and also provides additional protection from ambient electrical noise. ScTP is a four-pair, 24-American Wire Gauge (AWG), 100-ohm cable with a shield over all four wire pairs, while STP is a four-pair, 22-AWG, 150-ohm cable with a shield over each individual pair and then an overall shield. ScTP is characterized at frequencies to 300 MHz. Although easier to install than STP, ScTP cable suffers from the same lack of acceptance in North America as does STP. It certainly will be used in those instances when additional protection is warranted. But like STP, it is doubtful that ScTP will ever become the mainstream medium of choice.
Glass optical fiber
For many years, proponents of optical fiber solutions for information transfer have hailed it as the ultimate futureproofing media. The TIA/EIA standard offered 62.5/125-micron multimode optical fiber as one of three recommended horizontal media. Neither the distance limits nor the bandwidth capacity had been challenged by high-speed applications until the advent of 1,000-Mbits/sec Ethernet.
Studies now indicate that 62.5/125-micron multimode fiber's information-carrying capacity and its power-coupling efficiency to light-emitting diodes (LEDs) at short wavelengths are insufficient to meet the distance requirements of this application. Users now have to revisit the standards recommendations to evaluate their relevance to the future needs of the user's own network. They may have to consider changing to newer types of 62.5/125-micron fiber, or changing to 50/125-micron multimode optical-fiber cables. They may have to move from LEDs to vertical cavity surface-emitting lasers (VCSELs) for short wavelength (SX) transmission. They may opt to employ long wavelength (LX) transmission, and may choose singlemode fiber.
It appears that a universal futureproof media does not exist. The "new" 62.5/125-micron fiber will be more costly than singlemode fiber, and the 50/125-micron multimode optical fiber is more bend-sensitive and has greater loss at connectors than 62.5/125-micron fiber. Singlemode-fiber networks are thought to be more costly than multimode networks due to the light source and the connectors, but a new 1,300-nanometer (nm) VCSEL source may, in fact, lower the total cost below that of systems with the "new" 62.5/125-micron multimode fiber.
Wave division multiplexing (WDM) is a new technology that expands the data-carrying capacity of optical fiber rather than using a new media type for structured cabling systems. To increase the carrying capacity of fiber, the laser light that carries data through fiber-optic glass can be split into different colors, or more precisely, wavelengths, each of which carries a discrete data channel.
Today, the technology will support up to 40 different wavelengths. In the near future, 128 channels will be available. The best result of this technological development is that transmission facilities for new wavelengths can be retrofitted onto existing plants that connect to fiber already in the ground, which makes it the easiest way to increase bandwidth. This technology will obviate the requirements of additional fiber and will use the existing optical fiber and connectors.
Plastic optical fiber
Historically, plastic optical fiber (POF) has been relegated to low-speed, short-distance applications. Recent technical developments of graded-indexed POF have increased bandwidths to 3 GHz/100 meters. Recently developed singlemode POF, optical amplification in plastic fibers, new POF materials with low loss to 1,550 nm, higher power, and faster sources have allowed POF to realistically be considered for applications such as Fiber Distributed Data Interface (FDDI), Asynchronous Transfer Mode (ATM), Escon, Fiber Channel, and SONET.
But this medium is not endorsed by any standards body because the current technology is limited to a distance of 50 meters at the required bandwidth. Endorsement within a written standard is crucial for market acceptance. It will be perhaps five years before low-cost POF will be commercially available. If and when a standards body sanctions POF, it should offer a more robust system for applications currently served by copper media, at a cost below that of glass optical fiber.
Much has been written about the prospect of wireless networking replacing fixed-media structured cabling systems in the future. Currently, cost and low bandwidth have left wireless technology with approximately 1% of the number of deployed Ethernet ports. The features of wireless networking are beguiling to those who are involved with the design, installation, and maintenance of structured cabling systems. With wireless, there are no more concerns about running cable to inaccessible locations, and no more concerns about cable types. But for all the magic of wireless networking, there are downsides. Although a standard for wireless networking exists - IEEE 802.11b put out by the Institute of Electrical and Electronics Engineers (New York City) - complete interoperability among all wireless-LAN vendors remains unattained.
Narrowband networking equipment needs Federal Communications Commission licensing, and unfocused infrared networking equipment can be rendered unreliable by interference caused by other light sources, such as sunlight. Spread-spectrum networking equipment overcomes these problems to some extent but suffers from relatively low data rates.
IEEE 802.11b stipulates an 11-Mbit/sec data-transfer rate. An Australian company has recently developed a wireless system that claims to support 54 Mbits/sec. In an open office plan, propagation of the radio waves may be limited to distances of 200 feet to 500 feet. In a closed-wall office environment, propagation may be limited to as little as 100 feet.
Undoubtedly, the cost of wireless networking will be reduced and the bandwidth will increase. Wireless networking can serve admirably in numerous applications, but it is doubtful that the technology will provide the solution to ever-increasing information-transfer-speed requirements in environments where the work areas are relatively fixed for extended time periods (1 to 3 years).
Coaxial cable has been the medium of choice for wideband applications ranging from high-fidelity audio to television to baseband and broadband communications. Coaxial cable was the primary media for 10Base-5 and 10Base-2 Ethernet. The advent of higher-bandwidth UTP cable and connector technology replaced coaxial cable in commercial networks and has relegated its primary use to legacy networks and cable television.
For many years, coaxial cable was used in the IBM 3270 networks that represented a major share of data communications in commercial facilities. The ease of installation and the economics drove the market decidedly in favor of UTP. But before we close the chapter on coaxial, it does still offer possibilities.
Coaxial networks can support higher bandwidths than UTP, and can operate with less-sophisticated electronics. One could argue that it will be difficult for coaxial cable to gain acceptance with installers and users because of installation and maintenance difficulties and higher cost. But suppose that the same rigorous development programs were applied to coaxial systems as have been applied to UTP? Certainly, connectors could be developed that are as easily installed as UTP connectors, and cable could be manufactured that was lighter in weight and smaller in diameter than existing coaxial cable. When compared to the cost of more sophisticated electronics, installation detail, testing requirements, and ongoing maintenance of UTP in higher bandwidth applications, coaxial could possibly be more than competitive.
Other than the required performance, a structured cabling system should be economical both for material price and labor. Molex Premise Networks undertook a cost-comparison based on a sample network of 96 workstations (see table, below). We assumed that each workstation was approximately 100 feet from the telecommunications closet (TC). The TC contains four 24-port jackfields and 96 patch cords. The fiber-optic cable considered for comparative purposes had 62.5/125-micron multimode fiber. The prices are retail, assuming comparatively small quantities.
Numbers being what they are, one cannot just look at the cost of materials when making a purchasing decision. The applications to be run and the anticipated requirements must be factored into any decision. For instance, Category 5 cable is characterized at frequencies to 100 MHz, while Category 6 is characterized at 21/2 times bandwidth. Trying to double the bit rate of a signal using the same bandwidth will considerably increase electronics costs, as it will take more-sophisticated equipment to decipher the transmitted signal.
Unless there is a need to reduce electromagnetic interference (EMI), most users cannot justify the up to 300% price premium of STP and ScTP over UTP. The use of fiber, at 4 to 41/2 times the cost of UTP, may be justified on several levels: EMI cancellation, increased bandwidth, and longer distances. In addition to the material cost for the optical-fiber media, however, using optical fiber for a 10/100-Mbits/sec Ethernet system increases the cost of the hub and network interface cards (NICs) by 280% over copper-based NICs. A wireless system is slightly more costly than a copper-wired system, but has much less bandwidth. With each order-of-magnitude increase in bit rate, the cost of the electronics increases on the order of 600% to 1,000% initially; after volume increases, the cost tends to taper off to approximately 300% to 500%.
UTP will likely be with us for applications at or below 100 MHz well into the future. At one time, it was thought that UTP's capability was limited to transmission frequencies less than 1 MHz. Although it is difficult to forecast what new technology will be developed, it is safe to say that, given the amount of research being done in the area of electronic communications, it is quite possible to extend its capability with a new dielectric, new manufacturing processes, or other such advances. On the receiving end, it is likely that electronics will be developed to increase the sophistication with which signals are interpreted, or a new breakthrough in signal processing may be developed that allows increased bit packing.
Shielded twisted-pair and screened twisted-pair media will also continue to find their use in high-EMI environments for a long time to come. Their cost premium and increased installation costs are a disadvantage, but they offer solid performance at frequencies higher than those that UTP can accommodate. The possibilities of using coaxial cable systems will not be fully realized until manufacturers invest in the development of installer-friendly coaxial cable and connectors.
Wireless technology will undoubtedly advance and continue to support greater bandwidth requirements. It will probably see an even greater share of the market, especially in residential environments. But it does not appear likely that wireless will increase to the extent that it will support the requirements of application protocols already on the horizon.
There already appears to be a slow-but-steady migration in the industry toward optical fiber throughout the network as a precautionary measure. In general, this is prudent when there are few budgetary restrictions. Users should be interested in the information-transfer system as a tool to provide productivity for their enterprise, rather than support one technology versus another technology. And, even though in terms of application protocol Gigabit Ethernet will be with us for awhile, and Category 5E and Category 6 will support it just fine, we do know that in the next 5 to 10 years, 10-Gigabit Ethernet will require a totally optical-fiber infrastructure.
Debate continues on employing the new multimode fiber versus singlemode fiber, short wavelength (SX) versus long wavelength (LX) transmission, and using WDM for new fiber installations. The development of a 1,300-nm VCSEL will enhance singlemode fiber's chance of deployment in both backbone and horizontal usage. It seems that the ultimate solution will probably be a combination of many of the current optical-fiber options.
In the final analysis, manufacturers of structured cabling products must always be aware of new developments in electronic components, breakthroughs in materials technology, and innovations in signal transmission. They must also be cognizant of all new application protocols being developed. And, through it all, manufacturers have to be able to support the new technology and applications with products that can transmit the data flawlessly, without interruption or degradation.
The one guarantee that can be made is that speeds and data transmission requirements won't be going down. It is a global economy and end users will capitalize on any advantages they can get in terms of the speed with which data is processed. Lloyd Mariner is chief operating officer of Molex Premise Networks (Hudson, NH).