Toward the uplink correction:

application of adaptive optics

techniques on free-space optical
communications through the
atmosphere

Noelia Martínez
Luis Fernando Rodríguez-Ramos
Zoran Sodnik

Noelia Martínez, Luis Fernando Rodríguez-Ramos, Zoran Sodnik, “Toward the uplink correction: application
of adaptive optics techniques on free-space optical communications through the atmosphere,” Opt.
Eng. 57(7), 076106 (2018), doi: 10.1117/1.OE.57.7.076106.
Optical Engineering 57(7), 076106 (July 2018)

Toward the uplink correction: application of adaptive

optics techniques on free-space optical communications
through the atmosphere
Noelia Martínez,a,* Luis Fernando Rodríguez-Ramos,a and Zoran Sodnikb
a
Instituto de Astrofísica de Canarias, Vía Láctea s/n, San Cristobal de La Laguna, Spain
b
European Space Agency, European Space Research and Technology Centre, Noordwijk, The Netherlands

Abstract. The optical ground station (OGS), installed in the Teide Observatory since 1995, was built as part of
ESA efforts in the research field of satellite optical communications to test laser telecommunication terminals on
board of satellites in Low Earth Orbit and Geostationary Orbit. As far as one side of the link is settled on the Earth,
the laser beam (either on the uplink or on the downlink) has to bear with the atmospheric turbulence. Within
the framework of designing an adaptive optics system to improve the performance of the free-space optical
communications at the OGS, turbulence conditions regarding uplink and downlink have been simulated within
the object-oriented matlab adaptive optics toolbox as well as the possible utilization of a laser guide star to
measure the wavefront in this context to cope with the angular difference in uplink and downlink beams due
to the point-ahead angle required for the uplink. Simulations have been carried out by reducing available
atmospheric profiles regarding both night-time and day-time measurements and by having into account
possible seasonal changes. An AO proposal to reduce atmospheric aberrations and, therefore, ameliorate FSO
links performance is presented and analyzed in this paper. © 2018 Society of Photo-Optical Instrumentation Engineers (SPIE)
[DOI: 10.1117/1.OE.57.7.076106]

Keywords: FSO communications; uplink correction; adaptive optics; laser guide star.
Paper 180213P received Feb. 8, 2018; accepted for publication Jul. 11, 2018; published online Jul. 28, 2018.

1 Introduction background radiation on their performance.4 In case of bidi-

One of the hottest topics in the field of space communications
is the future replacement of the radio-frequency links by the
optical technology, which offers many advantages in terms of
mass, power, system flexibility, and cost to be considered as
a good candidate for the space communication race. Koishi
et al.1 summarized the advantages of optical communications
in their article about high data rate optical links, demonstrating
the superiority of this technology in comparison with RF
(Table 1). One of the most remarkable features in the optical
links is their capability to carry substantially more information
than RF: the typically used Ka-band in radio-frequency (32 to
38 GHz) allows bandwidths of 500 MHz, whereas for optical
systems at 1550-m wavelength, the bandwidth may be 1000
times larger.2
A free-space optical (FSO) communication system con-
sists of a line-of-sight technology that transmits a modulated
laser beam through the medium for broadband communica-
tions,3 performed from satellite to satellite, from ground to
ground, or from satellite to ground, and vice versa. The field
of interest at this point of the research is this last scenario:
satellite-ground and ground-satellite FSO communications,
known from now on as downlink and uplink, respectively.
FSO systems used to work near the 850-nm spectral
range, although as a consequence of the continuous pursuit
of bandwidth increase, 1550 nm has been selected as a
common wavelength for ultrahigh speed optical links: opti-
cal devices using the 850-nm spectral range cannot operate
above 2.5 Gbps, due to the power limitations imposed for
eye safety, and due to the larger influence of the solar
rectional links, it is common to have different wavelengths
for downlink and uplink.5
Optical communications systems use high coherent
modulation schemes because they provide higher sensitivity,
longer relay distance, larger communications capacity, and
better receiver selectivity, in comparison to systems using
intensity modulation direct detection. However, coherent
modulation is vulnerable to the atmospheric turbulence
since wavefront and amplitude of laser carrier signal are
distorted by this turbulence with time in the free space.6
While propagating through the atmosphere, the laser beam
experiences several effects along its transmission path due
to random refractive index fluctuations in the atmosphere
itself (consequence of inhomogeneities in the temperature
and pressure), causing the deterioration of the received signal
quality which undergoes fluctuations in both intensity and
phase, impairing the link performance. Atmospheric propa-
gation effects cause the FSO channel to be highly variable
with a rate of change as high as 1 kHz and with power
fades >10 dB.7
Adaptive optics systems measure and correct the aberra-
tions caused by the atmosphere and are extensively studied
and used in astronomical ground observations. Recently,
plenoptic cameras have extended their functionality also to
this field by playing the role of wavefront sensors (WFS).
The use of plenoptic optics for wavefront measurement
was first described by Clare and Lane8 for point sources.
Then, Rodríguez-Ramos et al.9 extended its capabilities up
to the computation of atmospheric turbulence tomography

*Address all correspondence to: Noelia Martínez, E-mail: noelia@iac.es 0091-3286/2018/$25.00 © 2018 SPIE

Optical Engineering 076106-1 July 2018 • Vol. 57(7)

 Martínez, Rodríguez-Ramos, and Sodnik: Toward the uplink correction: application of adaptive optics techniques on free-space optical. . .
Table 1 Comparison of optical links with radio-frequency
communications.1
Optical communication RF communication
Satellite Small- and medium-scale Large-scale satellite
satellite
Antenna 10 to 30 cm class 2 to 5 m class
Data rate 1 to 40 Gbps 200 Mbps to 800 bps
Resolution Very high resolution High resolution
for adaptive optics. Plenoptic sensors present the advantage
of measuring optical aberrations from extended objects as
opposed to conventional Shack–Hartmann whose perfor-
mance dramatically impairs when no point-source is avail-
able. Details about the selection of the plenoptic camera
as a WFS for the AO system is described in Sec. 3.
Wavefront sensors need a reference close to the area of
interest (a cosmic object for ground observations at the
expected place where the satellite is going to be when the opti-
cal communication takes place) in order to measure turbulence
effects along these lines-of-sight and to use this measured sig-
nal for the control of a deformable mirror (DM), which can
compensate, to a certain level, the found aberrations. Usually,
it is unlikely to find a natural star bright enough to play this
role; the problem is solved then by using artificial references,
so-called laser guide stars (LGS). LGSs are commonly used in
astronomy to avoid the lack of natural stars up on the area of
interest; they are produced by laser beams projected to the
atmosphere to create focused spots of light as if they were
a natural point-source (LGSs only appear as a point-source
when launching and measuring it on axis, in any other
case, the LGS would appear as an elongated source). There
are two types of LGS: Rayleigh guide stars, which consist
of pulsed lasers that make profit of the Rayleigh backscatter-
ing from air molecules produced at altitudes up to 20 km; and
sodium guide stars, tuned lasers which produce an output
wavelength of 589 nm and a focused beam at altitudes of
90 km by exciting the sodium atoms existing at this layer.
The proposed adaptive optics corrector has been designed
to use either a Rayleigh guide star or a sodium guide star
(depending on its availability at the telescope) as it is stated
in Sec. 3.
The AO plenoptic system presented in this research
focuses primary on the uplink precompensation as downlink
correction would only require a conventional adaptive optics
system in the receiver telescope on the Earth’s surface. This
first study addresses preliminary concepts and assumptions
by simulating the current scenario as real as possible, taking
into account real atmospheric profiles, telescope optical
parameters, and the main features of the physical elements
which, eventually, will integrate the AO system. The main
goal of this study is a feasibility assessment of the adaptive
optics precompensation in an optical link from one ground
station to a satellite.
1.1 Optical Communications in Turbulence Media
The atmosphere can be thought of as a medium into
which energy is injected in the form of thermally induced
Optical Engineering
convection or wind shear. The upper scale at which this
energy is dissipated is known as the outer atmospheric scale,
commonly designated as L0 and representative of the largest
size eddies involved in its flow, that upon reaching a critical
point, break down to smaller scales by a cascading process
until an inner scale size l0 is reached.10
Considering a laser source placed in the deep space that
spreads to hundreds or thousands of kilometres, its beam
width exceeds by far the atmospheric characteristic spatial
scales, which are in the range of a few tens of meters, there-
fore all the atmospheric eddies will be smaller and the
main effects induced by them on the downlink will be
those related to this situation: beam broadening, scintillation,
and loss of spatial coherence. However, an upward propagat-
ing laser, originated from an optical ground station and trans-
mitted to space, will imply a beam width within the inertial
range of the atmosphere (the scale size bounded by the outer
scale L0 and the inner scale l0 is referred to as the inertial
scale, and it typically ranges from 10 to 100 m down to
0.1 to 1 cm). As a consequence, the uplink will be mostly
affected by large angular displacements: beam wander and
fluctuations in the angle-of-arrival.10
By applying AO techniques to the optical communica-
tions links, beam wavefront aberrations are suppressed to
the possible extent: Wilson et al.11 and Hemmati10 demon-
strated through simulations up to a 6-dB improvement in
SNR when using adaptive optics in daytime with high back-
ground environment for downlink communications.
This paper specifically focuses on an uplink corrector
system, demonstrating the reduction of both the scintillation
and the beam spot size (what it can be translated into an
enlargement of power density) at the receiver.
1.2 Satellite Pointing Angles
The pointing and tracking systems become the keys to suc-
cess in FSO communication links; as the WFS to measure the
atmospheric turbulence will be placed on the same axis than
the optical links used to perform the communications, the
pointing angles (also known as look angles) are the ones of
interest.
From the ground station point of view, the position of a
satellite within its orbit is defined by the look angles: azimuth
and elevation; the azimuth is the angle of the direction of
a satellite, measured in the horizon plane (a tangent plane
to the surface of the Earth at observer’s position), from
geographical north in clockwise direction (from 0 deg to
360 deg); and the elevation is the angle between a satellite
and the ground station horizon plane (from 0 deg to 90 deg).
A geostationary satellite is characterized by a circular
geosynchronous orbit in the plane of the Earth’s equator
with a radius of ∼42;164 km measured from the center of
the Earth. With GEO satellites, the look angles to point
the emitter (in case of uplink) or the receiver (in case of
downlink) of the ground station only need to be adjusted
once as the satellite will remain in a given position perma-
nently. However, in order to propagate a laser beam to the
spacecraft (uplink), the speed of light needs to be taken
into account as well as the time it gets to reach the satellite
from the Earth; a laser beam projected from a ground station
must point ahead of the object by an angle of 2v∕c, where c
is the speed of light and v is the orbital speed.12 Figure 1
shows the satellite pointing geometry.
076106-2 July 2018 • Vol. 57(7)
 Martínez, Rodríguez-Ramos, and Sodnik: Toward the uplink correction: application of adaptive optics techniques on free-space optical. . .
Fig. 1 Engagement geometry for propagating a laser beam to a satellite.
GEO satellites move with an orbital velocity ðvÞ of
∼3.07 km∕s; therefore, its point-ahead angle would be
around 20 μrad (4 arcsec), whereas satellites in Low Earth
Orbit (LEO) would vary their velocity depending on their
altitude: the International Space Station is considered the
study case for this study at an altitude of around 400 km
and orbital speed of 7.668 km/s, implying a point-ahead
angle of 10.5 arcsec.
This pointing geometry will have severe implications
over the uplink corrector design as it will be critical for
the selection of a reference source to measure the wavefront
aberrations on the propagation path.
1.3 Atmospheric Isoplanatic Angle
The isoplanatic angle defines an angular patch up on the sky
within which the degree of correlation between the complex
amplitudes of the light traveling through two corresponding
locations falls to a suitable threshold value, such as 0.5.13
Isoplanatic angles are determined mainly by atmospheric
turbulence at intermediate and higher altitudes. Two well-
separated light waves (i.e., binary stars) traveling toward
the telescope will tend to pass through separated turbulence
structures, which will cause a decorrelation between the
resulting aberration on both stars and imposing a limit on
the isoplanatic angle; however, if the same light waves
traveled through lower altitudes, the turbulence structure
would be the same as they are going through it and, there-
fore, aberrations would be basically equal.
Following the example of the binary star, the optical com-
munication link traveling from a satellite to a ground station
may suffer different turbulence conditions than the one trav-
eling from the same location on Earth to the same satellite
due to the point-ahead angle. The relationship between the
isoplanatic angle on a specific location and the point-ahead
angle difference of the downlink and the uplink has been
studied and simulated in this paper in order to know whether
one link could be corrected with the wavefront measure-
ments on the other one or, on the contrary, some additional
mechanism would be necessary to quantify and reduce the
atmospheric influence, such as LGS launched to the sky
point where the uplink will take place. It is important to
Optical Engineering
mention that, whenever an LGS is used as a reference to mea-
sure the incoming wavefront (such as the usual astronomy
scenario), no tip-tilt (TT) information could be extracted
from it; on the contrary, when using the downlink signal
from the satellite, this TT is feasibly measurable; this detail
could play an important role on the system design (Sec. 3).
2 Simulations: Procedures and Results
Communication links have been emulated within the object-
oriented matlab adaptive optics (OOMAO) toolbox by taking
into account their expected location and the point-ahead
angle between both, in such a way that the downlink has
been simulated as a source at azimuth 0 in. and zenith
angle 0 in., and the uplink as a source launched to azimuth
0 in. but zenith angle equals to the point-ahead angle in GEO
and LEO satellites (4 in. and 10 in., respectively, for simu-
lation purposes).
The OOMAO simulator is a Matlab toolbox that has
been developed by Conan and Correia14 with the aim of
simulating the behavior of a complete AO system. It is
based on class objects: atmosphere class, telescope class,
source class, WFS class, DM class, and imager class. Each
of them needs to be set up with the proper parameters in
order to be as representative of the real situation as possible.
The telescope has been modeled with the Optical Ground
Station (OGS) optical parameters, including the central
obscuration (see Table 2). The OGS is a telescope owned
by the European Space Agency (ESA) and located at the
Teide Observatory (OT) in Tenerife, in Spain, mostly used
for communication terminals testing.
The “ogsAtmosphere” function has been defined, within
the “atmosphere class” framework, with the purpose of
binding all OT atmosphere types together. The atmospheric
profiles have been extracted from OT nocturnal profiles15 as
well as from SHABAR measurements in case of diurnal
analysis16 (Figs. 2 and 3).
Every data set has been integrated into the simulator by
concentrating the overall atmosphere distribution in only
four layers; therefore, the available C2n values have been
used to calculate the global Fried parameter (corresponding
to the atmosphere type under study) as well as the fractional
076106-3 July 2018 • Vol. 57(7)
 Martínez, Rodríguez-Ramos, and Sodnik: Toward the uplink correction: application of adaptive optics techniques on free-space optical. . .

Table 2 OGS telescope optical parameters. Fried parameters related to each atmospheric layer. Global
Fried parameter, r0 , has been computed with Eq. (1) and
fractional Fried parameters, fr0 , have been calculated by
Telescope diameter 1m
the procedure described in Ref. 17 and summarized by
Central obstruction 20% Eq. (2)
Optical configuration Ritchey–Chrétien (RC) focus Z −3∕5
EQ-TARGET;temp:intralink-;e001;326;697 r0 ¼ ð0.423k2 C2n ðhÞdhÞ ; (1)
Focal length RC 13.3 m

Field-of-view 2.5 arcmin

where k is the wavenumber k ¼ 2  π∕λ,

Fig. 2 Average C 2N profiles obtained from the individual measurements at OT (Spain) by SCIDAR
instrument from 2002 November to 2009 January.15 Statistics gathered taking into account seasonal
changes.

Optical Engineering 076106-4 July 2018 • Vol. 57(7)

 Martínez, Rodríguez-Ramos, and Sodnik: Toward the uplink correction: application of adaptive optics techniques on free-space optical. . .

Fig. 3 Average C 2N profiles obtained from the individual measurements at several Sun elevations by
SHABBAR instrument during 2014 at OT, in Spain.16 Statistics gathered taking into account seasonal
changes (data gaps between January and May).

C2 ðhÞΔh propagation channel implies a movement of the atmospheric

f r0 ¼ R n 2 : (2) screen at its corresponding wind direction and velocity.
dhCn ðhÞ
EQ-TARGET;temp:intralink-;e002;63;285

The validity of the procedure and the simulation tool is

This simulation implements the Fresnel propagation of
a Gaussian-shaped beam from the emitter plane to the first
atmospheric layer and from this layer to the next one until the
upper bound (at ∼10; 000 m in this model), plus the phase
change found at each of the layers (due to the atmosphere
itself); the Fresnel propagation extends from the last atmos-
pheric layer to the receiver location at around 38,000 km for
the GEO case and 400 km in the LEO scenario. Talanov scale
transformation18 is imposed to fix a constant ratio between
the beam size and the dimensions of the discretization grid.
In all simulations, the receiver is placed at the geometrical
focus of the beam.
The time behavior is taken into account by moving
each phase screen perpendicular to the beam in accordance
with the wind velocity profile (also extracted from the OT
atmospheric profiles). Every independent realization of the
proved by comparing the LEO intensity profile within the
receiver plane to the simulated beam profiles gathered by
Andrews et al.19 For this comparison, the beam intensity pro-
file has been analyzed using the same metrics: the W 0 ∕r0
ratio, where W 0 is the radius of the Gaussian beam and
r0 is the Fried parameter. Figure 4 shows the simulated inten-
sity profiles that are really similar to the ones in Ref. 19.
Two types of simulation procedures have been developed
in order to check two main features that will have crucial
implications on the AO system design for optical communi-
cations. First, some simple simulations have been carried out
to know whether both communication links would be inside
the isoplanatic area (isoplanatic simulation) or not, with the
purpose of justifying the use of an LGS propagated to the
uplink expected location in contrast to the use of the down-
link signal itself as a reference source; afterward, a more
complex methodology has been built to get knowledge of

Optical Engineering 076106-5 July 2018 • Vol. 57(7)

 Martínez, Rodríguez-Ramos, and Sodnik: Toward the uplink correction: application of adaptive optics techniques on free-space optical. . .

Fig. 4 Uplink intensity profiles within the receiver plane at 400 km. Notice the differences depending on
the W 0 ∕r 0 ratio: when this parameter is small W 0 ∕r 0 ≪ 1, the Gaussian beam shape is perfectly
received; tilt effects are predominant when the ratio is close to one; and in case of strong turbulence
(W 0 ∕r 0 ≫ 1), the beam is distributed over the receiver plane.

Fig. 5 RMSE of the difference between phases of one downlink at telescope zenith and one uplink get-
ting away 1 in. to 15 in. from zenith. Seasonal and day length variation analysis: (a) winter night profile,
(b) winter day profile, (c) summer night profile, and (d) summer day profile.

Optical Engineering 076106-6 July 2018 • Vol. 57(7)

 Martínez, Rodríguez-Ramos, and Sodnik: Toward the uplink correction: application of adaptive optics techniques on free-space optical. . .

the uplink correction performance after measuring the wave- measurements on the downlink itself as a reference source,
front on the downlink or on the launched LGS (closed-loop which will be separated from the first one by a point-ahead
simulation). All simulations and derived results are presented angle, or measuring the wavefront on an LGS launched to
in following sections. the uplink location. LEO satellites, with a 10-in point-ahead
angle has been the case under study as it has been considered
2.1 Isoplanatic Area at the OT the worst scenario (in comparison to GEO satellites).
Two metrics to analyze the performance of the uplink
The analysis of the isoplanatic area at the OT consists of
assessing whether each of the links would suffer from similar correction have been selected: the long-term scintillation
atmospheric wavefront and therefore, one of them could be index [see Eq. (3)] seen by a 30-cm receiver (treated as a
used as a reference source to correct the other one. point receiver for the calculations) at a given height over
In order to achieve these conclusions, the wavefront of the Earth’s surface (400 km for the simulation), and the spot
both sources has been simulated with the available data size of the beam, represented by the Dσ parameter [see
(atmospheric profiles, telescope, and sources), and it has Eq. (4)].
been checked whether each case study fulfils the isoplanatic
condition of 1 rad as maximum root-mean square difference Table 3 Uplink correction performance: simulation framework.
of the two wavefronts (the one suffered by the downlink and
the one suffered by the uplink).20
Results have been analyzed by taking into account the Atmosphere OT atmospheric profiles (see Figs. 2 and 3)
main influencing variables on this study: solar time and Telescope OGS telescope optical parameters (see Table 2)
season when the atmospheric profiles were acquired. In
addition, the inferred conclusions are evaluated under the Uplink source λ ¼ 1550 nm
premise of the point-ahead angle related to each orbit case.
Notice that all simulation scenarios have been performed Zenith = 10 in.
by placing the downlink at telescope zenith, and as it is well- Downlink λ ¼ 1064 nm
known, this situation is the best one concerning the amount source
of turbulence; if the results are not satisfactory in this point- Zenith = 0 in.
ing direction, they will not be either at lower telescope
LGS λ ¼ 589 nm
elevations.
Simulation results are summarized in Fig. 5. The local Height = 90 km
time is proven to have a significant influence on the size
of the isoplanatic patch: Figs. 5(a) and 5(c) (corresponding Photon return flux ¼ 7.7 × 106 photons∕s∕m2 21
to nocturnal atmospheric measurements) show that, at the
Na layer thickness = 20 km22
GEO point-ahead angle (4 in.) distance, the border of
the isoplanatic patch is almost reached, meaning that the Na profile (see Fig. 6)22
uplink may be corrected by the wavefront measurement
on the downlink and that correction may be good enough. Plenoptic WFS 20 × 20 microlenses
However, for the LEO case (10-in point-ahead angle), Sensor resolution = 300 × 300 pixels
both sources will be outside the isoplanatic area and
correcting one with the incoming wavefront on the other Deformable 21 × 21 actuators (373 useful actuators)
one introduces anisoplanatism errors; in this last case, an mirror
LGS, launched at the same direction as the uplink, would
be the best way to measure the turbulence the uplink is
exposed to. Moreover, when dealing with diurnal links,
the turbulence is much higher and concentrated at the ground
layer; therefore, the isoplanatic patch becomes smaller.
Figures 5(b) and 5(d) show results of day-time simulations,
pointing out the need of an LGS as a reference source for
the uplink correction. Seasonal changes have also been
analyzed, although no big differences during the night
have been encountered as the OT location in the Canary
Islands (Spain) has a quite stable temperature all year
long. However, the difference becomes more noticeable in
the case of diurnal links, as during summer, the Sun heats
intensely the Earth’s surface, whereas during winter, the
ground does never get so hot.
The isoplanatic study conclusions remark the needs of
propagating an LGS to the expected uplink location in
order to perform an optimal uplink precompensation.

2.2 Uplink Correction Performance

The closed-loop simulation has been performed to evaluate Fig. 6 Na LGS height profiling obtained during 2016 at OT using the
the quality of the link corrections carried out with wavefront LGS profiling from IAC80 telescope.22

Optical Engineering 076106-7 July 2018 • Vol. 57(7)

 Martínez, Rodríguez-Ramos, and Sodnik: Toward the uplink correction: application of adaptive optics techniques on free-space optical. . .

Each analysis has been performed with 500 independent
iterations to characterize the statistical behavior of the propa-
gation channel. Both possible operation procedures (using
downlink versus LGS as a reference source) consider the
use of the downlink signal for the TT correction, due to
the impossibility of extracting TT data from LGS
hI 2 i − hIi2
σ 2I ¼
hIi2
EQ-TARGET;temp:intralink-;e003;63;686
Simulation parameters are listed in Table 3: the atmos-
phere model comes from OT profiles and telescope model
has been built with OGS telescope optical parameters, as
stated before. The LGS model has been assumed to be
the European Southern Observatory’s Wendelstein sodium
Laser Guide Star Unit (WLGSU), whose profile and photon
return were measured at OT during 2016.21,22 AO correction
is inferred by the 373 modes of the DM under consideration.
; (3) Figure 7 shows the simulation results when studying

the long-term scintillation index at a typical LEO height

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (400 km) and launching the laser from several telescope
R þ∞ R þ∞ diameters (from 0.01 to 1 m). It is pointed out how correcting
−∞ R −∞ R ½Iðx; yÞðx − x̄Þ2 dxdy
Dσ ¼ þ∞ þ∞ ; (4) the uplink by the turbulence information on its location
−∞ −∞ Iðx; yÞdxdy
EQ-TARGET;temp:intralink-;e004;63;642

(propagating an LGS) would reduce the scintillation index

by almost two orders of magnitude in the case of a winter
where I represents the light intensity. night atmosphere [Fig. 7(a)] and, even in a very turbulent

Fig. 7 Uplink scintillation index with no AO correction (red), corrected by the measurements on the down-
link position (blue), and by the measurements on the uplink location (green). Uplink and downlink are
separated by 10 in. (LEO satellite). Day length variation analysis: (a) nocturnal atmospheric profile and
(b) diurnal atmospheric profile.

Fig. 8 Uplink spot width with no AO correction (red), corrected by the measurements on the downlink
position (blue), and by the measurements on the uplink location (green). In this case, uplink and downlink
are separated by 10 in. (LEO satellite). Day length variation analysis (a) nocturnal atmospheric profile and
(b) diurnal atmospheric profile.

Optical Engineering 076106-8 July 2018 • Vol. 57(7)

 Martínez, Rodríguez-Ramos, and Sodnik: Toward the uplink correction: application of adaptive optics techniques on free-space optical. . .

scenario (diurnal atmosphere), the improvement is also

clear with a reduction of one order of magnitude average
[Fig. 7(b)]. In the winter night scenario, the two first ana-
lyzed Tx diameters (1 and 10 cm) are not big enough to
make profit of the AO correction.
The analysis of the resulting spot size at the receivers alti-
tude −400 km in this case is shown in Fig. 8. Measurements
are shown in equivalent arcseconds seen by the telescope
located at ground level. The spot size is calculated for the
short term (the measurement of the spot size at each indepen-
dent realization of the propagation channel iteration). In case
of nocturnal links [Fig. 8(a)], the improvement in the perfor-
mance is represented by a reduction of 41.12% (from 2.245
in. down to 1.324 in.) when using the LGS as a reference
source, whereas there is almost no difference when cor-
recting the uplink with the downlink information. However,
in diurnal scenarios it is when the performance improvement
becomes more noticeable, with a reduction in the spot size of
almost 50% (from 4.858 in. with no correction, down to
2.421 in. when using an LGS as reference) [see Fig. 8(b)]
when propagating the artificial reference to the expected
uplink location and no real difference when using the down-
link signal.

3 Uplink Wavefront Corrector System: The

Proposal
By the beginning of 2017, the preliminary design phase of
the instrumentation needed to reach the Adaptive Optics cor-
rection on the uplink has started, based on the simulation
outcome.
The uplink wavefront corrector system (UWCS) is a
proof-of-concept instrument, whose aim is the improvement
of Earth-satellite communications (uplink), at the OGS in
the OT; it will measure and correct the perturbations the link
suffers while traveling through the atmosphere. It is currently
in conceptual design phase.
The UWCS has been designed to demonstrate the advan-
tages of uplink correction by AO techniques, as well as to test
the feasibility of using a plenoptic camera as WFS with LGS,
which will be launched to create an artificial reference for
turbulence measurement and correction at the expected
uplink location.
The selection of the plenoptic camera as a WFS instead of
the conventional Shack–Hartmann is based on the following
premise: as it has been stated before, the use of a LGS
implies that some other mechanism needs to be found to
measure TT aberrations on the upward propagated beam;
in case of using a natural guide star (NGS) for this purpose,
both natural and artificial stars will fall inside the plenoptic
camera field-of-view (one of them will be obviously out-of-
focus, although it would not be a problem due to fact that
the plenoptic WFS is capable of measuring the turbulence
over extended objects). A Shack–Hartmann WFS would
not allow this feature and two SH-WFS would be needed
(one dedicated to the LGS and the other one to the NGS).
However, in the event of using the downlink beam itself
to measure the TT, it would imply a WFS detector change
in order to fulfil the wavelength requirements and, as the
LGS would have a different wavelength, this detector
would be less sensitive to either the downlink or the LGS
signals. Further tests will be performed to assess in detail
these two possibilities.
Fig. 9 UWCS conceptual design. DM is only applied to the commu-
nications laser in order to specifically evaluate the uplink correction.
In a practical system, the DM could also be included in the downlink
path. BS, beam splitter; WFS, wavefront sensor; DM, deformable
mirror; LGS, laser guide star.
The UWCS conceptual design is shown in Fig. 9. The
system is composed of a plenoptic camera as WFS, a
laser for the creation of the Rayleigh LGS (WFS reference),
a DM to precorrect the uplink of an FSOC laser in such a way
the on-purpose added predistortions on the link will counter-
act the atmospheric aberration while traveling through it;
another WFS analyzing the DM output in order to initialize
the system; and a control PC to operate the system at the
proper frequency.
The AO system has been designed for its integration at the
Ritchey–Chrétien focus of the OGS telescope with a 13.3-m
focal length at this optical configuration.
A Rayleigh LGS type has been selected due to the current
nonavailability of a sodium LGS at the OT. However, an
Na-LGS could be introduced in the setup by modifying
the launching optics.
A sketch of the beam path is shown in Fig. 10. The tele-
scope light input and output enters and exits the system
through the BS1 beam splitter. The LGS laser creates a
Rayleigh star up on the sky by traveling through the M2
mirror, the BS2 and BS1 beam splitters; synchronously, the
WFS measures the atmospheric turbulence by using this
Rayleigh LGS and one natural star or the downlink signal
as a reference. The WFS receives the collected light by
the whole telescope aperture entering through the polarizing
BS1 beam splitter, which only allows backscattered light

Optical Engineering 076106-9 July 2018 • Vol. 57(7)

 Martínez, Rodríguez-Ramos, and Sodnik: Toward the uplink correction: application of adaptive optics techniques on free-space optical. . .
Fig. 10 UWCS beam path sketch.
from the Rayleigh LGS into the detector, and the M2 mirror
to reduce the used space on the optical breadboard.
Once the wavefront is well known and the DM receives the
commands to actuate, it modifies the FSOC laser beam
(which travels through the M3 mirror until the DM), adding
the aberration to counteract the atmospheric phase. The FSOC
laser beam exits the telescope after the DM through the BS2
and BS1 beam splitters. In this case, the self-propagating
path of the LGS is not precompensated.
The light in the BS2 beam splitter is also conducted to the
DM WFS, which measures the initial state of the DM in
order to perform the proper calibration.
Both lasers (FSOC laser and LGS) share the telescope
aperture, therefore focus difference needs to be handle by
placing both launching optics on linear stages; the LGS
beam is collimated when arriving to the aperture whereas
the communications laser has a divergence angle with the
aim of further moving the focus plane.
Design requirements have been already defined to fulfil
the expected operational performance, although further
development phases will follow in 2018 and 2019.
Two laser types are already available at the OGS telescope
and the characteristics of both sources have been analyzed
in order to choose the best option in this very conceptual
design: a continuous wave-doubled Nd:YAG laser at 532 nm
with 18-W output power and a Hawk pulsed laser at 1064
and 532 nm with 8-W output power and pulse frequency
between 2 and 20 kHz. As the guide star is launched through
the whole telescope aperture, the LGS laser needs to be
pulsed or chopped to avoid light backscattering from lower
atmospheric layers. Hence, the Nd:YAG laser needs to be
used with an acousto-optic modulator whereas the Hawk
laser is already designed for pulse operation. In order to
act as a reference for the wavefront measurement and con-
sequent DM interaction matrix calculation, the laser power
Optical Engineering
needs to be at least 8 W in the visible wavelength (532 nm
green) to generate enough response from the Rayleigh return,
based on measurements at OT during 2017. Tests have been
carried out to select the most suitable laser for the UWCS:
due to the fact that the Nd:YAG laser could not be hanged
from the Ritchey–Chrétien focus (it has operational require-
ments of mechanical stability and clean environment), the
Hawk laser has been selected as an LGS source.
The system shall work at 2 KHz and the most crucial ele-
ments to cope with this requirement are the WFS and the
DM; therefore, their selection and design need to be handled
with special care.
On one hand, the WFS will consist of an Firstlight
OCAM2S camera plus a specifically customized 8 × 8 lens-
let array. The camera selection is based on its maximum
working frequency (2 KHz) and its capability to receive an
external trigger signal to properly synchronized the acquis-
ition with the LGS pulses. The WFS needs to be sensible in
the visible in general, paying special attention to the 532-nm
wavelength with the purpose of working with Rayleigh guide
stars. The microlenses array has been designed and custom-
built to fulfil the optical requirements of the RC focus, as the
F number of the optical set-up needs to be equal to the F
number of the ensemble “camera + lenslet array” in order
to avoid the blurring effects the microlenses introduce.23
The OCAM2S camera has 240 × 240 pixels of a 24-μm
size; therefore, by placing 8 × 8 microlenses in front of
the sensor, the subimage (created by each microlens) will
have 30 × 30 pixels of 24 μm each, meaning each microlens
in the array will take up 720 μm; by relating this length (one
microlens diameter) with the F number to achieve (F∕13.3),
the focal length of the array needs to be 9.57 mm. The built-
in microlenses array has 10-mm focal length to fulfil this
requirement.
On the other hand, for the DM selection, it has to be taken
into account that the optimum DM size is determined by the
relationship D∕r0, where D is the telescope diameter (1 m
in OGS scenario) and r0 is the Fried parameter. By consid-
ering most favorable and unfavorable cases (r0 ¼ 20 cm and
r0 ¼ 2 cm, respectively), ALPAO DM241 with 241 actua-
tors has been chosen. The UWCS needs to be able to control
an absolute minimum of modes equivalent to the number of
actuators expected to be present in the DM. The figure of 241
is considered as a goal.
The UWCS laboratory prototype is currently being built
and will be future tested at Instituto de Astrofísica de
Canarias (IAC) as a previous step to telescope verification.
4 Conclusions and Future Development
FSO links through Earth’s atmosphere are subject to phase
distortions superimposed by turbulence. Adaptive optics
techniques can help to mitigate these perturbations and that
is the aim of this research: deeply characterizing the propa-
gation channel in order to use this information to suppress
the atmospheric contribution from the optical links.
Once FSO communications through the atmosphere have
been profoundly studied from a theoretical point of view, one
adaptive optics system to improve their performance in the
OGS has been proposed and simulated with the OOMAO
toolbox. It has been demonstrated that, both downlink and
uplink are not usually inside the isoplanatic area, meaning
that the upwardly propagated laser will travel through a
076106-10 July 2018 • Vol. 57(7)
 Martínez, Rodríguez-Ramos, and Sodnik: Toward the uplink correction: application of adaptive optics techniques on free-space optical. . .
different atmospheric patch than the light coming from the
satellite. Therefore, an artificial reference at the expected
uplink location (instead of using the downlink signal itself)
seems to be the best solution to measure the incoming wave-
front. However, results have also shown that the wavefront
on the downlink is not as different as expected in comparison
with the uplink turbulence, and this fact could be used to deal
with TT corrections; this option would imply a different
detector in the WFS camera depending on the downlink
wavelength.
The UWCS has been conceived by following the premises
from simulation outcomes as a pathfinder instrument to dem-
onstrate AO uplink correction. Development phases will
follow with deeper detail design.
Future tests at the OGS in Tenerife (Spain) are planned to
be performed during the end of 2018 and 2019, in order to
analyze each of the subsystems separately as well as the per-
formance of the plenoptic WFS with the LGS and an out-of-
focus natural guide star falling within the camera field of
view, before the final design review of the UWCS.
Acknowledgments
The authors would like to thank the Instituto de Astrofísica
de Canarias (IAC) and the European Space Agency (ESA)
for cofunding this research, as well as Carlos Correia for
the support and maintenance of the OOMAO simulator.
References
1. Y. Koishi et al., “Research and development of 40 Gbps optical free
space communication from satellite/airplane,” in Int. Conf. on Space
Optical Systems and Applications (ICSOS), pp. 88–92 (2011).
2. W. D. Williams et al., “RF and optical communications: a comparison of
high data rate returns from deep space in the 2020 timeframe,” NASA/
TM—2007-214459 (2007).
3. A. A. B. Raj, J. A. V. Selvi, and S. Raghavan, “Terrestrial free space
line of sight optical communication (TFSLSOC) using adaptive
control steering system with laser beam tracking, aligning and position-
ing (ATP),” in Int. Conf. on Wireless Communication and Sensor
Computing, pp. 1–5 (2010).
4. S. Jasmine, S. Robinson, and K. Malaisamy, “Investigation on free
space optical communication for various atmospheric conditions,” in
2nd Int. Conf. on Electronics and Communication Systems (ICECS),
pp. 1030–1034 (2015).
5. H. Kaushal and G. Kaddoum, “Optical communication in space:
challenges and mitigation techniques,” IEEE Commun. Surv. Tutorials
19, 57–96 (2017).
6. J. Cao et al., “Performance analysis of a coherent free space optical
communication system based on experiment,” Opt. Express 25,
15299–15312 (2017).
7. N. M. Namazi et al., “Synchronization and detection of binary data in
free-space optical communication systems using Haar wavelet transfor-
mation,” Opt. Eng. 45, 015001 (2006).
8. R. Clare and R. Lane, “Wave-front sensing from subdivision of the focal
plane with a lenslet array,” J. Opt. Soc. Am. A 22, 117–125 (2005).
Optical Engineering 076106-11
9. J. Rodríguez-Ramos et al., “The CAFADIS camera: a new tomographic
wavefront sensor for adaptive optics,” in 1st AO4ELT Conf., p. 05011
(2010).
10. H. Hemmati, Deep Space Optical Communications, Wiley, Hoboken,
New Jersey (2006).
11. K. Wilson et al., “Daytime adaptive optics for deep space communica-
tions,” in 10th Int. Conf. of Pacific Basin Countries, p. 117 (2004).
12. R. R. Parenti, “Adaptive optics for astronomy,” Lincoln Lab. J. 5(1),
93–114 (1992).
13. T. S. McKechnie, General Theory of Light Propagation and Imaging
through the Atmosphere, Vol. 196, Springer, Cham (2015).
14. R. Conan and C. Correia, “Object-oriented Matlab adaptive optics tool-
box,” Proc. SPIE 9148, 91486C (2014).
15. B. García-Lorenzo and J. J. Fuensalida, “Statistical structure of the
atmospheric optical turbulence at Teide Observatory from recalibrated
generalized SCIDAR data,” Mon. Not. R. Astron. Soc. 410(2), 934–945
(2011).
16. J. M. de la Rosa et al., “Daytime turbulence profiling for EST and its
impact in the solar MCAO system design,” Proc. SPIE 9909, 99096X
(2016).
17. R. Conan, Object-Oriented Matlab Adaptive Optics: User Guide, https://
github.com/rconan/OOMAO/tree/master/User%20Manual (2013).
18. V. Talanov, “Focusing of light in cubic media,” Sov. J. Exp. Theor. Phys.
Lett. 11, 199 (1970).
19. L. C. Andrews et al., “Strehl ratio and scintillation theory for uplink
Gaussian-beam waves: beam wander effects,” Opt. Eng. 45, 076001
(2006).
20. A. Glindemann, Principles of Stellar Interferometry, Springer, Berlin
(2011).
21. R. Holzlöhner et al., “Comparison between observation and simulation
of sodium LGS return flux with a 20W CW laser on Tenerife,” Proc.
SPIE 9909, 99095E (2016).
22. J. Castro-Almazan et al., “Na LGS height profiles at Teide Observatory,
Canary Islands,” (2017).
23. L. Rodríguez-Ramos, “Utilización de la cámara plenóptica como sensor
de frente de onda para Optica Adaptiva en Astrofísica,” PhD Thesis,
Universidad de La Laguna (2015).
Noelia Martínez is a PhD student, currently working at Instituto de
Astrofísica de Canarias (IAC) on the topic “Adaptive Optics for
Free-Space Optical Communications.” She received her BS degree
in technical industrial engineering from the University of Santiago
de Compostela, Spain, in 2012, and her MS degree in industrial engi-
neering from the University of Leon, Spain, in 2014. Her primary areas
of interest are astronomical and space instrumentation, focusing onto
the electronics and control fields of expertise.
Luis Fernando Rodríguez-Ramos is a head of the Electronics
Department of the IAC since March 1991. He has been involved in
a number of outstanding projects in the astronomical instrumentation
field: EDIFISE, AOLI, HARMONI, GTCAO, among others, as well as
serving as consultant in technology matters for industry and ESA
(VEGA launcher for small satellites).
Zoran Sodnik joined the European Space Agency (ESA-E STEC) in
1993 as senior optical engineer, responsible for managing R&D
activities and supporting ESA programs (ARTEMIS, DARWIN, LISA,
SMART-1, SMART-3, AIM, etc.). His main activities are in the field
of optical communication technologies; he managed the development
of ESA’s optical ground station in Tenerife, Spain, for which he became
the responsible station manager in 1995. Since December 2013, he
is heading the Opto-Electronics Section (TEC-MME) at ESTEC.
July 2018 • Vol. 57(7)