1. Introduction
Communication technology has experienced a
continual development to higher and higher carrier
frequencies, starting from a few hundred kilohertz at
Marconi's time to several hundred terahertz since we
employ lasers in fiber systems. The main driving force
was that the usable bandwidth - and hence
transmission capacity - increases proportional to the
carrier frequency. Another asset comes into play in
free-space point-to-point links. The minimum
divergence obtainable with a freely propagating beam
of electromagnetic waves scales proportional to the
wavelength. The jump from microwaves to light
waves therefore means a reduction in beamwidth by
orders of magnitude, even if we use transmit antennas
of much smaller diameter. The reduced beamwidth
does not only imply increased intensity at the receiver
site but also reduced cross talk between closely
operating links and less chance for eavesdropping.
Space communication, as employed in satellite-to-
satellite links, is traditionally performed using
microwaves. For more than twenty five years,
however, laser systems are being investigated as
alternatives. 1-3) One hopes that mass, power
consumption, and size of an optical transceiver
module will be smaller than that of a microwave
transceiver. Also, fuel consumption for satellite
attitude control when quickly re-directing antennas
should be less for optical antennas. On the other hand,
a new set of problems had to be addressed in
connection with the extreme requirements for
pointing, acquiring, and tracking the narrow-width
laser beams.
In this tutorial we will first discuss the basics of an
optical free space link (Sect. 2) and then point out the
differences to terrestrial fiber systems and to
microwave links in Sect. 3. Section 4 presents the
requirements for and the available technologies to
implement transmitters, receivers, optical antennas, as
well as the PAT system (PAT...pointing, acquisition,
and tracking). Next we sketch application scenarios,
and we conclude with both a glimpse onto past and
future system technologies.
2. System Layout
A scenario typical for the transmission system in
question asks for point-to-point data transfer between
two spacecraft. The distances to be
bridged may extend anywhere from a few hundred
kilometers to 70 000 km (e.g. in near-earth
applications) up to millions of kilometers in case of
signals transmitted by a space probe.4) Today the data
rates in mind range from several hundred kbit/s to
some 10 Gbit/s.
Terminals for optical communication in space are
mostly designed for bi-directional links, at least
concerning the optical tracking function. They
comprise both a transmitter and a receiver that
generally share the optical antenna. Another
peculiarity is the necessity of beam steering (or
pointing) capability with sub-microradian angular
resolution and possibly with an angular coverage
exceeding a hemisphere.
These requirements lead to a transceiver block
diagramlevels of the duplexer are an essential prerequisite.
Duplexers can be based on spectral discrimination (i.e.
filtering), on polarization diversity, or on both. Hence
a common suggestion is to use left hand and right
hand circularly polarized light for the two directions,
respectively. This also makes the transmission
insensitive against rotation of the terminals along their
antenna axes. Because polarization duplexing will
provide only some 15 dB of isolation, wavelength
duplexing must be designed into the system in any
case.
For the general case that both terminals experience
a relative velocity along the line-of-sight, vD, the
Doppler effect will yield a frequency shift ∆f in the
received signal. As long as vD << c (c ... velocity of
light), one has ∆f = vD/λ where λ is the carrier
wavelength. In a LEO-GEO* link, vD may amount up
to some ± 8⋅103 m/s. Because of the small wavelength,
the resulting Doppler shift is large and amounts up to
± 7.5 GHz at λ = 1.06 µm for the example cited. Such
a large frequency shift might be negligible in a direct
detection receiver (as long as no extremely narrow
optical filtering is applied). In a heterodyne receiver,**
however, the frequency shift has to be compensated by
either tuning the local laser oscillator, by tuning the
electrical oscillator in a second intermediate frequency
stage, or by both.
In space applications - even more than in undersea
fiber systems - reliability and lifetime is of special
importance. As examples, the laser source itself or a
(cooled) detector may represent a weak point
concerning reliability and thus require redundancy.
Other subunits, like the telescope or the coarse
pointing assembly may be too bulky and present such
a high fraction of the mass budget that a single failure
point is accepted in their case.
3.2 Differences to microwave systems
At a first glance, the equation governing the
amount of power received in an optical directional
link, PR, is the same as one knows from microwave
links, namely
Here PT is the optical output power generated at the
transmitter, GT and GR are the gain values of the
transmit and receive antenna, λ is the carrier
wavelength, R the distance between the terminals and
the factors LT and LR cover the loss within the transmit
and receive terminal. However, the last factor, LP,
which accounts for loss caused by non-ideal pointing,
may correspond to several dB in a free-space laser
link: Because of the extremely small beamwidths
involved in optical links, transmit and receive antenna
will, in general, not yield their maximum gain. Despite
the implementation of an active tracking control loop
*
LEO..low earth orbiting (satellite), GEO..geostationary (satellite),
see also Sect. 5
**
see Sect. 4.2
to align the antenna axes, some mispointing will
persist and the receive intensity will vary statistically.
To a first approximation, the antenna gains GT, GR
are related to the diameters of the (circular) transmit
and receive antenna, DT, DR as
2
Substituting (2) into (1) reveals the 1/λ2-dependence
of receive power PR which makes the optical regime
so attractive compared to microwaves. Equation (2) is
applicable in case of diffraction limited antenna
operation. The full beam divergence then obtained is
on the order of
The very small beamwidths θ at optical frequencies
(some 5 µrad for typical values of λ and DT) are, of
course, the reason for the high antenna gain achievable
(some 115 dB). However, this advantage is not gained
for free: Establishing and maintaining contact with
extremely narrow beams is a tough task, especially if
transmitter and receiver change their relative position
(see Sect. 4.4).
One critical aspect of intersatellite laser
communications with narrow beams results from the
need to introduce a point ahead angle. Because of the
finite velocity of light (c) and the relative angular
velocity of two communication terminals moving in
space, the transmit beam must be directed towards the
receiver's position it will have at some later time. This
point ahead angle is given by 5)
where vR is the relative velocity component of
transmitter and receiver, orthogonal to the line-of-
sight, as illustrated in Fig. 3. Point ahead is generally
required in both dimensions. It amounts up to 40 µrad
for a GEO-GEO link and up to 70 µrad for a LEO-
GEO link and may thus be appreciably larger than the
beamwidth. The point ahead angle can be introduced
in either the receive or the transmit path of each
transceiver and must be adjustable if vR varies with
time. It is difficult to design a control loop for
automatic adjustment of point ahead. Therefore today's
concepts rely on the calculation of point ahead angles
from known ephemeris data and on open loop
implementation.
4. Requirements and technology
4.1 Data transmitter
The main parameters characterizing the optical
source are wavelength, output power, transverse mode,
polarization, linewidth, and modulation capability. A
smaller wavelength requires increased surface quality
of optical elements which in turn asks for bulkier
devices if diffraction limited operation is essential.
Thus the mass of the antenna (and hence the load for
the coarse pointing assembly) is strongly influenced
by the choice of λ. Also, the wavelength dependence
of the sensitivity of available optical receivers must be
considered. The output power will have to be in the
range of 100 mW and 1 W, depending on the link
distance and data rate. It should be available in a
single transverse mode to achieve maximum on-axis
antenna gain, and in a single longitudinal mode to
obtain optimum spectral efficiency. For coherent
reception, phase noise is detrimental and thus a narrow
linewidth of both the transmitter laser and the local
laser oscillator in the receiver is required. The usually
linear state of polarization emitted by the laser source
is to be converted into circular polarization before the
beam leaves the terminal (see Sect. 3.1). Modulation
may be achieved directly (e.g. in case of diode lasers
and moderate data rates) or with an external
modulator. Especially in connection with a subsequent
optical booster amplifier, the insertion loss introduced
by an electro-optic or acousto-optic modulator may be
tolerable. As with fiber systems, binary modulation
formats are envisaged for space links. In connection
with a coherent receiver, phase shift keying (and
possibly frequency shift keying) is an attractive
alternative to on-off keying, as it makes better use of
the carrier power.
4.2 Data receiver
For space applications, good receiver sensitivity is
an extremely valuable asset, not at least because no in-
line amplification is possible. It is often characterized
by the minimum number of input photons per bit to
achieve a bit error probability of 10-6. If other sources
of noise than that due to the quantum nature of
radiation are negligible, a direct detection receiver
needs n = 6.6 photons/bit. As an example for a
coherent receiver, a homodyne receiver with PSK
modulation would require n = 5.6 photons/bit.*** To
what extent this quantum limit is reached in practice
depends on the engineer’s ability to make negligible
the effect of other noise contributions, as there is
– excess noise in avalanche photodiodes (APDs),
– optical preamplifier noise (amplified spontaneous
emission),
– transistor noise and circuit noise in the receiver
electronics,
– laser phase noise,
– transmit-receive cross coupling,
– background radiation.
Today direct receivers employing APDs can be
used up to 2.5 Gbit/s. Their sensitivity is determined
by electronic and by multiplication noise and may be
less than 100 photons/bit at low data rates.6) With
optical preamplification by an Erbium-doped fiber
amplifier, direct receivers have shown sensitivities of
40 photons/bit at 10 Gbit/s.7)
With coherent reception, the received optical field
is transposed into the electrical regime (intermediate
frequency, IF) by mixing it with the field of a local
laser oscillator.8) A photodetector serves as mixer
element. Information is preserved not only about
amplitude but also about frequency and phase of the
received field, hence frequency and phase modulated
optical signals can be detected, too.
As optical mixers
have sensitive areas with dimensions large compared
to the wavelength, in the optical regime the spatial
modes of received and local field have to be matched
to obtain maximum IF signal. Matching requires
identical polarization and asks for equal amplitude and
phase distribution, the latter two optimized with
respect to the mixer element.
Coherent receivers perfectly reject radiation from
other than the nominal input direction. Equally well
they discriminate against unwanted spectral
components by their IF filter. Therefore they are a
priori less sensitive against background radiation and
cross talk.
An experimental heterodyne receiver with
phase shift keying at 565 Mbit/s has demonstrated a
sensitivity of 22 photons/bit.9)
4.3 Antennas
The transmit antenna is essentially a telescope
which magnifies the diameter of the beam emitted by
the laser (or by a booster amplifier). This beam is
generally well modeled by a Gaussian intensity
distribution. The antenna will not only introduce
truncation via its finite diameter DT but may also cause
some central obscuration, depending on the telescope's
construction. These two effects reduce the ideal on-
axis antenna gain given by equ. (2) by typically 1.5
5
dB.10) The antenna pattern resembles that of an Airy
pattern. Alignment tolerances of the optical elements
constituting the telescope are usually very tight, as the
output beam has to be perfectly collimated for
maximum gain.
The main specifications of the optical antenna are:
diameter of primary mirror (or lens), magnification,
aberrations, wavelength dependence of throughput,
sensitivity to temperature changes and gradients, and
stray light level. Usually, refractive telescopes are
envisaged in case of small diameters while reflective
systems are preferred for diameters exceeding several
centimeters. With increasing antenna aperture it
becomes more and more difficult (and expensive) to
meet specifications. Large antennas will also increase
the mass and size of an optical transceiver
considerably, as the telescope and the coarse pointing
assembly do contribute appreciably to these
characteristics. Presently it is felt that the diameter of
diffraction limited antennas should not exceed some
25 cm for free-space laser links. Coarse pointing may
be accomplished via gimbal mounting the antenna or
by a separate unit consisting of two orthogonally
mounted steering mirrors or one gimbaled reflector.
4.4 Pointing, acquisition, and tracking.
To establish an optical link in space, a
sophisticated spatial pointing and acquisition
procedure must be initiated. Information on the
position of the two space terminals has to be
available. Still, because of position uncertainty and
incomplete knowledge of the spacecraft's orientation
(attitude uncertainty), one terminal's beam width has
to be widened deliberately as to illuminate the second
terminal despite the uncertainty in position. A spatial
search operation by the (narrow beam) receive path of
the second, and subsequently, of the first terminal
have to follow before acquisition is completed and
switching to the tracking mode can occur. Wide-field-
of-view acquisition detectors in the form CCDs are
most helpful.
During data transmission, the angle between the
line-of-sight and the transmit beam axis must be kept
to within a fraction of the transmit beamwidth θ which
may be as small as a few µrad. To maintain sufficient
alignment of the transmit and receive antennas despite
platform vibrations, both terminals have to be
equipped with a tracking servo loop. Optical beacons
have to be provided in both directions to render input
information for the control loops. The data carrying
beams themselves may serve as beacon, or separate
optical beams may be implemented, e.g. in a one-way
link. Tracking should ensure a mispointing of typically
less than 1 µrad. Whenever the tracking loop signals
optimum receive position, the transmitted beam (or
beacon) will be correctly directed to the opposite
terminal. This would require a perfect coaxial
alignment for the optical transmit and receive path
within each transceiver. However, some bias, or point
ahead angle, has in general to be introduced into the
alignment, as was discussed in Sect. 3.2. To ensure
short acquisition time and adequate tracking accuracy,
sufficient optical power for the acquisition and the
tracking process must be received.
5. Application scenarios
One of the first scenarios considered was a bi-
directional, symmetric link between two geostationary
satellites (GEOs). The orbital distance between the
GEO satellites may lie anywhere between a few
degrees and some 120°, corresponding to distances
between a few thousand kilometers and 75 000 km
. Such a link has the attractive features of
a single (or very seldom) acquisition process, of a
nominally zero Doppler shift, and of low angular
tracking velocities. Connections to ground stations
could be performed with microwaves.
Large data streams generated on a low-earth-
orbiting satellite (a LEO, with a distance to ground of
less than 1000 km) may advantageously be transmitted
to a GEO acting as a relay before being directed to the
earth via microwaves (see Fig. 4b). Distances for this
asymmetric link may be as large as 45 000 km. The
concept allows continual data transfer to a single earth
station for at least half a LEO orbit.
Another use of a laser data link was already
included in the upper part. Characterized by
very large distances (e.g. millions of kilometers) and
by relatively low data rates (e.g. some 100 kbit/s),
such a link would serve to transfer data from
interplanetary and deep space probes to relay satellites
orbiting the earth. This relay could be equipped with a
large receive telescope. Further transport to ground
stations would use microwaves. As an alternative, an
optical ground station would receive the probe's data
after passage through the atmosphere.
For satellite networks now being planned or
established to serve mobile data transfer,
interconnectivity at very high data rates could be
achieved by optical links. Frequency
allocation problems - as they persist increasingly for
radio links - are practically non-existent, with the
merit of negligible mutual interference. Another
advantage is the expected smaller mass and volume of
optical terminals.
6. System technologies
The almost three decades of efforts towards
intersatellite laser links have seen various
technologies,1-3) starting from those based on lamp-
pumped, mode-locked Nd:YAG lasers,11) on CO2
lasers 5) operating at λ = 10 µm, on GaAlAs diodes
(0.85 µm), up to those employing diode-pumped
Nd:YAG lasers (λ = 1.06 µm) 12) and InGaAsP
semiconductors operating at λ = 1.5 µm.
Only a few experimental systems have been
launched so far. The European space agency, ESA, has
put a terminal on SPOT IV, a LEO earth observation
satellite.13) It employs a diode laser at λ = 0.85 µm and
shall transmit data at 50 Mbit/s. The counter terminal
still awaits its launch on board of the GEO satellite
ARTEMIS. The development of this system, dubbed
SILEX (semiconductor laser intersatellite link
experiment), started as early as 1985. Japan will
participate in this experiment by launching, in 2001, a
dedicated satellite named OICETS. This LEO satellite
is equipped with an optical terminal to communicate
with ARTEMIS. - Between 1994 and 1996 a laser link
was tested between a terminal placed on the Japanese
test satellite ETS-VI and ground stations in Tokyo and
in California, although the satellite did not reach the
intended GEO orbit but a highly elliptical one.14) The
down link operated with a diode laser, the up link with
an Argon laser.
For future applications, systems based on
Nd:YAG lasers 12) and on diode lasers at 1.5 µm in
connection with Erbium-doped fiber amplifiers 7) are
investigated presently. With the specific properties
inherent to these laser sources, they lend themselves
especially to coherent detection and to optically pre-
amplified direct detection, respectively.
In the future one should take into consideration not
only recent technological developments like optical
demodulation of phase modulated signals, the use of
low-duty-cycle return-to-zero coding, or a
combination of both. One should also give serious
thoughts to use the large, mature, and reliable
technology base commercially available in the 1.5 µm
band. Only then one can hope to achieve economy in
medium-scale applications like intersatellite networks.
