George- The Gigamux uses a passive approach. Below is an excerpt from a recent article from EE Times:
WDM-or, more specifically, dense WDM-originated as a long-distance telecommunications technology connecting one central office to another, said Ali Abouzari, manager of global telecommunications for the optoelectronics components business of Lucent Technologies' Microelectronics Group (Allentown, Pa.). In the past year or so, however, efforts have begun to adapt it to lower data rates for metropolitan areas. Some proponents of DWDM think that it has the most growth potential among optical technologies, compared with previous attempts at fiber-to-the-desk or fiber-to-the-home.
"The problem with both of those is that they attempted to reach price parity with copper, and they failed," he said. "DWDM, however, is an all-optical technology" that can leverage the installed fiber base.
The existing 2-Gbit/s fiber is limited to about 2.5 Gbits/s because of electronic-component speed limits. To get more traffic down that fiber, an alternative to laying more fiber is a parallel approach, in which additional 2.5-Gbit links can be established over the same fiber to make "virtual fibers," said Don Buell, GigaMux product manager for Osicom Technologies Inc. (Santa Monica, Calif.)
For smaller carriers, the expense of laying new fiber can be prohibitive, ranging from $70,000 to more than $500,000 per mile, "depending on whether it's in the country or downtown Manhattan," said George Ballog, vice president of sales and marketing for Lightwave Microsystems (Santa Clara, Calif.)
The problem WDM is solving in the long-haul transmission range of 10 to 30 km is very real as far as cost goes, said Bill Woodruff, director of Sonet/SDH products for Vitesse Semiconductor Corp. (Camarillo, Calif.). But it's not the answer for everyone.
"Below 10 km, if there's a conduit in place, it's probably cheaper to install another fiber," he said. "But over 30 km, it's probably cheaper to use WDM. In between, it's situation-specific."
Most current WDM systems convert from the optical domain to the electrical domain, involving a fairly high semiconductor content.
Osicom's DWDM GigaMux product plugs into the network backbone, providing more trunks for gigabit-speed packet multiplexers to use. Its inputs and outputs are user-selectable down to OC-1 speeds of 52 Mbits/s and consume a single channel, regardless of speed. Individual port speeds can be upgraded separately without reconfiguring the GigaMux.
Since each port is independent of the others, it can carry a different type of traffic-for example, video on one, Sonet on another and Fast Ethernet on a third. In general, Sonet multiplexers are larger, more complex and costlier than DWDM multiplexers because of the software support needed, said Buell.
New breed
DWDM components are primarily a new breed of laser built to transmit in a very narrow spectrum as well as filters that can select just one wavelength, said Buell. Component vendors use entirely different schemes for building those filters. Osicom's is a passive frequency-oriented method. Others include refraction grids, temperature-controlled ovens and active components, but such methods require redundancy to be built into the systems.
A passive method is more conducive to deploying DWDM in metropolitan areas, Buell claimed, because only a single-user channel, not entire links, can fail. Osicom does use transmitters, but only on a channel card, and its receivers are avalanche photodetectors.
A time-division multiplexing (TDM) multiplexer's top speed is limited by the speed its components can handle, but a frequency-division multiplexing (FDM) system, such as DWDM, is not as limited, as long as its component technology is passive, as are Osicom's light combiners, according to Buell.
The main question regarding any WDM components is whether they can be manufactured in volume at reasonable yields, said Matt Steinberg, senior analyst for Ryan Hankin Kent (South San Francisco, Calif.). So far, the only vendor that can do that with polymer-based optical components is Akzo Nobel, a Dutch company making optical switches, he said. All of the newer techniques, including those of Bookham, Lightwave and LightChip, have several applications in optics and optical networking, not just WDM, and all are using technologies based on silicon-wafer-processing techniques. "The real kicker there is, if they can build it and get volumes, then they should also have low cost," he said. "Cost will be especially important for the metropolitan-network applications, where the number of users to share the cost of equipment is much lower than in a long-haul network."
The next stage in DWDM will be add/drop multiplexing, which combines passive optical multiplexing with optical switching, to direct the multiple WDM channels, said Ballog. Lightwave has combined optical design with a polymer-materials technology and semiconductor techniques to make waveguide structures directly on a wafer. The silicon acts as a platform only.
The material acts as an optical pipe, but it can also be made optically active by applying a voltage across the material to change the index of refraction. Then the light can be modulated or switched from one path to another, or just modulated at very high speeds. Switching speeds are at 40 GHz or higher, he said. The main objection to using polymers in optical devices has been lack of temperature stability. Lightwave's polymer is a polyimide, with a low dielectric and high temperature stability, because of its very high glass transition temperature, which means it can remain stable above solder temperatures, Ballog said. |
