Global Sources
EE Times-Asia
Stay in touch with EE Times Asia
?
EE Times-Asia > Networks
?
?
Networks??

Vendors ramp products for 40Gbps networks

Posted: 13 Feb 2002 ?? ?Print Version ?Bookmark and Share

Keywords:networking component? optoelectronic? fiber?

Though 40Gb networking components have yet to make their way to market in any appreciable number, optoelectronics experts see few technical barriers impeding a smooth transition from 10Gbps to 40Gbps data rates.

In fact, several companies active in the market, including Mintera Corp., say the transition will be easier than the switch from 2.5Gbps to 10Gbps. They spoke at a briefing organized recently by Information Gatekeepers Inc.

"We have seen major technology changes over the past 10 years," Benny Mikkelsen, vice president of optical R&D at Mintera, told EE Times. "In the upgrade from 2.5Gbps to 10Gbps we have seen in-line dispersion compensation, precise knowledge of fiber parameters [fiber type, dispersion, polarization-mode dispersion] and distributed Raman amplification, which allows for high power in fiber transmission."

Rather than being a question of whether technology is available to realize 40Gb transmission, the real issue is how to do it economically, said Carsten Videcrantz, Mintera's product-marketing manager. "The key message here is that all of the building blocks exist today from several vendors, and at a fairly low cost," he said.

"Take some in-line interfaces?long haul from a router or switch?and then multiplex them into a 40G stream. Then it's able to go very long routes," said Videcrantz, indicating Mintera's basic approach to 40G long-haul systems. "This is definitely WDM, so we can support up to 80 wavelengths in the current system, for a full capacity of 3.2Tb. We are multiplexing four 10Gb data steams into a single 40Gb data stream."

The rationale behind the move to 40Gbps is cost. Particularly, the opportunity to use existing installed fiber at a higher data rate is a large cost-saving opportunity for carriers. At the moment there is a fiber glut in that existing fiber is not being used at anywhere near its capacity while traffic is growing at 100 percent per year, Videcrantz observed. One big opportunity for optical-component makers is the fact that installed fiber is still ideal for advanced concepts such as WDM.

Fiber capacity is following a road very similar to Moore's Law for semiconductors. "Every time you quadruple the data rate you typically pay 2.5-times more for that piece of equipment," Videcrantz said. "So you are saving 40 percent in cost per transmitted bit?that is the major driver for the higher data rate."

At the heart of this four-times increase in capacity is the transponder that converts between electrical and optical signals. "Four 10G client signals come in and we multiplex those signals to a single 40Gb-data-stream [SONET]-type of signal. We then add overhead for performance monitoring so the real line width is 43Gb, not 40Gb," he explained.

The 43Gb data, in parallel form, is then serialized with a mux, amplified and fed to an optical modulator to convert the electrical signal to an optical signal. On the receive side, dispersion-compensation modulators adapt to the dispersion variations in the transmission fiber, then to the PIN diode, clock and data recovery, and then to a demultiplexer.

Engineers at Mintera have found that once component vendors are committed to developing a cost-effective 40G system and see that volume orders are in the offing, they are quite cooperative in dropping the cost of components. "You may hear from system vendors that 40G is too expensive, the technology is not mature," Videcrantz said. "This is not true. If you are a system house today who just wants to sample or wants some info on 40G, you can call a component provider and ask for 40G quotes. You will get the real quotes if the component vendor finds out that you are a serious player.

"There's a lot of hype out there for 40Gb?companies are claiming that they have the technology?but they don't have it," he said. "We have spent a tremendous amount of time figuring out those who have it and those who don't."

That's not to say there are no technical issues facing 40G beyond the building of a transponder. One is the chromatic dispersion in existing fiber.

In fact, dispersion is an essential ingredient in WDM. In dispersion-free fiber, different wavelength channels experience an effect known as four-wave mixing. This phenomenon links the channels together and creates noise. It turns out that chromatic dispersion suppresses four-wave mixing and is therefore essential as the number of wavelength channels increases.

The problem, however, is that the dispersion is not uniform and some special optical processing is required to manage the variations. "Chromatic dispersion and dispersion slope?which means the dispersion in the fiber is wavelength dependent?are not the killers for 40Gb that many claim," Videcrantz said. "Dispersion is a must for 40Gb?if you did not have it you could not transmit at 40Gbps over any significant distances."

While special tunable dispersion compensators exist, the ideal solution is to build that function into the transponder. "You have to design a smart transponder that takes this into account on the transmit part and the receive part," Videcrantz said. "It is key to have a technique in your transponder that adapts to band dispersion. Dispersion might be slightly different depending on which wavelength is used."

In addition, he said, "dispersion is slightly temperature-dependent, which means that a component in the receiver must know how to adapt to temperature changes over the year."

There is, however, another type of dispersion that is more difficult to deal with: PMD, or polarization-mode dispersion. This problem occurs in only a small fraction of the installed fiber, so it is not as critical an issue as chromatic dispersion.

Some companies have built PMD compensators, but the effort may not pay off economically in the end due to the small amount of fiber involved. "We do not believe that PMD compensators are an enabler for 40Gb," Videcrantz said. "There is only a small fraction of the installed fiber out there where PMD is going to be the showstopper. It will come into play at certain routes, where you only have old fibers, but it will be a niche market.

"And six years ago people were claiming that 10Gb would not happen because of high PMD," he said. "Of course, it did happen." Indeed, PMD has been completely eliminated in the new fiber that is now being installed.

Finally, there is the problem of signal degeneration over long distances. While the signal can be easily amplified in the optical domain whenever needed, eventually noise increases and overwhelms the signal. Compensating for that problem requires that data be translated back to the electronic domain, where it can be restored. It then must be converted back to the 40Gb optical stream, a process that is very costly.

Mintera said that its equipment can certainly handle distances of 1,000km without the need for regeneration. "We can now go up to 2,000km in newer types of fiber," said Videcrantz. For example, fiber specifically constructed to control dispersion effects will allow the longer distances. "This proves that 40Gb can go long distances, not simply with research equipment but with real components that are field-deployable today," he said.

Videcrantz noted that every time the optical industry contemplates a fourfold increase in data rate, there is argument about whether it can be done. This happened in the early years of the semiconductor industry, when transistor counts kept growing exponentially.

"Now the debate is whether it is possible to get to 160Gbps?that's something for the researchers?now that commercially deployable 40Gb systems are arriving," he said.

? Gail Robinson

EE Times





Article Comments - Vendors ramp products for 40Gbps net...
Comments:??
*? You can enter [0] more charecters.
*Verify code:
?
?
Webinars

Seminars

Visit Asia Webinars to learn about the latest in technology and get practical design tips.

?
?
Back to Top