|
Not since Alexander Graham Bell's invention of the
telephone has communications experienced such meteoric
or revolutionary development, and nearly all of it has
been made possible by electro-optics.
Ironically, it was Bell himself who invented one of
the earliest light-wave communications devices in 1880.
Bell's "photophone" used a flexible diaphragm to modulate
a beam of sunlight and a selenium photodetector as a
receiver. Current fluctuations generated in the photoconductive
selenium by the modulated light beam were fed through
a transducer to recreate the original sound. But what
made the photophone impractical for long-distance communications
was the incoherent light and its free-space transmission
through the atmosphere. The advent of lasers in 1960
and low-loss optical fiber in 1970 eliminated all these
barriers.
Modulating the Light Source
Fiberoptic communications systems offer a unique set
of solutions (and problems) to the monumental task of
moving mountains of information from one location to
another. However, like any form of communication, the
signal to be transmitted must be encoded onto the carrier
at the source (transmitter) and decoded from the carrier
at the destination (receiver).
When the carrier is a light wave, signal encoding
is physically done either through direct modulation
or external modulation of the light source. For example,
varying the current of a laser diode (and therefore
its light output) is a form of direct modulation.
Within this modulation method exist a variety of signal-encoding
schemes that can be classified as either analog or digital.
Analog transmitters change the amplitude (intensity),
phase, or frequency of a light wave in a smooth, continuous
fashion, while digital transmitters shift these same
attributes between distinct states, like the dots and
dashes of Morse code.
The type of modulation and encoding used in a fiberoptic
transmitter depends on a number of factors, but some
light sources are better suited for certain schemes
than others. For instance, the broad spectral output
of light-emitting diodes (LEDs) precludes them from
modulation techniques that require a stable monochromatic
wavefront such as phase and frequency modulation. For
LED fiberoptic transmitters, intensity modulation is
the best direct encoding method. This is also true for
low-coherence diode lasers. Nevertheless, frequency
or phase modulation can be achieved indirectly with
these light sources if the modulation is first performed
on an electronic subcarrier, and the subcarrier is then
used to modulate the intensity of the source.
The compact, solid-state structure of LEDs and diode
lasers, as well as their compatibility with direct intensity
modulation, has made them overwhelmingly successful
for fiberoptic communications (particularly diode lasers).
And the invention of stable, tunable, single- frequency
diode lasers such as distributed-feedback (DFB) lasers
and distributed-Bragg- reflector (DBR) cavities has
stimulated the growth of coherent fiberoptic communications
systems for external modulation of phase or frequency.
Coherent communications systems actually require at
least two single-frequency lasers, one at the transmitter
and one at the receiver. With this arrangement, modulated
light from the transmitting laser can be heterodyned
(or homodyned) with the brighter light from the receiver's
laser, which is called a local oscillator. The result
is a hundredfold improvement in receiver sensitivity
over simpler systems that detect the light signal directly
(direct detection).
Modulation schemes for coherent communications systems
include amplitude shift keying, frequency shift keying,
phase shift keying (PSK), and differential phase shift
keying. Homodyne PSK offers the highest sensitivity
of any coherent detection system.
Another big advantage of coherent optical communications
systems is that they allow a narrower channel spacing
for wavelength-division multiplexing (WDM). Multiplexing
refers to any of several techniques used to pack more
information on a single fiber by simultaneously transmitting
several signals over the same fiber. To avoid senseless
gibberish at the receiving end, each signal is uniquely
tagged in a way that the receiver can recognize. WDM
accom plishes this by delivering each signal on a slightly
different laser frequency that is then optically filtered
by the receiver.
Besides WDM, two other important kinds of multiplexing
are time-division multiplexing (TDM) and frequency-division
multiplexing (FDM). TDM segregates samples of each signal
into sepa rate time slots that the receiver can clock
off individually. With FDM, each signal is carried on
a separate subcarrier frequency that can be electronically
filtered out by the receiver.
Semiconductor Sources and Detectors
In addition to the demands of modulation and encoding,
light signals often must travel many miles of glass
before they reach the receiver. If the signal is to
maintain detectable strength and fidelity over those
distances, the glass must have low loss (scatter, absorption,
and so forth) and low dispersion at the wavelength of
the light source. This turns out to be an impossible
task for most fiberoptic installations because step-index
silica fiber has zero dispersion at a wavelength of
1.3um and minimal loss (0.16 dB/km) at 1.55um.
However, fiber loss and dispersion are of little concern
in applications such as intraoffice communications and
local-area networks. Here, distances are short and data
rates are usually low, allowing low-cost communications
systems equipped with aluminum gallium arsenide LEDs,
multimode fiber, and silicon photodetectors to be used.
The LEDs of these so-called first generation systems
are either surface or edge emitting and emit light in
the 0.87um region where fiber dispersion and loss are
both high. This wavelength also happens to be well suited
for silicon PIN and avalanche photodiodes.
For higher data rates or longer distances, sources
and detectors must operate near the 1.3 or 1.55um regions
which are defined as second and third generation fiberoptic
systems, respectively. At the 1.3um, zero-dispersion
wavelength, LEDs fabricated from indium gallium arsenic
phosphide can be used for short-haul communications
up to about 5 km. These second generation sources can
accommodate data rates of several hundred megabits per
second, but longer distances or faster data rates demand
the higher coherence and modulation speeds of indium
gallium arsenic phosphide lasers which can be tailored
to emit light at 1.3 or 1.55um. Of course, 1.55um indium
gallium arsenic phosphide sources are designed for long-haul
applications, where signal attenuation becomes important.
Silicon photodetectors cannot be used for second-
or third-generation fiberoptic communications systems,
because the 1.1-eV bandgap makes silicon relatively
transparent at these wavelengths (cutoff wavelength:
1.13um). At 1.3 and 1.55um, PIN and avalanche photodiode
(APD) photode- tectors must be fabricated from narrower
bandgap semiconductor materials such as InGaAs and germanium
(Ge). These detectors have good responsivities and fast
response times, which make for sensitive receivers with
band-widths up to 60 GHz. Schottky barrier photodiodes
provide even faster response times, leading to bandwidths
in the 100-GHz range.
When integrated into a complete fiberoptic communications
system, semiconductor light sources and detectors can
achieve bandwidths of 20 GHz or more for a wide range
of applications.
Fiber Lasers and Amplifiers
The critical challenge of long-haul fiberoptic communications
over distances of more than 100 km is how to amplify
the optical signal. Until recently, the solution has
always been to place repeaters at regular intervals
along the fiber link. For example, a 7500-km transatlantic
link uses about 100 repeaters, one every 75 km.
Repeaters are electro-optic devices that work a little
like light wave communications systems in reverse. A
photodiode (usually a low-noise PIN or APD detector)
first converts the attenuated light wave into an electronic
signal that is amplified and regenerated. A diode laser
then transforms the electronic signal back into light
for further transmission down the fiber.
The problem with repeaters is that the complex circuitry
fixes the electronic speed of each device making upgrades
difficult and costly. Repeaters also complicate WDM
systems because the channels must be handled in parallel.
In 1985-1986, however, researchers at the University
of Southampton (Southampton, England) developed an all-optical
way to amplify the 1.55um light of long-haul fiberoptic
systems. By doping a 3-m-long silica fiber core with
erbium and optically pumping it at 650 nm, they generated
125 dB of gain for a 1.53um signal. Their achievement
touched off a worldwide investigation into fiber amplifiers
for optical communications that has culminated in a
variety of important developments, not the least of
which is the evolution of all-optical transoceanic fiber
links.
Commercial erbium-doped fiber amplifiers typically
are end-pumped by a semiconductor laser at either 980
or 1480 nm. Pump radiation is introduced to the core
by a dichroic coupler (beam- splitter), leaving the
signal and pump waves to travel through the core together.
Because the narrow core keeps the pump radiation concentrated
in a small volume, just a few milliwatts of pump power
will generate gain at the signal wavelength. To avoid
unwanted resonance absorption of the signal wave by
unexcited erbium atoms, the entire length of the doped
fiber through which the signal travels must be pumped.
The advent of these optical amplifiers has already
helped define a whole new fifth generation of fiberoptic
communications devices and systems to succeed the fourth-generation
of coherent optical communications systems. For example,
the search for an effective optical amplifier at 1.3um
has produced praseodymium-and neodymium-doped fiber
amplifiers. One of the most promising candidates at
this important wavelength seems to be praseodymium-
doped ZBLAN (a fiuorozirconate compound).
Another system under development is the distributed
fiber amplifier which consists of a lightly doped fiber
core that, when pumped, generates just enough gain to
overcome the natural attenuation of the glass. Long
lengths of these "no-loss" fibers can be pumped via
an oversized, transparent cladding (cladding pumping).
Still another fifth generation device is the fiber
laser, which is basically a fiber amplifier placed between
two parallel mirrors. Conventional mirrors can be used,
but a more elegant and practical approach is to form
the mirrors from Bragg gratings written directly into
the photosensitive core. A great deal of research on
fiber lasers has yielded a plethora of new devices including
upconversion lasers with outputs at the blue end of
the spectrum.
But the most exciting field of inquiry into fifth-generation
devices involves solitons and other nonlinear fiberoptic
effects. Optical solitons consist of femtosecond light
pulses with high peak powers that can travel through
thousands of miles of fiber without changing their temporal
shape. This is because the nonlinear optical effects
of these intense pulses essentially render the fiber
"dispersionless."
Before long, optical transmitters based on these nascent
technologies will be spewing out data at trillions of
bits per second. And that's enough to satisfy even the
most voracious commuter on the information highway (at
least for a picosecond or two).
|
