Free30Dayqs468x60
Magazine Events Forum Research Careers Company Profiles Women About Us
 

cover story
editorial
efocus
industry psychology
insider
leadership
legal advice
profile
science & technology
technology editorial



By Majid L. Riaziat Ph.D

 

One bit of information is a yes or no answer to any question. Telecommunication networks transmit bits of information by sending electronic 0 and 1 signals. These signals are used to reconstruct voice, images, etc. at the destination point. A typical conversation is transmitted through the telephone line by 64,000 bits of information per second (64 kbps). In contrast, a TV channel uses approximately 6,000,000 bits of information per second or 6 Mbps (A picture is worth a thousand words!).

Telecommunication networks transmit telephone conversations, FAX, cable TV, Internet data traffic, etc. They use copper wires, radio waves, microwaves, or more recently, laser light traveling in an optical fiber. Each one is known as a transmission medium. Every telecommunication network has a network capacity that is the maximum bit rate it can transmit. Network capacity depends on the medium and the technology used. Table 1 compares the capacities of a number of communication technologies.

Telecommunication networks consist of multiple communication links. Each link has a transmitter, a receiver, and a transmission medium. Signals traveling between the transmitter and the receiver are attenuated by the medium and may need to be amplified or even fully regenerated along the way. Amplification and regeneration add cost to the installation and operation of the link. Optical fiber is the lowest-loss guiding medium available today. Infrared light at the wavelength of 1550 nm experiences the minimum attenuation of approximately 0.25 dB per kilometer of propagation. Figure 1 shows the attenuation of infrared light through silica fiber as a function of its wavelength. Telecommunication companies that historically provided telephone service, and even those that provided cable TV are increasingly relying on fiber links for information transmission


Table 1. Some telecommunication technologies and teh approximate capacity associated with each.

Depending on transmission distance and the clients served, fiberoptic networks are referred to as submarine, long-haul, metro, access, or enterprise. Long-haul networks typically transmit information through a few hundred to a few thousand kilometers. Metro networks serve metropolitan areas and employ links of a few tens to a few hundreds of kilometers. Access networks span a few kilometers between a central office and the consumer or the commercial customer. Finally, enterprise networks provide connectivity within a building or multiple sites of a company.

Development Since the 1990’s
The dramatic increase in data traffic that started in early 1990’s led to a revolutionary growth in fiber optic telecommunications. Prior to 1990’s, network capacity was driven by voice traffic that grew at the rate of 4 to 7 percent per year. In contrast, since the late 90’s, data traffic has been growing at the estimated annual rate of 70% to 200%. In year 2000 for the first time data traffic volume in the US overtook voice traffic. Even in the current economic downturn, data traffic continues to grow at the approximate annual rate of 70%. It should be pointed out, however, that as of this date, voice traffic is still where most telecommunication companies make their profits. Residential data traffic in particular is not a moneymaking venture.

The bulk of the growth in long-haul telecommunications has been made possible by advances in fiber optic technology. Specifically, the adoption of dense wavelength division multiplexing or DWDM (discussed below) has played a crucial role. Currently, with DWDM technology, a single strand of optical fiber is capable of carrying a data rate equivalent to approximately five communication satellites (Table 1). The capacity of the optical fiber can easily increase by another factor of ten within the next few years. Optical fiber provides the largest capacity and, in many cases, the most economical medium for long-haul telecommunications.

In order to appreciate the global pervasiveness of optical fiber, consider the fact that by the end of year 2000 the total length of installed fiber in the world grew to 180 million kilometers. This is enough to circle the earth 4,500 times.

Since early 1990’s the information carrying capacity of a single optical fiber has increased by a factor of more than 100 (Figure 2). The explosive growth of Internet data traffic created the demand for this increased capacity. In 1992 the code for the World Wide Web was first made available, and by 1996, the number of Web sites had grown to over 230,000. It was apparent that the Internet data traffic would grow much faster than the speed of the electronic circuitry that had previously supported capacity growth (Moore's Law). Data traffic growth required major innovations rather than incremental electronic improvements. These major innovations came in the form of DWDM and EDFA.

DWDM and EDFA
Wavelength Division Multiplexing, WDM, or its dense version, DWDM places a number of discrete wavelengths or optical channels from multiple lasers on a single optical fiber. Typically, the laser light sources operate in the low-loss 1550 nm infrared window of the optical fiber, and have narrow enough spectral widths to occupy the same fiber without interfering with adjacent wavelengths. Adjacent wavelength spacings have been shrinking over the years. Currently, 100 GHz spacing is typical. In 1989, a WDM system was deployed with two channels carrying 3.4 Gbps per fiber. Today’s standard DWDM employs over 128 wavelengths at 10 GB/s for 1.28 Terabit transmission over a single fiber. This equals 165 million equivalent voice circuits.

The key enabler of DWDM was the EDFA or the Erbium-Doped Fiber Amplifier that first appeared on the market in 1993. EDFA amplifies the optical signal in multiple DWDM channels simultaneously without having to convert them into electrical signals first. EDFA eliminates the need to have a large number of expensive regeneration centers. The cost reduction achieved this way makes it practical to combine or multiplex tens to hundreds of channels or wavelength onto a single optical fiber. The typical wavelength range of the EDFA defines the “C” band in fiber optic communications shown in Figure 1.


Figure 1. Optical communication bands currently in use. the solid curve is the typical attenuation of infrared light per kilometer of propagation in silica optical fiber as a function of wavelength.

Technology Trends
In order to meet the growing demand for transmission capacity, three major areas of development are being actively pursued concurrently: (1) Increasing spectral efficiency or information transmitted per bandwidth used; (2) expanding utilized optical bandwidth; and (3) extending transmission distance between signal regenerations. These areas are addressed individually in the following discussion. The question of transparent vs. opaque networks is also addressed.

Spectral Efficiency:
The one terabit per second capacity of optical fiber cited above (100 channels at 10 Gbps data speed per channel) occupies 10 terahertz of bandwidth. This corresponds to a spectral efficiency of 0.1 bps/Hz. The practical limit of spectral efficiency using current protocols is thought to be 3 bps/Hz. Therefore, capacity enhancement by a factor of 30 can be achieved by improved spectral efficiency alone. One way to improve spectral efficiency by a factor of 4 is to increase the data speed per channel from 10 Gbps to 40 Gbps while maintaining 100 GHz channel spacing. Major experiments and field trials are currently underway to accomplish this network upgrade. Figure 3 shows optoelectronic integrated circuits designed to work with 10 and 40 Gbps optical links.

Utilized Bandwidth:
Utilized communication bands of the optical fiber are the medium and long wavelength bands shown in Figure 1. The most efficiently used band is 40 nanometers around 1550 nm wavelength known as the C band. Wavelengths close to 1400 nm are not used due to the absorption peak caused by residual moisture in the fiber. Wavelengths around 1300 nm are sparsely used primarily for metro applications. L and S bands in the long wavelength region will be fully utilized in the near future as new fiber amplifiers become available. The water absorption peak will be eliminated in future fibers. There is no fundamental barrier against using the full bandwidth of 1000 to 1700 nm. This wavelength span is over 80 times what is utilized today. Combined with improved spectral efficiency, fiber capacity can grow by over 2000 times with incremental advances in technology. If and when this potential is realized will depend on economic factors.


Figure 2. Laboratory demonstration of fiber capacity and the product of fiber capacity times distance since 1993.



Transmission Distance:
The presence of dispersion and cross-talk place limits on transmission distance between signal regenerations. Since higher data rates are more severely affected, a tradeoff exists between system capacity and transmission distance. In order to gauge the progress in optical network technology, it is often the capacity times distance that is monitored. Figure 2 shows the highest capacity times distance products achieved in laboratory experiments since 1993. Note the improvement by a factor of 100.


Figure 3. 10 and 40 Gbps optoelectronic integratea circuits fabricated at OEpic, Inc.



Opaque vs Transparent Optical Networks:
There is a definite advantage in avoiding electronic regenerations in the network. Electronic regeneration is expensive and the equipment needs to be upgraded every time data transmission speeds increase. For this reason some researchers would like to see networks that are fully “transparent”. This means that optical information is not converted to electronic information anywhere between initial generation and final destination. All switching and routing is done in the optical domain. As a consequence, the system is insensitive to data rate. On the negative side, signal impairments accumulate along a transparent network. This requires sophisticated centralized planning of each link. Furthermore, a transparent network still lacks system intelligence, i.e., no information is available about data traffic. Performance monitoring and fault detection are quite complicated.

At the other extreme is the “opaque” optical network where electrical regenerations are abundant. All system functions including traffic monitoring and fault detection are done electronically in a straightforward manner. The system is expensive and resistant to data rate upgrades. This was the only architecture available in the early days of optical networks. Today’s optical networks are a “hybrid”, meaning that there are “islands of transparency” but still many system functions are performed electronically. There is a steady trend toward generating more transparency. However, practical communication systems are likely to remain of the hybrid type for the foreseeable future.

Cost
Optical fiber is made of silica glass and costs much less than copper wire or coaxial cable used in telephone and cable TV transmission. Also, as data rates per fiber increase, the price per bit of transmitted information drops. Nevertheless, the price of a broadband connection has not dropped far enough yet for the average household to subscribe to a fiber connection. It is envisioned that a 100 Mbps connection to the home will satisfy the entire household’s multiple-service information needs, including telephone, Internet, and video-on-demand for the foreseeable future. The average household is willing to pay approximately $100 per month for such a connection. Currently, a provider’s cost to deliver this bandwidth is between 50 to 100 times higher. One can therefore conclude that once cost reduction by a factor of fifty is achieved, a very large new market will open for fiber-to-the-home services. Fiber-to-the-home will provide a versatile and scalable conduit for bi-directional transmission of all types of information.

Conclusion
We are observing the beginnings of a revolution in fiber optic communications. The enormous information carrying capacity of the optical fiber has been known for a long time but was not demonstrated in field implementations until the mid 1990’s. It was during the last decade that the telecommunication fiber surpassed the communication satellite in information carrying capacity. We are still far from reaching the capacity limits of the optical fiber. An additional improvement by a factor of 2000 can be achieved without introducing any radically new technology. Cost reduction goes hand in hand with increased capacity, and in the near future, the average household will be able to afford the advantages of a broadband fiber optic connection.
Majid L. Riaziat is the CTO and a cofounder of OEpic, Inc., a manufacturer of high-speed integrated optoelectronic devices for fiber-optic communications. He was previously the director of Central Research at Varian Medical Systems in Palo Alto, California. He has 19 years of experience in technology development and acquisition. Dr. Riaziat received his Ph.D. in Applied Physics from Stanford University in 1983.

Majid L. Riaziat is the CTO and a cofounder of OEpic, Inc., a manufacturer of high-speed integrated optoelectronic devices for fiber-optic communications. He was previously the director of Central Research at Varian Medical Systems in Palo Alto, California. He has 19 years of experience in technology development and acquisition. Dr. Riaziat received his Ph.D. in Applied Physics from Stanford University in 1983.

Oct 2001 - How did Silicon Valley come about?

 

Advertising Form - Subscription Form - Media Kit - Contact Us - Site Map - feedback - Privacy Policy
Magazine - Events - Forum - Research - Careers - Company Profiles - Women - About Us
® 2004 Teksia, Inc. All Rights Reserved.