shows fiber has room to grow
Technology Research News
Although fiber optic lines have been in
commercial use for many years, their top theoretical capacities have always
remained rather murky. This is because many optical signals of slightly
different frequencies, or colors, can flow over one fiber at once, and
it's difficult to predict the effect of the different frequencies interfering
with each other.
A pair of researchers from Lucent Technologies' Bell Lab has simplified
the problem in order to make a prediction. Their method shows that fiber
capacities can theoretically increase by an order of magnitude over today's
fastest links. The method may also provide insight into the way biological
When a string of signals traverses a communications channel made of wire,
the signals fade in a mathematically predictable way, slowly degraded
by random noise. In optical fiber, however, the noise is due to crosstalk
from other channels. Instead of having an additive, or linear effect,
the various crosstalk noise signals interact, changing the signal pulses
in more complicated ways.
This cocktail party environment, which randomly changes the speed of various
parts of the signals, has historically proven difficult to predict. "The
expected output when two [signals] are simultaneously present is not the
sum of the expected outputs when the signals are transmitted individually,"
said Partha Mitra, a physicist at Bell Labs.
The researchers came up with a way to characterize the effect mathematically
by changing the problem into one that was linear and could therefore be
worked out. They essentially calculated the multiplicative effects of
crosstalk on a linear signal. "The actual set of ideas we used in the
paper emerged through discussion of what happens to a set of initially
monochromatic beams propagating through the fiber in the presence of a
background of... channels," said Mitra.
"The nonlinearities caused the crosstalk. The effect of the crosstalk
is to randomly modulate the propagation speed of the channel. The simplification
we made... was to [use] a linear channel, but one that [was modulated
in complicated ways] due to the information being carried by the other...
channels," he said.
This simplification made it possible for the researchers to estimate the
peak spectral efficiency of fiber, which in turn allowed them to estimate
its top capacity based on today's top bandwidths and how far the information
needs to travel. Spectral efficiency is a measure of the top speed that
pulses, or bits of information, can travel down a fiber-optic line, divided
by the amount of wavelength space between the optical information channels
that carry slightly different colors of light.
According the researchers' calculations, it is theoretically possible
to put more than an order of magnitude more information through existing
lines before running into physical limits due to crosstalk. "The peak
spectral efficiency from normal fiber is three bits per second per hertz,"
or wavelength cycle, said Mitra. "This leads to a capacity estimate of
150 terabits per second."
Today's fiber lines have a top capacity of about 1.6 terabits of information
per second using 160 channels in a single fiber; speeds of ten terabits
per second have been demonstrated in the laboratory.
The trick to sending more information over fiber lines is striking a balance
in signal power. The more powerful a signal is, the less likely it is
to degrade quickly, but the more likely it is to produce crosstalk. This
is very different from sending information over a wire line, where boosting
the power simply increases the time it will be able to travel before degrading.
Because our eyes are essentially lightwave sensors, they encounter some
of the same issues; scientists looking into the way vision systems work
could also benefit from the crosstalk calculations, said Mitra. "That
research will benefit from better understanding of the qualitatively new
effects that arise from having nonlinearities in the channel, which, needless
to say, are present in the nervous system," he said.
Considering the effect of fiber nonlinearities is something that has not
been done before and is very useful, said Amnon Yariv, a professor of
applied physics, engineering and applied science at at the California
Institute of Technology. "The common wisdom is that when one elevates
the signal power one can propagate farther or increase [the] data rate
for the same distance. This is not true at the power levels employed today
and the practical implications are very important," he said.
The researchers are now working on designing more efficient systems based
on the research. "Our next step is to use the increased understanding
to design better systems, for example with higher spectral efficiencies,
or to use the available phase space to optimize [something else]," he
The research could have an immediate impact in terms of how people think
about designing systems that propagate optical signals, said Mitra. "I
think it will lead to better understanding of modulation and coding schemes.
In practical terms, it means that optical fibers are a good way to build
scalable networks, because the capacities can be increased substantially
by adding appropriate equipment," he said. It's technically possible to
make substantial progress towards the ultimate in spectral efficiencies
"on a several year time scale," he added.
Mitra's research colleague was Jason B. Stark. They published the research
in the June 28, 2001 issue of the journal Nature. The research was funded
by Bell Labs.
Timeline: Now; 3-5 years
TRN Categories: Optical Computing, Optoelectronics and Photonics;
Story Type: News
Related Elements: Technical paper, "Nonlinear Limits to
the Information Capacity of Optical Fiber Communications," Nature, June
Tool reads quantum bits
Study shows fiber
has room to grow
Search tool builds
atoms advance quantum chips
Electron beam welds
Research News Roundup
Research Watch blog
View from the High Ground Q&A
How It Works
News | Blog
Buy an ad link