SAN JOSE, Calif. A group of as many as 30 vendors have taken the first giant step toward defining a standard for long distance networking at 100 Gbits/second. The Optical Internetworking Forum has chosen a modulation scheme in an effort that aims to kick start development on integrated optical modules.
"Carriers are already conducting trials because they want to deploy 100G systems as soon as possible and we are feeling that pressure," said Karl Gass, chair of an OIF working group that helped pick the so-called Dual-Polarization Quadrature Phase Shift Keying (DP-QPSK) technology.
"Choosing a modulation scheme is just the beginning of this project," said Gass, who is also a researcher at Sandia National Labs. "We need to define a whole architecture for a 100G photonics module, so decisions have to be made about issues such as where to draw interface lines between parts of these modules.
"Such issues could take six months of discussion," he added.
The integrated modules will be used in future optical transmission and switching systems that aim to send traffic between nodes as far as 1,000 to 1,500 kilometers apart.
"Companies will bring 100G products to market soon. They will be big, clunky and have relatively high power dissipation, but they will work," said Joe Berthold, former president of the OIF and current 100G project editor. With this spec, "component vendors can see their way to take all the modulators for a 100G transmitter or receiver and put them on one chip," he added.
Berthold described today's high-speed optical modules as "very expensive brass boxes" that include as many as eight demodulators--each in a separate package—along with multiple splitters and other components. "What's needed is investment from the component industry in integrated photonics" for 100G, he said.
"Long-haul dense wavelength division multiplexing has been stalled at 40 Gbits/sec for some time now, even as carriers have pushed hard for an upgrade to 100 Gbits/sec," said Loring Wirbel, director of the EE Times Market Intelligence Unit. "The OIF's decision brings a common approach to modulation methods, and it will also help lower the cost of transceiver components," he added.
The choice of DP-QPSK provides module makers direction in how to tackle the problem. The dual polarization splits the 100G problem into two 50G poles, one vertical and the other horizontal. The QPSK approach splits light on the fibre into four streams corresponding to four phases of light modulated at 90-degree angles.
In this approach systems can encode two bits of data on each symbol. The scheme can ride in a 50-100 GHz width of optical fibre spectrum.
"We are creating an implementation that lets vendors pack 40 to 80 channels of 100G signals into a single fibre," said Berthold, a vice president of network architecture at Ciena Corp. "That could expand the amount of capacity on a single fibre from 1-2 Tbits/second today to as much as 8 Tbits/s," he added.
The OIF is still evaluating forward error correction techniques for its approach. It may also spawn a working group to define digital signal processing techniques for 100G receivers. Ultimately members hope to develop a multi-source agreements that define the transmit and revive modules.
The modules could carry 100G Ethernet traffic based on a standard still in development in the IEEE's 802.3ba group. "We are completely complementary with the IEEE effort and depend on them to create a standard for signals we can capture and modulate," Berthold said.
If the IEEE group picks a scheme that uses four 25 Gbit/s channels, it will map easily to the OIF standard. But the OIF module architecture will work with whatever approach the IEEE defines, he said.
The OIF approach will also work with a new level of the optical hierarchy in development at the ITU-T. The so-called OTU-4 spec in development aims to encapsulate 100G Ethernet traffic.
As many as 50 people have participated in OIF calls on the 100G spec which started in May. They represent as many as 30 companies including carriers, systems and component makers.
"We have selected an implementation approach supported by a critical mass of photonic component vendors and users," said David Stauffer, who chairs the OIF's physical and link layer working group. "Network element vendors have already begun work [on products that will] accelerate the adoption of 100G in long distance [optical] transmission," he said.