Relief for the bandwidth-starved

Th 0105lwfeat03f1

Fiber technology becomes readily available and economically feasible in metropolitan-area networks.


It seems as though in the blink of an eye, the entire world has become starved for bandwidth. Unfortunately, this ravenous desire to do "more over wire" has hit at the same time that telecommunications service providers are tightening their belts.

That's especially true in metropolitan-area networks (MANs). Even with increasing user demands for "more, more, more," economic realities are forcing providers in this market segment to delay major capital expenditures. Many are choosing to do what they can to meet demands by enhancing, rather than tossing out, existing infrastructures. Interestingly, one of the more cutting-edge options in this enhancement process-optical technology-is for the first time becoming both an economically feasible and readily available alternative for MANs. That's largely due to the recent introduction of new laser technology into fiber communications systems.

At the highest level, today's telecommunications infrastructures can be thought of as a series of specialized layers, characterized by topology and geography. The base level of this hierarchy includes transoceanic and long-haul sections for communicating over long distances (e.g., across the Atlantic or across the United States). The next two layers include the metropolitan backbone level, connecting neighboring urban centers (e.g., the cities around San Francisco Bay), and the access level, connecting to business and residential users. Within the user premises themselves, LANs extend the communication fabric to everything from phones to PCs.

Optical technology has already advanced to the point where it could radically influence and greatly improve every part of these infrastructure layers. Today, it is most prevalent in the long-haul arena. However, that is all going to change. Th 0105lwfeat03f1

The infrastructure of telecommunications networks consists of a number of specialized layers transparent to the users and characterized by their topology and geographical implementation. As shown here, the core includes transoceanic and long-haul sections, while the metro backbone and access layer connect to business and residential users. Within the user premises, LANs extend the communication fabric to the end equipment: phones, PCs, appliances, wireless cell devices, etc.

Bandwidth improvements at every level have taken place over the past decade. Unfortunately, most improvements have happened at a relative snail's pace, especially when compared to the potential of the latest fiber-optic technologies. For example, in a traditional transoceanic or long-haul network, available bandwidth could grow by a factor of about 10 every 10 years. With today's fiber-optic transport systems, bandwidth has been pumped up by a factor of 100 every five years. Today's LANs present a similar scenario; traditional LAN topologies can support bandwidth growth at a factor of about 10 every five years. However, fiber increases previous bandwidth availability exponentially.

The biggest bandwidth gaps today appear to be in the metro and access networks. They still lag, by two to four times an order of magnitude, behind the bandwidth growth in the WAN and LAN. Bandwidth in the MAN has been growing at the same rate as older WAN and LAN technologies (a factor of 10 every five years.) That means it lags 10 years behind the current LAN infrastructure and 20 years behind modern WANs. For example, LANs saw the ready availability of 10 Mbits/sec as long ago as 1990, while the MAN had to wait until 2000 to see wide deployment of similar bandwidth.

It's even worse for access networks, which lag 10 years behind the MAN, 20 years behind the LAN, and 30 years behind the WAN in bandwidth enhancement. For example, 1 Mbit/sec was readily available in LAN deployments in 1985. The access network had to wait until 2000 to see wide deployment of equivalent bandwidth.

Today, that translates into a bottleneck at the MAN. The gigabit-per-second traffic being produced in many enterprises and transported over the LAN or WAN gets throttled down to mere megabits per second when it hits the T1 speeds of the copper access pipe in the MAN.

The increased use of optical technology can radically change the way metro networks are built. Fiber access removes the copper access bottleneck. What's more, because of the deregulation in telecommunications in most countries, fiber is now widely deployed worldwide.

But these facts and figures still raise many questions. Is fiber everywhere a real possibility? What about cost? What about existing infrastructures? Could fiber deployed throughout the metro and access infrastructures become the ultimate meal for the bandwidth-starved? For the first time, the answer to this last question is a resounding "yes." Why? Because it is finally possible to smash the price and availability barriers that have traditionally dogged fiber. How? With laser technology.

Recent breakthroughs in laser technology are making it possible to greatly expand the information transmission capacity of a fiber network beyond current optical pump technology at a significantly lower cost. These laser technology advances may well be the single-most important improvement in moving fiber to the MAN and beyond, potentially transforming the underlying way that voice, data, video, and Internet traffic are carried across the world. Th 0105lwfeat03f2

Metro and residential access networks have lagged by two to four times an order of magnitude in the growth of bandwidth in the WAN and the LAN. In the WAN (submarine and long-haul), available bandwidth has been growing in the past by a factor of 10 every 10 years. With fiber-optic transport systems now migrating to DWDM to achieve higher-bandwidth expansion, available bandwidth is now growing by a factor of 10 every five years and even by a factor of 100 every five years. For example, a single-wavelength OC-192 link at 10 Gbits/sec will be upgraded by DWDM systems operating at 40 OC-48 or 80 OC-192 for a total bandwidth of 96 or 800 Gbits/sec, respectively.

Today, the two most common approaches for enhancing the available capacity of a single optical fiber are time-division multiplexing (TDM) and DWDM. Until recently, SONET-based TDM deployment has been the preferred choice for increasing information bandwidth. However, developments in optical-component technology, such as optical amplification, passive optical devices, optical switching fabrics, and tunable lasers, have made DWDM systems increasingly attractive.

Fiber-optic transport systems are now migrating to DWDM to achieve greater bandwidth expansion. For example, a single-wavelength SONET link at OC-192 (10 Gbits/sec) will be upgraded by DWDM systems operating 40 OC-48 (2.4-Gbit/sec) wavelengths or 80 OC-192 wavelengths for a total bandwidth of 96 or 800 Gbits/sec, respectively.

Today, DWDM links are powered by transmitter lasers and fiber amplifiers using laser pumps. Most of the transmitters use discrete frequency sources with costly external high-speed modulators; as the number of channels grows, so does the number of sources and modulators. Most of the pumps use edge-emitting diode-laser technology. Such offerings are limited to about 250 mW of reliable power in singlemode fiber, which has been a tremendous barrier-the larger the number of DWDM channels used, the greater the demand for pump power.

Herein lies the latest breakthrough. Using a completely different approach, new laser technology has solved the "power pump" problem found in DWDM amplifiers. These new laser approaches make it possible to produce a much higher power (300 mW), perfectly circular beam that emits from the surface of the chip though a larger aperture. That compares with the hard-to-use elliptical beam and much smaller area of the edge-emitters and allows cheaper optics and easier alignment to fiber, which is circular in cross section.

It is also twice as powerful as current optical laser-pump technology and significantly increases the efficiency of testing. In fact, new laser technologies make it possible to "wafer test" chips before costly packaging by probing the laser while it is still on the gallium arsenide wafer.

Another advantage of the new technology is that it lends itself to the production of tunable lasers for use in transmitters, optical add/drop multiplexers (OADMs), and/or optical-switching systems. For example, these new lasers eliminate the need for multiple backup fixed-frequency lasers. Single digitally tuned transmitters can be used instead.

These new lasers increase the flexibility and options of OADMs, as well. Current OADMs use pre-provisioned arrays of lasers, each dropping a fixed frequency. In the new OADMs, tunable lasers can drop on demand any number of channels at any number of frequencies.

Further, fast tunable chips have the potential to revolutionize the architecture of the entire communications infrastructure by making possible wavelength routed and switched packet-based services. That will drastically reduce the cost of a fiber network while still improving the overall performance of a packetized network. The advent of new semiconductor laser technology paves the way for the universal deployment of fiber.

The days of fiber being relegated to the long haul are gone. Universal hunger for voice/data/video bandwidth can only be satiated by fiber, and that means it is moving everywhere, especially in the MAN. Thankfully, lasers and laser technology can come together to solve the price and availability issues that previously kept fiber out of this arena. The union of fiber and new laser production techniques is a promise that those craving bandwidth will never go hungry again.

Vincent Schmidt is director of business development at Novalux (Sunnyvale, CA).

More in Network Design