|

Imran Hayee
PHOTO BY BRETT GROEHLER
|

Hayee's research team Standing: Imran Hayee and Vincent
Zhang
Sitting: Rami Haddad
PHOTO BY BRETT GROEHLER
|
|
Virtually everyone who has ever gone online has been caught in
a Web traffic jam. Half-loaded pages and frozen computers litter
the shoulders on the information superhighway, and the traffic that
engenders the slow-downs is only increasing as consumer demand for
high-bandwidth services--graphic-rich Internet, audio and video
streaming, and online gaming--continues to intensify.
The ubiquitous optical fibers that line the walls and floors of
our buildings and snake beneath our roads and under the oceans are
the primary highway for these services. They transmit information
as light pulses along glass or plastic wire or fiber, and according
to Imran Hayee, associate professor of electrical and computer engineering
at the University of Minnesota, Duluth, they are capable of meeting
the increasing demands. "Relatively speaking, optical fiber
is unlimited in its capacity, if it can be optimized," said
Hayee. "There is a limit, but it is above our current and near-future
needs."
Those needs include helping to transmit signals from wireless devices
like cell phones and PDAs. Wireless networks currently carry signals
for a relatively short distance before switching the signals over
to optical fiber networks, which are quicker and less expensive.
Ultimately, the explosion of wireless devices, along with the growth
of voice over internet protocol (VoIP) and video on demand (VOD),
will further add to the traffic on fiber-optic lines.
Five years ago optical fibers could carry 2.5 gigabits of information
per second (Gb/s) on each of many wavelengths (colors) of light.
Today they carry 10Gb/s on each wavelength with a few unused wavelengths
to spare. But the current 10Gb/s system is like a crowded highway
in a booming metropolis with car pool lanes providing the potential
for additional capacity. Ultimately, a new system will need to be
built--imagine a new highway around the expanding suburbs and a
commuter rail line through the densest areas. Unfortunately, that
upgrade is many years and billions of dollars in the future. "To
fully upgrade today's telecommunication infrastructure it will be
necessary to replace thousands of kilometers of existing fiber and
amplifiers in the ground with new fiber and new amplifiers,"
said Hayee.
Until then, Hayee's research aims to find a way to coax the existing
fiber-optic system to carry 40Gb/s on those few spare wavelengths
of light. If he is successful, this system should be able to handle
traffic for years to come and meet the incredible expectations of
consumers. "Ultimately, we want our computers to be like a
crystal ball," he said. "We want to look into our monitor
and see in time and space [information from past to present and
from servers everywhere on the planet]. Optical fiber networks can
do this."
Keeping a finger on the pulse distortions
Until the late 1980s, optical fibers transmitted data on a single
optical channel using transmitters and receivers to send and receive
the signals and amplifiers to boost the signals as they deteriorated
en route. But improved transmitters, better amplifiers, and the
introduction of devices capable of discerning different wavelengths,
allowed the system to handle more capacity. Wavelength-selective
devices use laser-light wavelengths as signature addresses of origin
that can be routed and recovered out of the chaos of multiple signals
allowing for the transmission of many independent signals simultaneously
on one fiber.
Hayee's research focuses on the use of wavelength dimension multiplexing
(WDM), a technique in which each of the multiple wavelengths of
light carry the same amount of information in parallel on a single
hair-width fiber without interfering with each other. There are
other ways to expand bandwidth in the current system, but WDM has
the ability to use more of the latent potential in the existing
fiber.
According to Hayee, the major hurdles to reaching 40Gb/s WDM transmission
on optical fibers are fiber dispersion and nonlinearity, both of
which cause signal degradation. Fiber dispersion is the spread of
optical laser pulses in the fiber over time, causing pulses to interfere
with neighboring pulses of the same wavelength. Fiber nonlinearity
manifests itself as severe pulse distortion by causing multiple
wavelengths of light to interact with each other, thereby deteriorating
the information bits on multiple channels.
Hayee's team is proposing to combat these difficulties through a
technique (nodal dispersion mapping technique) where a specific
amount of predispersion compensation is added at the transmitter
end and a specific amount of post dispersion compensation is added
at the receiver end--like buoys in no-wake zones guiding ships out
of and into ports. This allows for the control of expansion and
contraction of 40Gb/s pulses, minimizing nonlinear distortion and
preventing the pulses from interfering with each other.
"The initial research will use computer modeling and simulations,"
said Hayee. The next step will be to build a transmission test bed
in the laboratory to confirm the results of the computer simulations.
Fortunately Hayee has significant experience with this type of effort,
having worked in private industry for five and a half years developing
telecommunications products and techniques to improve the performance
of trans-oceanic optical fiber communications systems. He holds
more than a dozen U.S. patents (issued and pending), and he is continuing
along the same research path in his UMD lab as he pursued in the
optical telecommunications industry.
If all goes well, this research can bridge the gap between network
capacity and burgeoning demand until a new network is installed,
helping ease the traffic that frustrates us all.
WRITTEN BY BRIAN LIEB
|