For fast communication, divide and conquer

Hank Hogan, hank.hogan@photonics.com

By splitting up one big, tough problem into many smaller and easier parts, researchers at the University of California, Davis, may have created the basis for ultrafast optical communications. Team leader S.J. Ben Yoo noted that the new approach could offer speeds in the multiple-terabits-per-second line rates, orders of magnitude faster than is now possible.

“The device can eventually scale up to 10,000 times the speed of standard electronics at approximately 10 GHz,” Yoo said.

The new technique combines spectral slicing with parallel detection or generation of an optical signal. Done in one direction, devices employing this approach could perform real-time optical waveform measurement, useful for investigating ultrafast optical phenomena or receiving an incoming signal. When run the other way, those devices could be used for transmission, creating a communication link.

For researchers, the method solves the problem of how to measure full-field ultrafast optical waveforms. In making such measurements, it’s not enough to capture intensity changes. Phase must be measured as well. Drawbacks of existing methods that do both are that they update too slowly to gauge what is happening on subpico-second timescales, require too much pulse energy or can measure only for a few nanoseconds’ duration.

Yoo credited the idea for a new solution to graduate student Nicolas K. Fontaine, lead author of a Nature Photonics paper covering the group’s work and published online on Feb. 28, 2010.

In it, the researchers described their scheme and demonstrated it using a silica planar lightwave circuit and balanced photodiodes. The first component sliced an incoming arbitrary light waveform into separate spectral segments. The researchers combined those slices with a coherence-ensuring reference signal from an optical frequency comb, with one reference signal per spectral slice.

Using this silica planar lightwave circuit, researchers divided an incoming arbitrary optical waveform into more manageable parts. The approach could form the basis for terabits-per-second line rates. Courtesy of S.J. Ben Yoo, University of California, Davis.

They fed everything into the second part, the photodiodes, using four-quadrature spectral interferometry for measurement. Finally, they took that output, sent it through CMOS digital signal processing circuitry and reconstructed the original incoming signal.

After building the prototype, they tested it, generating an array of complex and different waveforms in the 1.55-μm telecommunications waveband. The test waveforms included static, rapidly changing, bright and dark varieties, Yoo said. “The measurement system works extremely well and captures complex waveforms across a wide dynamic range.”

In doing this, they demonstrated an instantaneous bandwidth >160 GHz, with record lengths of 2 µs. This length-to-resolution ratio of more than 320,000:1 represents the largest of any single-shot, full-field measurement technique, the group stated in the paper.

Yoo noted that the technique could be used for a number of applications, including for ultrahigh-speed or highly secure optical communication, lidar and multicolor coherent femtosecond spectroscopy. Of these, he said that high data rate communication is likely to be the most compelling.

One advantage of the new approach, aside from speed, is that it potentially can be done in a single-silicon photonic integrated circuit. Being able to achieve this will make the resulting chips more attractive and useful for all applications.

The process of integration will involve two steps, the first being the merging of the silica planar lightwave circuit and balanced photodiodes into a hybrid-integrated chip. That intermediate integration, Yoo said, the group could do relatively soon.

The second step will involve combining the digital signal processing circuitry with everything on the first intermediate chip. That involves making the fabrication of the photonic components process fully compatible with large-scale CMOS manufacturing.

But it will be worth the additional work, Yoo predicts. “Integration of both photonic and electronic circuits on the same silicon CMOS platform opens new possibilities.”