Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Available online at www.sciencedirect.

com

Advances in Space Research 47 (2011) 802–810


www.elsevier.com/locate/asr

Link budget and background noise for satellite quantum


key distribution
Andrea Tomaello a,b,⇑, Cristian Bonato a,b,c, Vania Da Deppo a,b, Giampiero Naletto a,b,
Paolo Villoresi a,b
a
Dipartimento di Ingegneria dell’Informazione, Università degli Studi di Padova, Via Gradenigo, 6/B, 35131 Padova, Italy
b
Istituto di Fotonica e Nanotecnologie, CNR, IFN UOS Padova, Via Trasea, 7, 35131 Padova, Italy
c
Huygens Laboratory, Leiden University, P.O. Box 9504, 2300 RA Leiden, The Netherlands

Received 12 January 2010; received in revised form 2 November 2010; accepted 3 November 2010
Available online 18 November 2010

Abstract

Optical quantum communication exploiting satellites is the most promising field to enable global quantum communication.
Bonato et al. (2009) discussed the feasibility of quantum key distribution (QKD), for vertical space to Earth and Earth to space links
at different time conditions and transmitter and receiver specifications, outlining a set of needed technical requirements.
In a real scenario a satellite is seen at any time at a different position from an Earth station: we analyze the link attenuation and the
signal to noise ratio (SNR) for its visibility time. In particular, we study the atmosphere effects on the beam propagation for different
altitudes and zenith angles, modelling the dynamic quantum link for an orbiting satellite.
Our results show the feasibility of QKD from low-Earth orbit satellites to Earth during night time; QKD uplinks are much more
difficult to achieve, and their implementation is feasible reducing the field of view of the receiving telescope. Feasibility of QKD with
satellite at higher altitude is a technological challenge due to the narrow portion of sky that an Earth station can effectively use for
communication.
Ó 2010 COSPAR. Published by Elsevier Ltd. All rights reserved.

Keywords: Satellite quantum communication; Optical communications; Quantum key distribution; Satellite quantum link model

1. Introduction atmosphere. Now we are interested to study and simulate


a real satellite quantum link, in which a satellite is seen
In this paper we present the model of an optical quan- from an Earth station, at a different positions in the time
tum communication link between a satellite and Earth expressed by a set of zenith angle, azimuth angle, and link
and vice versa, considering the portion of satellite orbit distance. In this scenario, also the link condition evolves in
seen from and Earth station. time, establishing an effective link budget and a portion of
Bonato et al. (2009) found the best environmental con- sky that allows the quantum communication.
dition and the hardware requirements for satellite QKD, To model this time-evolving link, we introduce the depen-
modelling the propagation of a laser beam in a turbulent dence of the atmosphere parameters to the zenith angle.
Our analysis is mainly focused on a few classes of
⇑ Corresponding author at: Dipartimento di Ingegneria dell’Informaz- satellites:
ione, Università degli Studi di Padova, Via Gradenigo, 6/B, 35131 Padova,
Italy.  Europe global positioning system (Galileo).
E-mail addresses: tomaello@dei.unipd.it (A. Tomaello), bonato@  USA global positioning system (GPS).
molphys.leidenuniv.nl (C. Bonato), dadeppo@dei.unipd.it (V. Da Deppo),
naletto@dei.unipd.it (G. Naletto), paolo.villoresi@dei.unipd.it (P. Villor-
 Low Earth Orbit (LEO).
esi).  Geostationary Earth Orbit (GEO).

0273-1177/$36.00 Ó 2010 COSPAR. Published by Elsevier Ltd. All rights reserved.


doi:10.1016/j.asr.2010.11.009
A. Tomaello et al. / Advances in Space Research 47 (2011) 802–810 803

The model as is can be extended to other satellites. here wST is the short-term beam width, while b is the
Each class of satellites has been examined for the follow- instantaneous beam displacement from the unperturbed
ing aspects: Galileo satellites because are a new interesting position.
constellation which can be exploited for new applications, For a collimated beam, the long-term beam width is:
GPS because are the counterparts of Galileo, LEO satel-    2
L2 4L
lites because are at the lower altitude and can allow high w2LT ¼ w20 1 þ 2 þ 2 ; ð3Þ
Z0 kr0
SNR; GEO because they are at a fixed position in the
sky, and so they ideally have no time limits for where Z0 is the Rayleigh parameter of the beam, L is the
communications. propagation distance and r0 is the Fried parameter (for
the uplink), given by (Tyson, 1997):
2. Signal and noise " Z L  5=3 #3=5
2 2 Lz
r0 ¼ 0:423 secð#zn Þk C n ðzÞ dz ; ð4Þ
As for any communication system, to analyze and h0 L
model a system, we first need to know the attenuation of
the link and the noise introduced in the system. Moreover, where #zn is the zenith angle and h0 is the altitude of the
in quantum communication information is coded at the Earth station.
level of single photons (Sergienko, 2006), so the signal can- Considering the geometry formed by an orbiting satellite
not be increased: therefore a sufficient signal to noise ratio and an Earth station, for a circular or a near circular orbit
(SNR) can be achieved only reducing the link attenuation we can estimate the effective link distance as:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
and the background noise.
2RT cos #zn þ ð2RE cos #zn Þ2 þ 4ððRSAT þ RE Þ2  R2E Þ
L¼ ;
2.1. Signal attenuation 2
ð5Þ
As discussed by Bonato et al. (2009), the main factor where RE and RSAT are respectively the mean Earth radius
limiting the performance of free-space optical communica- and the satellite’s altitude
tion is atmospheric turbulence. The refractive index inho- The refractive index structure constant C 2n is (from
mogeneities induced by turbulence increase the beam Andrews et al. (1995)):
spreading to an extent significantly larger than what caused
by diffraction. Depending on their size, the turbulence C 2n ðhÞ ¼ 0:00594ðv=27Þ2 ðh  105 Þ10 eh=1000
eddies can cause two main effects (Dios et al., 2004): beam þ 2:7  1016 eh=1500 þ Aeh=100 ; ð6Þ
wandering, if their dimension is large compared with the
14
beam size; beam broadening on the contrary. On short where A = 1.7  10 and v = 21 m/s.
timescale, the beam is broadened and deflected in different Following the same model, the short-term width can be
directions; integrating over time-scales larger than beam- estimated as:
wandering, the global effect is a broadening of the beam.  ( "  1=3 #)2
2
In a real scenario, from an Earth station a satellite is L 4:2L r0
w2ST ¼ w20 1 þ 2 þ 2 1  0:26 ; ð7Þ
seen rising, orbiting and setting at different positions in Z0 kr 0 w0
the sky, related to its orbit. At each time the satellite posi-
A dominant beam displacement in Eq. (2) can be ideally
tion can be defined, in the Earth station reference system,
completely compensated via an accurate pointing system,
by zenith and azimuth angles and by link distance.
leading to a long-term width that reduces to short-term
The thickness of the atmospheric layers crossed by the
width. On the other hands, turbulence compensation is effec-
beam depends on the zenith angle and, as we will show
tive only if b is higher than wST. As shown by Toyoshima
in the following, influences the beam propagation and the
et al. (2005), for a real space link, the beam displacement is
link budget.
related to the residual tracking error and the coalignment
error due to turbulence, hence a fine pointing and tracking
2.1.1. Uplink
system converges in wLT ffi wST. In our simulation and
A Gaussian beam of waist w0 and intensity I0, has an
analysis we deal with a worst-case scenario, where the beam
average long-term spot that tends to a Gaussian spatial dis-
displacement is uncompensated.
tribution of intensity (Andrews et al., 1995; Bonato et al.,
The power P at receiver can be estimated as:
2009; Toyoda, 2005):
Z R
2 =w2 2 2
hIðr; LÞi ¼ I 0 e2r LT ; ð1Þ P ¼ 2pI 0 qe2ðq =wLT Þ dq; ð8Þ
0
where r is the radial distance from the center axis of the
where R is the entrance aperture radius; it follows the link-
beam, L is the propagation distance to the receiver, wLT
efficiency g as:
is the long-term beam width defined as:  
2 2

w2LT ¼ w2ST þ 2hb2 i; ð2Þ g ¼ g0 1  e2R =wLT : ð9Þ


804 A. Tomaello et al. / Advances in Space Research 47 (2011) 802–810

The factor g0 can be set to 0.1 at the zenith to take into ac- Earth albedo (Bonato et al., 2009). Coupling the related
count the detection efficiency, the pointing losses and the model’s equation and doing the math, we obtain the num-
atmospheric attenuation (Aspelmeyer et al., 2003). More ber of background photons collected by the receiver system
in general g0 can be related to the zenit angle by the follow- as a function of bandwidth units and time units Dm [nm], Dt
ing equation (Hemmati, 2008) that is an accurate approxi- [s]:
mation up to 70°:
N DAY ¼ aE R2@S ðIFOVÞ2 H Sun ; ð12Þ
4:34sð0Þ secðhzen Þ
g0 ¼ LP LTX LRX  10ð
 10 Þ; ð10Þ
where aE is the Earth albedo; R@S is the radius of the tele-
where LP, LTX, LRX are the pointing losses, the transmitter scope in space; IFOV is the telescope field of view; HSun is
losses and the receiver losses that from Aspelmeyer et al. the solar spectral irradiance (photons s1 nm1 m2).
(2003) can be set respectively at 0.2, 0.8 and 0.8. s is the
depth of path at the zenith; that can be evaluated by the 2.2.2. Uplink (night-time operation)
atmospheric transmittance (ffi0.8 at 800 nm): In the same way, we can analyse the background noise
T ¼ es ; ð11Þ for a night-time operation uplink. At this time the domi-
nant noise sources are Earth black-body emission, diffused
moonlight and scattered light from human activities.
2.1.2. Downlink The noise related by moonlight can be evaluated by:
The beam travels through the atmosphere only in the 2
last path of the channel. Its size is then large enough to ðIFOVÞ
N NIGHT ¼ aE aM R2M R2@S H Sun ; ð13Þ
make the beam wandering effect negligible compared to d 2EM
the beam spreading due to the short-term effect. The turbu-
where aM is the Moon albedo; RM is the Moon radius; dEM
lent eddies appear to be much smaller than beam diameter.
is the Earth–Moon distance.
Moreover, compared with the uplink, the beam spreading
The noise component due to Earth black body radiation
appears to be less strong.
can be negligible for a real QKD system: the numerical esti-
This consideration implies a reduced attenuation for the
mation shown that it is three order of magnitude less that
downlink with respect to the uplink.
of moonlight (Bonato et al., 2009).
For QKD experiments, in which the working sites are
2.2. Background noise astronomical observatories located in suitable positions,
the human light is not a problem.
2.2.1. Uplink (day-time operation)
In a day-time operation uplink communication scenario,
2.2.3. Downlink noise
the main source of background noise is the Sun: more in
In a downlink scenario: a receiver telescope based on an
particular, it is the light diffused by the Earth and collected
Earth station, points at a satellite in the sky.
by the telescope (see Fig. 1).
Following Bonato et al. (2009) and Miao et al. (2005),
A simple model of this process can be obtained assum-
the noise power received by the telescope is:
ing a Lambertian diffusion and taking into account the
P b ¼ H b Xfov pR2@E Dm; ð14Þ
where Hbk is the brightness of the sky background
[W m2 sr1 lm1]; Xfov is the field of view of the telescope
in [sr]; R@E is the receiving telescope radius at the Earth sta-
tion; Dm is the optical bandwidth [lm].

2.3. SNR

The SNR, defined as the ratio between the single pho-


tons of signal and the noise photons at the receiver, in
the our scenario becomes:
g
SNR ¼ ; ð15Þ
eN
where g and eN are the number of signal and noise photons
at a detection time Dt and for a bandwidth Dm.
For the uplink in night time operation we have:
Fig. 1. Model for the uplink day-time background noise estimation. The g0 ð1  expð2R2@S =w2LT ÞÞd 2EM
same model can be adapted for night-time operation considering the SNR ¼ ; ð16Þ
Moon as light source noise. aE aM R2M R2@S IFOV2 H Sun DmDt
A. Tomaello et al. / Advances in Space Research 47 (2011) 802–810 805

and in the same way for the downlink in night time


operation:

g0 ð1  expð2R2@E =w2LT ÞÞhm


SNR ¼ ; ð17Þ
H b Xfov pR2@E DmDt

where h is the Planck constant.


We emphasize that, in this formulation the contribution
due to imperfect QKD system is neglected.

3. QKD from satellite


Fig. 2. Long term beam width wLT as function of link distance L.
With the model presented in Section 1, we have the tools Simulation at k ¼ 800 nm, Earth telescope radius rT = 75 cm, space
to evaluate the channel properties: attenuation and SNR. telescope radius R = 10 cm.
The simulation results will give us the capability to set
the feasibility or unfeasibility of QKD from satellite and
to satellite. the unperturbed expected position. On the same way, a
We will mainly focus on the following families of beam affected at the end of its path will be barely modified.
satellites: To give an overall view of the beam propagation, we
report the simulation of the beam displacement b in
– Europe global positioning system (Galileo). Fig. 3. Here it is clear that the beam displacement is negli-
– USA global positioning system (GPS). gible for downlink and that the main input affecting the
– Low Earth Orbit (LEO). beam is the broadening due to turbulence eddies, therefore
– Geostationary Earth Orbit (GEO). related to short term beam effects. Increasing the link dis-
tance as the same time, for the uplink a beam displacement
The acronyms in brackets will be used in the graphs to monotonically increasing with the distance is related to
refer to satellites families. atmosphere at the beginning of the path.
The analysis of the link as a function of the zenith angle,
will be useful to estimate the link budget in a real scenario,
in fact, the orbit of a satellite seen from an Earth station, 3.2. Uplink Attenuation and SNR
can be represented as a set of terns (zenith angle, azimuth
angle, link distance) evolving in time. Bonato et al. (2009) have shown that the most suitable
operation time to establish a space-Earth quantum link is
the night time both for up and down links, this mainly due
3.1. Beam simulation
to the achievable low level of background noise (sky bright-
ness is on the order of: 1.5  105 W m2 Sr1 lm1).
As first we simulated the effects of the atmosphere on the
In this section we simulate the case in which the Earth
beam propagation for uplink and downlink.
station play the role of transmitter and the space terminal
In our simulation we focused on: a real Earth station
is the receiver. After presenting the effects on beam propa-
(MLRO observatory) which has a telescope radius of
gation will be estimated the SNR: Figs. 4 and 5 represent
75 cm and an IFOV of 0.016°, and on a reasonable (for
the actual state of the art) space terminal with a telescope
diameter of 20 cm and an IFOV of 100 lrad. Likewise we
choose a bandwidth Dm = 1 nm, and a detection time
Dt = 1 ns.
The long term beam width, as we previously said, repre-
sents intuitively the beam width seen over a long time expo-
sure picture. It is due to the integration of the beam
wandering effect. As reported in Fig. 2, the effect of beam
wandering is strongly different for the uplink and down-
link, up to about two orders of magnitude. A few ten thou-
sand meters of beam radius is reached in a very long
distance (like GEO satellite) in the uplink; while a beam
radius of the order of few hundred meters is calculated in
the case of downlink. This is due to the beam degradation
induced by atmosphere: in fact, a beam affected at the start Fig. 3. hb2i as function of the link distance L. Simulation at k ¼ 800 nm,
of its path will arrive at the receiver plane far away from Earth telescope radius rT = 75 cm, space telescope radius R = 10 cm.
806 A. Tomaello et al. / Advances in Space Research 47 (2011) 802–810

term effect grows up compared to a link at zenith. This is


mainly due to a longer path over the atmosphere layer.
This strongly affects the link attenuation (Fig. 6) and the
SNR (Fig. 7), inducing a limitation of 40° of zenith angle
for higher LEO: the SNR falls below zero preventing
QKD. For a LEO satellite, and for satellites in general,
the zenith angle also defines a boundary angle for which
a quantum link at angles greater than this is unfeasible.
Only the portion of sky delimited by the cone formed by
the zenith direction and the zenith boundary angle is suit-
able for quantum communication. Quantum communica-
tion for the simulated transmitter and receiver with GPS,
Galileo and in general GEO satellites, is unfeasible for
Fig. 4. Long term beam width wLT as function of the zenith angle for any zenith angle (Fig. 7).
different satellites in uplink. Simulation at k ¼ 800 nm, Earth telescope The dependence of the link budget to the zenith angle
radius rT = 75 cm, space telescope radius R = 10 cm. leads to the following consideration:

 For a QKD the key generation rate varies with the


zenith angle; if the cone is too narrow an high data
rate is needed to obtain a sufficient key length and
overcome the losses.
 The time from the satellite set to the limit in which the
SNR allows QKD must be optimized for pointing,
tracking and synchronization to take advantage on
quantum link when the satellite goes on the “good
SNR cone” (Fig. 8).
 The zone out of the good SNR cone, in which the
satellite is however visible from the earth station
would be used for classical communication, not only
for establishing the link with the satellite but also to
perform the public exchange of polarization basis as
expected in the BB84 and B92 QKD protocols.
Fig. 5. hb2i as function of the zenith angle for different satellites in uplink.
Simulation at k ¼ 800 nm, Earth telescope radius rT = 75 cm, space The results are much more promising in the case of night
telescope radius R = 10 cm. time operations, as long as a narrow field of view and a rel-
atively short link are used. In a real system, the challenge to
decrease the IFOV has to deal with the precision of point-
respectively the long term beam effect and the beam devia- ing and tracking; so it is quite difficult to do QKD in every
tion as a function of the zenith angle. As shown, increasing conditions, more for far satellites as GPS, Galileo, GEO;
the zenith angle, the beam deviation, and hence the long LEO satellites seem to be the best candidate.

Fig. 6. Attenuation g as function of the zenith angle for different satellites Fig. 7. SNR as function of the zenith angle for different satellites in
in uplink. Simulation at k ¼ 800 nm, Earth telescope radius rT = 75 cm, uplink. Simulation at k ¼ 800 nm, Earth telescope radius rT = 75 cm,
space telescope radius R = 10 cm. space telescope radius R = 10 cm. Dm = 1 nm, Dt = 1 ns. IFOV = 0.006°.
A. Tomaello et al. / Advances in Space Research 47 (2011) 802–810 807

Fig. 10. hb2i as function of the zenith angle for different satellites in
downlink. Simulation at k ¼ 800 nm, Earth telescope radius rT = 75 cm,
space telescope radius R = 10 cm.

Fig. 8. Model of the “good SNR cone” seen from the Earth station. 2.2 dB. In both cases the results, de facto prove the unfea-
sibility of quantum communication with the simulated
3.3. Downlink attenuation and SNR setup, but gives good wishes related to further technologi-
cal improvements.
By exploiting the results obtained above for the atmo-
sphere model we can study the case of a downlink in which
the transmitter is on space, Figs. 9 and 10 show the effects
on beam propagation at different zenith angle for different
satellite. The downlink scenario is more promising than in
the uplink: the beam displacement is negligible because the
beam is only broadened. For a LEO satellite the trend
reported in Fig. 10 shows that the effect due to altitude
variations is dominant if compared to the increased layer
thickness at higher zenith angle.
In Figs. 11 and 12 are reported respectively the simula-
tion results for attenuation and SNR: the SNR appears to
be almost constant at 36 dB for satellite orbiting at the
lower limit of LEO.
The “good SNR cone”, for a downlink covers a greater
portion of sky compared to uplink, allowing a longer time- Fig. 11. Attenuation g as function of the zenith angle for different
slot for quantum communication (Fig. 12). For Galileo sat- satellites in downlink. Simulation at k ¼ 800 nm, Earth telescope radius
ellites, the SNR is closed to 0.7 dB at zenith; the situation is rT = 75 cm, space telescope radius R = 10 cm.
most promising for GPS satellites: the SNR is around

Fig. 9. Long term beam width wLT as function of the zenith angle for Fig. 12. SNR as function of the zenith angle for different satellites in
different satellites in downlink. Simulation at k ¼ 800 nm, Earth telescope downlink. Simulation at k ¼ 800 nm, Earth telescope radius rT = 75 cm,
radius rT = 75 cm, space telescope radius R = 10 cm. space telescope radius R = 10 cm. Dm = 1 nm, Dt = 1 ns. IFOV=0.016°.
808 A. Tomaello et al. / Advances in Space Research 47 (2011) 802–810

The actual feasibility of QKD is related to the link dis- 1  el  lel
Dffi ; ð19Þ
tance. In the range of LEO orbits, a good SNR can be 1  egl
achieved without stretching the telescope field-of-view to where g is the link attenuation and l the mean photon
the technological limit. Small improvements on IFOV number.
can shift the link distance feasibility limit toward higher We remind that for a coherent state:
orbits, allowing quantum communication even with GPS P1
X1
Y n P n ðlÞen
and Galileo satellites. Ql ¼ Y n P n ðlÞ; El ¼ n¼0 ; ð20Þ
n¼0
Ql

4. Discussion where Yn = Y0 + 1  (1  g)n is the yield of n-photon


pulses, Pn(l) is the Poisson distribution of a weak laser
In the previous section we addressed the modelization source used instead of an ideal single photon source and
and link budget analysis for quantum space channel. In a en = Y0/2Yn is the error rate of n-photon signal.
practical implementation some considerations have also The multi-photon tagging drastically reduces the key-
to be taken into account to obtain an effective QKD generation rate if we consider that Eve intercepts all the
system. multi-photon pulses. An accurate analysis of the channel,
The secret key generation rate, that represents the prob- and so a more closed to real scenario estimation of D, could
ability of a signal pulse to become part of the final key, is lead to a less pessimistic value than predicted by Eq. (19):
crucial for the choice of the quantum protocol. this is the basic idea of decoy-state protocol.
In a practical QKD implementation, single photons are In a two-decoy-states protocol (Hwang, 2003), we have
implemented with weak coherent pulses: these sources have three intensities with different mean photon number: m for
a non zero probability of multi-photon pulses. This benefits weak decoy states, l for signal states (l > m), and vacuum
Eve, that can attack the protocol via a photon-number- state. Alice chooses one of these intensities and sends a
splitting (PNS). Unlikely, this is what happens in a high pulse to Bob with the set mean photon number. The use
losses channel: only the multi-photon pulses have a high of different intensity states provides information on the
probability to arrive at Bob receiver and can easily be channel attenuation or eavesdropping. The key generation
attacked by Eve. rate becomes (Ma et al., 2005; Moli et al., 2008):
Inamori et al. (2007) and Gottesman et al. (2004), R P q Ql f ðEl ÞH 2 ðEl Þ þ QL1 1  H 2 eU1 ; ð21Þ
showed that a secure key can still be distilled, if the portion
of tagged bit (tagged bit and tagged pulses will be used where
below with the same meaning) is known: on this way, a Em Qm em  e0 Y 0
few new protocols have been proposed, one of the most QL1 ¼ lel Y L1 ; eU1 ¼ ; ð22Þ
Y L1 m
famous is the decoy-states (Hwang, 2003). In a decoy-states
protocol different light intensities are used to probe the and:
 
channel attenuation and to accurately estimate the amount l m l m
2
l2  m2
of tagged bit. Y L1 ¼ Qm e  Ql e 2  Y0 ð23Þ
lm  m2 l l2
The model presented below, based on Ma et al. (2005),
extends the analysis to longer distance link than that used In Fig. 13 we reported the key generation rate simulated
in Bonato et al. (2009). According to this model the range for our downlink, using a BB84 protocol (l = 0.01) and a
for QKD appears as shorter with respect to what we previ- BB84 with two-decoy states (l = 0.4, m = 0.27), as a func-
ously estimated for standard protocols (i.e. BB84–B92), tion of the link distance.
and longer for decoy-states protocol.
From Ma et al. (2005) and Moli et al. (2008), we can
evaluate a lower bound for the key-generation rate:
   
El
R P q Ql f ðEl ÞH 2 ðEl Þ þ XQl 1  H 2 ; ð18Þ
X

X is defined as X = 1  D, where D is the portion of tagged


bits, that represents the Eve information; q is the protocol
efficiency (1/2 for BB84, 1/4 for B92), Ql and El are de-
fined in literature as the gain and the quantum bit error
rate of the signal state (QBER), f(El)  1.22 is the error
correction efficiency and H2 is the binary Shannon entropy.
In a worst-case analysis, for a weak coherent pulses Fig. 13. Downlink key generation rate: BB84 protocol (l = 0.01), B92
source, D is given by the ratio of the portion of multi-pho- (l = 0.01) and BB84 with two-decoy state (l = 0.4, m = 0.27), as a function
ton pulses over the non-empty pulses detected by Bob: of the link distance.
A. Tomaello et al. / Advances in Space Research 47 (2011) 802–810 809

Clearly a decoy states protocol appears to be an essen- allowing few tens of dB for the SNR in downlink and
tial choice to reach long distance QKD, that is limited to uplink.
few hundred kilometers with classical protocol. In a high Technological improvements in pointing and tracking of
losses link, the capability to determine the portion of satellite and so in the telescope field of view can extend the
tagged bit through a channel analysis allows the use of results to higher orbits: Galileo and GPS satellites are the
pulses with a mean photon number higher than a classical next candidate.
QKD protocol; in fact, in the latter case the value must be In a real implementation the non idealities are not neg-
kept low to reduce the probability of multi-photon pulses ligible, and they also play an important role in the choice of
and hence of pulses that are PNS attachable. For real the protocol: due to high losses of the channel, an accurate
QKD implementation, some different protocols that maxi- estimation of the link attenuation and tagged bits is neces-
mize the key generation rate have been proposed: one of sary to avoid an excessive and conservative reduction in the
the most recent is the B92 with strong reference pulse key generation rate due to a pessimistic evaluation of the
(SRP) by Tamaki (2008). The protocol uses the two weak channel.
coherent states of the B92 together with a strong reference
pulse to be robust against channel losses. Acknowledgments
Satellite based links are identified mainly by two proper-
ties, high losses and short duration, that give a short The authors are glad to acknowledge many fruitful dis-
expected key length. The Gottesman–Lo–Lutkenhaus–Pre- cussions with Prof. C. Barbieri, Prof. G. Cariolaro, Prof. S.
skill (GLLP) key generation rate formulas, used in this sec- Ortolani, Dr. F. Tamburini, Dr. I. Capraro and Dr. T.
tion, are valid on the infinite key assumption. Scarani and Occhipinti. This work has been carried out within the Stra-
Renner (2008) showed that, for a BB84 protocol with one- tegic-Research-Project QUINTET of the Department of
way post-processing, at least 105 signals need to be Information Engineering, University of Padova and the
exchanged and processed to obtain a nonzero key genera- Strategic-Research-Project QUANTUMFUTURE of the
tion rate and so an unconditional secure key. From a University of Padova.
ground station, a LEO satellite is visible for a few minutes
and only in a restricted portion of them the SNR is suitable References
for QKD: we can assume an effective communication time
of 180 s. A link rate of 550bps is hence required, and con- Andrews, L.C., Philips, R.L., Yu, P.T. Optical scintillation and fade
statistics for a satellite-communication system. Appl. Opt. 34 (33),
sidering the key generation rate for a 500 km link, we
7742–7751, 1995.
obtain a transmission frequency of 3 MHz for a BB84 pro- Aspelmeyer, M., Jennewein, T., Pfennigbauer, M., et al. Long distance
tocol and 100 kHz for a BB84 with decoy state. The same quantum communication with entangled photons using satellites.
considerations can be extended for uplinks in which, due to IEEE J. Sel. Top. Quant. Electr. 9 (6), 1541–1551, 2003.
higher losses, the requirements on transmission frequency Bonato, C., Tomaello, A., Da Deppo, V., et al. Feasibility of satellite
will be more restrictive. quantum key distribution. New J. Phys. 11, 045017, doi:10.1088/
1367-2630/11/4/045017, 2009.
Dios, F., Rubio, J.A., Rodrı́guez, A., et al. Scintillation and beam-wander
analysis in an optical ground station-satellite uplink. Appl. Opt. 43
5. Conclusion (19), 3866–3873, 2004.
Gottesman, D., Lo, H.-K., Lutkenhaus, N., et al. Security of quantum
key distribution with imperfect devices. IE EE Inform. Theory,
The presented model is useful to analyze a real satellite
doi:10.1109/ISIT.2004.1365172, 2004.
quantum link, and its time evolution. Hemmati, H. Near-Earth Laser Communication. CRC Press, 2008.
The dependence of the link budget with the orbit param- Hwang, W.-Y. Quantum key distribution with high loss: toward global
eters, described by the zenith angle and satellite altitude, secure communication. Phys. Rev. Lett. 91, 057901, doi:10.1103/Phys
identifies the positions of the satellite in the sky that allow RevLett.91.057901, 2003.
QKD, and so the available time for communication. Inamori, H., Lutkenhaus, N., Mayers, D. Unconditional security of
practical quantum key distribution. Eur. Phys. J. D 41, 599–627,
This is also important to define the communication pro- doi:10.1140/epjd/e2007-00010-4, 2007.
tocol and the time slot distribution for: pointing, tracking, Ma, X., Qi, B., Zhao, Yi., et al. Practical decoy state for quantum key
synchronization and quantum communication. distribution. Phys. Rev. A 72, 012326, doi:10.1103/PhysRevA.
In this paper we have demonstrated for different types of 72.012326, 2005.
satellite that the feasibility of QKD is at hand in various Miao, E.-l., Han, Z.-f., Gong, S.-s., et al. Background noise of satellite-to-
ground quantum key distribution. New J. Phys. 7, 215, doi:10.1088/
cases: best for LEO satellite where an SNR of 35 dB can 1367-2630/7/1/215, 2005.
be achieved. The simulations have pointed out the optical Moli, L., Rodriguez, A., Seco-Granados, G., et al. Quantum key
and satellite visibility conditions in order to achieve or distribution (QKD) using LEO and MEO satellites and decoy states.
exceed an acceptable SNR, highlighting the portion of IEEE Signal Process. Space Commun., doi:10.1109/SPSC.2008.
4686743, 2008.
orbit seen from an Earth station that can exploit QKD.
Scarani, V., Renner, R. Quantum cryptography with finite resources:
The results show that: with the telescope of the MLRO unconditional security bound for discrete-variable protocols with one-
Observatory and with a space terminal with a telescope of way postprocessing. Phys. Rev. Lett. 100, 200501, doi:10.1103/Phys-
10 cm of radius, only LEO satellite can be used for QKD, RevLett.100.200501, 2008.
810 A. Tomaello et al. / Advances in Space Research 47 (2011) 802–810

Sergienko, A.V. Quantum Communications and Cryptography. Taylor & Toyoshima, M., Yamakawa, S., Yamawaki, T., et al. Long-term statistics
Francis Group, 2006. of laser beam propagation in an optical ground-to-geostationary
Tamaki, K. Unconditionally secure quantum key distribution with satellite communications link. IEEE Trans. Antennas Propag. 53 (2),
relatively strong signal pulse. Phys. Rev. A 77, 032341, doi:10.1103/ 842–850, 2005.
PhysRevA.77.032341, 2008. Tyson, R.K. Principles of Adaptive Optics, second ed Academic Press,
Toyoda, M. Intensity fluctuations in laser links between ground and a 1997.
satellite. Appl. Opt. 44 (34), 7364–7370, 2005.

You might also like