Split pulses speed signals

By Kimberly Patch, Technology Research News

Optical communications, which already far outstrip the speed of electrical signals, have been getting even faster as researchers find ways to make laser pulses shorter and more frequent. These pulses, and spaces between them, represent the ones and zeros of digital information.

Engineers from Purdue University and NTT Photonics Laboratories in Japan have figured out how to make light carry information faster using a device ordinarily used to sort lightwaves into different colors, or wavelengths.

Waveguide gratings were commercially developed to sort out the wavelengths of slightly different colors that are used to send many sets of pulses over the same fiber line at the same time. The different color wavelengths are slightly different sizes and the commercial waveguide grating is used to sort them, after their travels, into discrete sets of pulses of the same size.

The waveguide grating consists of a fiber-optic line that shunts a light pulse into a flat glass slab where the light fans out. Twenty curved fiber-optic lines connect that slab to a second slab. At the other end of the second slab, some number of fiber-optic lines carry the pulses out.

In its commercial use, the curved fiber-optic lines sort out different wavelengths because their different sizes are tuned to the spectral, or color ranges of the wavelengths. "A particular output will have a certain wavelength, and a free spectral range later it will have another wavelength," said Daniel Leaird, a research engineer at Purdue University. The lines out carry the signals sorted by wavelength.

Leaird and his colleagues, however, have found a way to use the device to speed optical signals by turning one pulse into a train of 20 pulses. "These devices have been thought of [as] wavelength filters," said Leaird. "The inverse of.. spectral range is the time delay between guides... it's completely valid to think of [a waveguide grating] as just an array of delay lines."

The upshot is the waveguide grating can not only sort different colors in order to increase the bandwidth, or amount of information a line can carry at once, but also has the potential to increase the frequency of the pulses so that a line can carry information more quickly.

In the researchers' device each output channel carries a train of 20 pulses made from the one pulse coming in. This is possible because each line in the device is slightly longer than the next, and the light takes slightly longer to get through each successive line. "You can think of [the lines] as a series of arcs, and they get longer as you go to each adjacent guide," said Leaird.

The difference in length results in a rapid fire of 20 distinct pulses as they come out the other side, to the second flat glass slab, which passes on the 20-pulse train to all of the output guides. "If you have 10 output guides, that means you have 10 trains of 20 pulses," said Leaird.

In order to make the device work, the researchers had to start with an input pulse with a width, or duration, that was shorter than the delay resulting from the difference in length from one curved guide to the next. Otherwise, the pulses would overlap, merging back into a single, wider pulse.

The researchers made two types of pulse trains at speeds of 500 gigahertz and one terahertz. Commercial optic systems use 10-gigahertz optical signals. Ten gigahertz is 10 billion times per second while one terahertz is one trillion times per second.

It would take 317 years to transmit at a speed of one pulse per second the amount of information it takes a 10 gigahertz channel to transmit in one second, and 31,710 years to transmit at a speed of one pulse per second the information that goes through a one terahertz channel in one second.

Some lasers used in research create much shorter pulses than those produced by the Purdue-NTT device, but those lasers are too difficult and expensive to use for the communications networks. The researchers' approach could provide a practical method for speeding up commercial-grade communications lasers.

The researchers' work however, is just one step toward that end. In order to transmit actual information, the researchers need to be able to mix on and off pulses. "We'd like to go in and turn off individual [pulses] to make not just the clock source generation but actual data transmission. It's a pretty difficult problem but we have some ideas on ways to [use] this kind of arrayed waveguide grating technology... with high-speed modulator arrays," Leaird said.

Once this work is done, the device could be used to group slower electronic signals into one very high-speed optical channel, said Leaird. Today's fastest electrical signals, which use currents of electrons rather than photons, are about 100 times slower than the fastest commercial optical transmissions.

A super high-speed optical channel could also be used to transfer information between different parts of a supercomputer, said Leaird.

"It's nice work," said Warren S. Warren, a professor of chemistry and director of the Center for Ultrafast Laser Applications at Princeton University. "These are very interesting devices [that] have a variety of potential applications for making these very high repetition rate [pulses] from low-frequency lasers," he said. The work could be a way to cut down on the cost of building high repetition rate light sources, he said.

The device may also eventually prove useful in analog-to-digital conversion in order to record the motion of very fast events, said Leaird. For example, researchers are aiming to use ultrafast light sources to record the movement of individual molecules. Analog signals, because they're not made up of discrete ones and zeros like digital signals, must be broken up to be converted. Digital signals, like frames in a movie, will always be stop-motion samples of analog signals. The higher the sampling rate, however, the better the digital signal represents the analog signal.

"With shorter and shorter pulses, faster and faster rates... you can stop shorter and shorter events," said Leaird.

It will be about 10 years before waveguide gratings can be used for commercial purposes, Leaird said.

Leaird's research colleagues were Andrew Weiner from Purdue University and Shuai Shen, a Purdue graduate student who is now at Lucent Technologies, and A. Sugita, S. Kamei, H. Yamada, and M. Ishii and Katsunari Okamoto from NTT Photonics Laboratories in Japan. They presented the research at the Conference on Lasers and Electro-Optics in Baltimore on May 5, 2001. The research was funded by the Army Research Office.

Timeline:   10 years
Funding:   Government
TRN Categories:  Optical Computing, Optoelectronics and Photonics
Story Type:   News
Related Elements:  Technical paper, "High Repetition Rate Flattop Pulse Trains from an arrayed Waveguide Grating," presented at the Conference on Lasers and Electro-Optics (CLEO) in Baltimore, May 5, 2001.


July 4/11, 2001

Page One

Split pulses speed signals

Gender gap shows cyberspace bias

Software lets appliances speak softly

Molecular shuttle gains light throttle

Light-sensitive memory does not fade


Research News Roundup
Research Watch blog

View from the High Ground Q&A
How It Works

RSS Feeds:
News  | Blog  | Books 

Ad links:
Buy an ad link


Ad links: Clear History

Buy an ad link

Home     Archive     Resources    Feeds     Offline Publications     Glossary
TRN Finder     Research Dir.    Events Dir.      Researchers     Bookshelf
   Contribute      Under Development     T-shirts etc.     Classifieds
Forum    Comments    Feedback     About TRN

© Copyright Technology Research News, LLC 2000-2006. All rights reserved.