ARTICLES

PUBLISHED ONLINE: 24 JULY 2017 | DOI: 10.1038/NPHOTON.2017.116

Long-distance free-space quantum key distribution

in daylight towards inter-satellite communication
Sheng-Kai Liao1,2†, Hai-Lin Yong1,2†, Chang Liu1,2†, Guo-Liang Shentu1,2†, Dong-Dong Li1,2, Jin Lin1,2,
Hui Dai1,2, Shuang-Qiang Zhao3, Bo Li1,2, Jian-Yu Guan1,2, Wei Chen1,2, Yun-Hong Gong1,2, Yang Li1,2,
Ze-Hong Lin3, Ge-Sheng Pan1,2, Jason S. Pelc4, M. M. Fejer4, Wen-Zhuo Zhang1,2, Wei-Yue Liu3,
Juan Yin1,2, Ji-Gang Ren1,2, Hong Wang2,5, Qiang Zhang1,2,5*, Cheng-Zhi Peng1,2*
and Jian-Wei Pan1,2*

In the past, long-distance free-space quantum communication experiments could only be implemented at night. During the
daytime, the bright background sunlight prohibits quantum communication in transmission under conditions of high
channel loss over long distances. Here, by choosing a working wavelength of 1,550 nm and developing free-space single-
mode fibre-coupling technology and ultralow-noise upconversion single-photon detectors, we have overcome the noise due
to sunlight and demonstrate free-space quantum key distribution over 53 km during the day. The total channel loss is
∼48 dB, which is greater than the 40 dB channel loss between the satellite and ground and between low-Earth-orbit
satellites. Our system thus demonstrates the feasibility of satellite-based quantum communication in daylight. Moreover,
given that our working wavelength is located in the optical telecom band, our system is naturally compatible with ground
fibre networks and thus represents an essential step towards a satellite-constellation-based global quantum network.

S
atellite-based quantum communication has proven to be a feas- Working wavelength selection
ible way to achieve a global-scale quantum communication We first switched the working wavelength to 1,550.14 nm from the
network1–9. Very recently, a low-Earth-orbit (LEO) satellite 700–900 nm used in all previous experiments. The 1,550 nm wave-
was launched10 for this purpose. However, with a single satellite, length is known to be an atmospheric window. In fact, the trans-
an inefficient 3 day period11 is required to provide worldwide con- mission efficiency is slightly higher at 1,550 nm than at 800 nm,
nectivity. On the other hand, similar to how the Iridium system12 as shown in Fig. 1a, and from the solar spectrum in Fig. 1b we
functions in classical communication, a satellite constellation can see that the sunlight intensity at 1,550 nm is around five times
(SC) composed of many quantum satellites could provide global weaker than it is at 800 nm. Furthermore, the main type of scattering
real-time quantum communication. Such an SC is expected to of solar noise for links between a satellite and Earth or between two
operate with LEO satellites or high-Earth-orbit satellites such satellites is Rayleigh scattering, the intensity of which is proportional
as geosynchronous orbit (GEO) satellites. The probability of a to 1/λ 4. Therefore, Rayleigh scattering at 1,550 nm is only 7% of its
satellite being in the Earth shadow zone decreases rapidly with value at 800 nm. In total, the background noise with 1,550 nm light
increasing orbit height (Fig. 1). A LEO satellite system has a can be reduced to 3% of the background noise of 800 nm light. We
probability of ∼70% of being in the sunlight zone, whereas for a measured the noise count rate of 1,550 nm light in the daylight case
GEO satellite this rises to ∼99% (ref. 13). Meanwhile, the total by pointing a telescope at the sky to simulate satellite-to-Earth
channel loss between a LEO satellite and the Earth and between communication. The result was smaller by a factor of 22.5 than for
LEO satellites is typically ∼40–45 dB (refs 14,15). Therefore, to 850 nm light. Note that all existing free-space experiments without
test the feasibility of an SC-based quantum network, quantum a satellite, including this work, have been implemented on Earth,
communication through a channel with at least ≥40 dB loss in and the direction of the free-space communication is parallel to
daylight is essential. the Earth rather than pointing at the sky. In this situation, Mie scat-
There have been several pioneering experiments on daylight tering, which does not follow the 1/λ 4 relation, will be the main noise
quantum communication before our work16–22. Although the exper- source instead of Rayleigh scattering. Moreover, 1,550 nm is the
iments were novel, the maximum loss calculated from them was telecom-band wavelength and is widely used for fibre-optical com-
only ∼20 dB. The main cause of the unsatisfactory performance munication. Using the same wavelength for both free-space and
was the strong background noise from the scattered sunlight, fibre-optical communication is an optimal choice.
which was typically five orders of magnitude greater than the back-
ground noise during the night23. We can reduce this noise in three Upconversion detectors and spectral filtering
ways: working wavelength selection, spectrum filtering and Despite the advantages of operating at 1,550 nm, researchers have
spatial filtering. been reluctant to use this wavelength due to a lack of good

1
Shanghai Branch, National Laboratory for Physical Sciences at Microscale and Department of Modern Physics, University of Science and Technology of
China, Shanghai 201315, China. 2 Synergetic Innovation Center of Quantum Information and Quantum Physics, University of Science and Technology of
China, Shanghai 201315, China. 3 School of Information Science and Engineering, Ningbo University, Ningbo 315211, China. 4 Edward L. Ginzton Laboratory,
Stanford University, Stanford, California 94305, USA. 5 Jinan Institute of Quantum Technology, Shandong Academy of Information and Communication
Technology, Jinan 250101, China. †These authors contributed equally to this work. * e-mail: [email protected]; [email protected]; [email protected]

NATURE PHOTONICS | ADVANCE ONLINE PUBLICATION | www.nature.com/naturephotonics 1

© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
ARTICLES NATURE PHOTONICS DOI: 10.1038/NPHOTON.2017.116

a 1.0
Zenith angle = 0° Zenith angle = 30°
Sun
0.9 Zenith angle = 45° Zenith angle = 70°
0.8
0.7

Transmittance
0.6
0.5
0.4
0.3
0.2
Satellite
0.1 810 nm 1,550 nm
0.0
0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Wavelength (µm)

Satellite b Solar radiation spectrum

2.5

Spectral irradiance (W m−2 nm−1)

50% 2.0

1.5

LEO 1,200 km 810 nm

1.0
23%

Shadow 0.5
1,550 nm
0.0
0.6% 250 500 750 1,000 1,250 1,500 1,750 2,000 2,250 2,500

GEO Wavelength (nm)

Figure 1 | Satellite-constellation-based global quantum network. A global quantum network needs many LEO satellites or several geosynchronous orbit
satellites to create a satellite constellation. The time that a satellite is in the Earth shadow zone, which we call night, is inversely proportional to the orbit
height of the satellite. a, Transmittance spectra from visible to near-infrared light in the atmosphere at selected zenith angles. b, Solar radiation spectrum
from visible to near-infrared light.

commercial single-photon detectors for the telecom band. We have designed to be ∼100 μrad (refs 2,4,5). In most previous daylight
developed a compact upconversion single-photon detector (SPD)24 quantum key distribution (QKD) experiments, a similarly large
(Fig. 2e). In this detector, a telecom-band photon is mixed with a FOV was also used to improve the coupling efficiency at the receiver
strong pumping signal photon of 1,950 nm in a wavelength division (for example, 220 μrad)17. However, the larger the FOV, the larger
multiplexing (WDM) coupler and then sent to a fibre-pigtailed the noise introduced to the receiver. Using a single-mode fibre
periodically poled lithium niobate (PPLN) waveguide. The gener- (SMF) to couple the signal photons is an extreme way to improve
ated photons are collected by an anti-reflection-coated objective the SNR in spatial filtering, with a FOV of <10 μrad. With SMF
lens and then separated from the pump and the spurious light by coupling, the noise count introduced by stray light can be reduced
a dichroic mirror (DM), a short-pass filter and a bandpass filter. by a factor of several hundred.
A volume Bragg grating (VBG) with 95% reflection efficiency is At the same time, the coupling efficiency must be maintained at a
used to further suppress the noise from both spontaneous Raman high level to make the SNR of the experiment larger. In previous
scattering generated in the nonlinear process and the sunlight back- experiments25, ∼0.1% SMF coupling efficiency has been achieved,
ground. The sum-frequency generation (SFG) photons are then col- which is insufficient for quantum communication. We therefore
lected and detected by a silicon avalanche photodiode (SAPD). developed the SMF coupling technique further.
Using a pump power of 200 mW, the total system detection effi- Unlike previous free-space experiments, which required many
ciency is 8%, with an average dark count rate of ∼20 Hz, which is optical elements6,7, we used the minimal amount of optics to
the same as the intrinsic dark count of the SAPD. design the receiving telescope and to reduce the optical attenuation
Note that, in our experiment, the spectral bandwidth of the VBG and aberration. The focal length of the off-axis primary mirror was
filter is 0.05 nm (full-width at half-maximum, FWHM) at a centre set to 2,000 mm to optimize the SMF coupling efficiency. We
wavelength tunable near 864 nm, corresponding to a bandwidth located only one fast-steering mirror (FSM) and one DM between
of 0.16 nm at the signal wavelength. The phase-matching condition the primary mirror and the SMF, and we developed an optical track-
and VBG of the detector itself realize spectrum filtering. Meanwhile, ing system with 300 Hz feedback frequency to stabilize the SMF
stable distributed feedback lasers (DFBs) with a 2 GHz bandwidth coupling (see Supplementary Information). With this set-up we
are used as the signal source, which guarantees that no signal will could obtain a SMF coupling efficiency of >30% in the laboratory,
be filtered out. Such a narrowband filtering technology reduces while in the outdoor experiment the SMF coupling efficiency was
the noise by a factor of about 100 compared with the 3–10 nm reduced to 5% by horizontal air turbulence (with 3 μrad tracking
filters used in previous night experiments. precision). We emphasize that this 5% efficiency is still much
higher than the 0.1% efficiency achieved in previous experiments25.
Spatial filtering with single-mode fibre coupling Furthermore, in future ground-to-satellite experiments we will
To further improve the signal-to-noise ratio (SNR) we reduced the receive signals vertically, and the turbulence is weaker. An efficiency
field of view (FOV) for our receiving system for spatial filtering. For of 30%, similar to that obtained in the laboratory test, can
free-space quantum communication at night, the FOV is usually be expected.

2 NATURE PHOTONICS | ADVANCE ONLINE PUBLICATION | www.nature.com/naturephotonics

© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE PHOTONICS DOI: 10.1038/NPHOTON.2017.116 ARTICLES
c d 850 nm
Detector
PBS
WDM BS e
N PBS
20 m

671 nm

DM CMOS
FSM 53 km e
Fibre Input
Pump
OSA PPLN IF Lens
Lens

VBG

b
QINGHAI LAKE

532 nm

CCD
a 1,548 nm
FSM BS
Lens
IF TC OSA BS BS PBS
PBS
CMOS a
+ −
TC
WDM V
810 nm H LD: 1,550.14 nm 10 km

Figure 2 | Birds-eye view of the 53 km QKD experiment in daylight. Alice and Bob are located on either side of Qinghai Lake. a, The 1,550 nm laser diodes
(LDs) are encoded to four quantum states (|H〉,|V〉,|+〉,|–〉) by two polarizing beamsplitters (PBSs) and one beamsplitter (BS). An 810 nm beacon laser and a
1,548 nm reference laser are combined and sent to a triplet collimator (TC) for optical alignment and tracking. An optical spectrum analyser (OSA) is used
to calibrate the signal spectrum. WDM, wavelength division multiplexer. b, The sending terminal has a telescope system on a two-axis rotation stage and an
optical tracking system. CCD, charge-coupled device; FSM, fast-steering mirror; IF, interference filter. c, The receiving terminal has an off-axial parabolic
mirror and a SMF coupling module. d, Received photons are transmitted to the detector via a 20 m fibre. At the detection part, photons are measured with
two PBSs, one BS and four detectors. e, Upconversion single-photon detector modules. A narrow-bandwidth volume Bragg grating (VBG) is used to narrow
the working spectrum of the detectors and reduce the noise. For details of optical alignment and tracking, see Methods. PPLN, periodically poled lithium
niobate. Map data: Google, CNES/Airbus, Digital Globe, Landsat/Copernicus.

Field test across Qinghai Lake (mounted on a two-dimensional platform, Fig. 2b). We minimized
Figure 2 presents the set-up of our experiment on Qinghai Lake. We the sending divergence angle of the 1,550 nm beam to reduce
chose Qinghai Lake because it has a good visual point of view geometric loss. We chose a Schmidt–Cassegrain system with a d =
(length of 53 km). The sending terminal (Alice) was located at 254 mm primary mirror for sending, because the divergence angle
Heimahe village (N36°49′01.3′′, E99°44′51.3′′), Qinghai Province, is proportional to the wavelength and inversely proportional to the
China (Fig. 2a). Alice sent the 1,550 nm signal beam through a size of the primary mirror. The measured divergence angle was
53 km free-space link across Qinghai Lake to the receiving terminal 12 μrad, which is very close to the diffraction limit and four times
(Bob), which was located at Quanji village (N37°16′42.4′′, E99°52′59.9′′), smaller than our previous sending system6.
Qinghai Province. At the receiving terminal, the signal light was collected by a
At the sending terminal we developed a 1,550 nm light source receiving telescope (Fig. 2c), which consisted of a primary parabolic
with a decoy scheme26–28. We used four DFB lasers with a central mirror with a diameter of 420 mm and focal length of 2,000 mm
wavelength at 1,550.14 nm and FWHM of 0.02 nm to emit 500 ps and a SMF coupling module with an optical tracking system (see
pulses. We created a standard BB84 source by combining the Methods). The signal photons went through a 20-m-long SMF to
pulses with two polarizing beamsplitters (PBSs), one beamsplitter the detection system. A fibre BS was used to select a measurement
(BS) and one variable optical attenuator. At a clock frequency of basis, and two fibre PBSs together with four upconversion SPDs
100 MHz, the source randomly generates one of four polarization were used for detection (Fig. 2d). All detected signals were sent
states, |H〉, |V〉, |+〉 and |–〉, with one of three average photon into a time-to-digital converter (TDC) for analysis. The time
numbers per pulse (0.6, 0.14, 0). Here |H〉/|V〉 represents √

window was set to 1 ns, for timely filtering of the noise as in pre-
horizontal and vertical
√

polarization, | +〉 = (|H〉 + |V〉)/ 2 and vious experiments17,19. We also developed an efficient self-synchron-
| −〉 = (|H〉 − |V〉)/ 2. The intensity of the signal state was 0.6 per ized system based only on GPS, which could use fewer resources
pulse. We used two decoy states: the vacuum state and a state with than previous pulsed-laser synchronization systems4.
average photon number 0.14. The probability ratio of the signal In our QKD experiment, Alice and Bob extracted the final
and the two decoy states was 2:1:1. All random control signals were secure key out of the raw data following a standard decoy BB84
generated at high speed by a random number generator. The signal post-processing procedure27–29. The final key rate formula is
was then coupled into the SMF and collimated out into free space
with a triplet collimator (TC) and then sent to the telescope Rpulse ≥ qpμ {−Qμ f (Eμ )H2 (Eμ ) + Q1 [1 − H2 (e1 )]} (1)

NATURE PHOTONICS | ADVANCE ONLINE PUBLICATION | www.nature.com/naturephotonics 3

© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
ARTICLES NATURE PHOTONICS DOI: 10.1038/NPHOTON.2017.116

Table 1 | Experimental parameters and results.

T (s) Qμ Qv Y0 Eμ (%) Ev (%) R pulse R total (bits)
643 1.36 × 10–5 3.99 × 10–6 7.24 × 10–7 3.19 9.18 9.42 × 10–7 60,567
649 9.12 × 10–6 2.59 × 10–6 3.02 × 10–7 3.24 9.49 4.27 × 10–7 27,700
464 1.63 × 10–5 4.11 × 10–6 2.38 × 10–7 1.65 3.35 1.49 × 10–6 68,912
T is the effective time for QKD. Qμ and Qv are the gains for the signal states and decoy states, respectively. Y0 is the yield for vacuum states. Eμ and Ev are the quantum bit error rate (QBER) of the signal states and
decoy states, respectively. Rpulse is final key rate per clock cycle and Rtotal is the total final key size of the experiment.

where q = 1/2 is the basis reconciliation factor, pμ is the probability increase the efficiency of the upconversion detector to improve
of emitting signal states, Qμ and Eμ are the gain and error rate of the the secure key rate. Meanwhile, superconducting SPDs (SSPDs)
signal states, respectively, and f is the error correction efficiency. The have a higher detection efficiency and lower intrinsic dark count
low-density parity-check (LDPC) code was used for error correc- than upconversion SPDs. In our experiment, because of the
tion, with H2(e) = –elog2(e) – (1 – e)log2(1 – e) being the binary limited space and tough environment in the container, we chose
Shannon entropy function and Q1 (e1) the gain (phase error rate) an upconversion SPD for its compact size and room-temperature
when the source generates single-photon states. We carried out operation. With future miniaturization of SSPDs, we can definitely
and repeated QKD experiments from 15:30 to 17:00 local time for improve the performance of the entire system. Moreover, for future
several sunny days. Three typical groups of results are listed in satellite-to-ground communication with a very small zenith angle,
Table 1. We obtained 157,179 bits during 1,756 s effective time. Rayleigh scattering will dominate and we could have an even
The final key rate was 20–400 bits per second. The variation of better SNR than in our ground experiment with 1,550 nm light.
the final key rate was mainly due to channel loss changes in the Recently, a novel experiment34 has been carried out that implements
atmospheric environment. quantum state transfer from a GEO satellite to the ground and shows
All the data were collected in good weather, as was the case for the potential of continuous variable QKD in daylight. Overall, com-
the long-distance QKD experiments in refs 2–4 and 6. pared with discrete-variable QKD, for example using a homodyne
Unfortunately, implementing quantum communication in bad detector, continuous-variable QKD is more robust against background
weather is not possible with current technology. However, we note noise in daylight, but suffers from high channel losses43,44.
that there will be no turbulence or bad weather between satellites.
Only when considering quantum communication between a satel- Methods
lite and the ground must bad weather be taken into account. Methods and any associated references are available in the online
The internal modulation of the decoy/signal states guaranteed version of the paper.
our system security against photon number splitting30,31 and unam-
biguous-state-discrimination attack32. The total loss over our 53 km Received 1 March 2017; accepted 8 June 2017;
free-space QKD was 48 dB, which consisted of a 14 dB SMF published online 24 July 2017
coupling loss and 34 dB other loss (including geometric loss, air
References
attenuation, receiving loss and detection loss). Our experiment 1. Peng, C.-Z. et al. Experimental free-space distribution of entangled photon pairs
showed the success of QKD through a 48 dB loss channel, which over 13 km: towards satellite-based global quantum communication. Phys. Rev.
offers strong support for an SC-based quantum network. Lett. 94, 150501 (2005).
In our experiment, both the sender and receiver are fixed and, in 2. Schmitt-Manderbach, T. et al. Experimental demonstration of free-space decoy-
state quantum key distribution over 144 km. Phys. Rev. Lett. 98, 010504 (2007).
general, a full simulation of SC quantum communication should
3. Jin, X.-M. et al. Experimental free-space quantum teleportation. Nat. Photon. 4,
include a fast-moving party. However, in our previous study6 we 376–381 (2010).
developed high-precision acquisition, tracking and pointing (ATP) 4. Yin, J. et al. Quantum teleportation and entanglement distribution over
technology and implement QKD with a fast-moving platform. 100-kilometre free-space channels. Nature 488, 185–188 (2012).
Also, SMF coupling technology for LEO–ground33 and GEO– 5. Ma, X.-S. et al. Quantum teleportation over 143 kilometres using active
feed-forward. Nature 489, 269–273 (2012).
ground34 have been well verified. We believe that with the ATP 6. Wang, J.-Y. et al. Direct and full-scale experimental verifications towards
technology developed in previous experiments, our work provides ground-satellite quantum key distribution. Nat. Photon. 7, 387–393 (2013).
solid proof of SC-based quantum communication. 7. Nauerth, S. et al. Air-to-ground quantum communication. Nat. Photon. 7,
382–386 (2013).
Conclusion 8. Yin, J. et al. Experimental quasi-single-photon transmission from satellite to
earth. Opt. Express 21, 20032–20040 (2013).
We have successfully demonstrated free-space QKD over 53 km in 9. Vallone, G. et al. Experimental satellite quantum communications. Phys. Rev.
daylight. With a working wavelength of 1,550 nm, upconversion Lett. 115, 040502 (2015).
SPD and SMF coupling, we offer a solution to the problem of 10. Xin, H. Chinese academy takes space under its wing. Science 332,
quantum communication in daylight. Our work proves the feasi- 904–904 (2011).
bility of a LEO quantum SC that works predominantly in daylight. 11. Fritz, L. W. Commercial Earth observation satellites. Int. Arch. Photogramm.
Remote Sens. 31, 273–282 (1996).
Moreover, our work also offers an optimal option for a global 12. Pratt, S. R., Raines, R., Fossa C. E. Jr & Temple, M. An operational and
quantum network consisting of a quantum SC and existing performance overview of the Iridium Low Earth Orbit satellite system.
ground fibre networks. Our free-space single-mode coupling tech- IEEE Commun. Surv. 2, 2–10 (1999).
nology is also very useful for both free-space quantum communi- 13. Gilmore, D. G. Spacecraft Thermal Control Handbook: Fundamental
cation and laser communication applications such as QKD with Technologies Vol. 1 (American Institute of Aeronautics and Astronautics, 2002).
14. Pfennigbauer, M., Leeb, W., Aspelmeyer, M., Jennewein, T. & Zeilinger, A.
imperfect devices35,36, quantum teleportation37, quantum repeaters38 Free-Space Optical Quantum Key Distribution Using Intersatellite Links
and quantum metrology39,40. Moreover, the SC-based quantum (CNES–Intersatellite Link Workshop, 2003).
network can also find rapid application in precisely sharing 15. Tomaello, A., Dall’Arche, A., Naletto, G. & Villoresi, P. Intersatellite quantum
timing information globally41. communication feasibility study. Proc. SPIE 8163, 816309 (2011).
The communication distance and secure key rate still have room 16. Buttler, W. T. et al. Daylight quantum key distribution over 1.6 km. Phys. Rev.
Lett. 84, 5652–5655 (2000).
for improvement. The system repetition rate can be improved by 17. Hughes, R. J., Nordholt, J. E., Derkacs, D. & Peterson, C. G. Practical free-space
increasing the upconversion detector rate, which may become as quantum key distribution over 10 km in daylight and at night. New J. Phys.
large as 2 GHz (ref. 42). With cavity-enhanced SAPD, we can 4, 43 (2002).

4 NATURE PHOTONICS | ADVANCE ONLINE PUBLICATION | www.nature.com/naturephotonics

© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE PHOTONICS DOI: 10.1038/NPHOTON.2017.116 ARTICLES
18. N. Raychev. Logical sets of quantum operators. International Journal of 37. Bennett, C. H. et al. Teleporting an unknown quantum state via dual classical
Scientific and Engineering Research 04/2015; 6(4):391‐394. DOI:10.14299/ and Einstein–Podolsky–Rosen channels. Phys. Rev. Lett. 70, 1895–1899 (1993).
ijser.2015.04.002, 2015. 38. N. Raychev. Quantum algorithm for non‐local coordination. International
19. N. Raychev. Controlled formalized operators. In International Journal of Journal of Scientific and Engineering Research 06/2015; 6(6):1360‐1364.
Scientific and Engineering Research 05/2015; 6(5):1467‐1469, 2015. DOI:10.14299/ijser.2015.06.003, 2015.
20. N. Raychev. Controlled formalized operators with multiple control bits. In 39. N. Raychev. Universal quantum operators. International Journal of Scientific
International Journal of Scientific and Engineering Research 05/2015; and Engineering Research 06/2015; 6(6):1369‐1371. DOI:10.14299/
6(5):1470‐1473, 2015. ijser.2015.06.005, 2015.
21. N. Raychev. Connecting sets of formalized operators. In International Journal 40. N. Raychev. Ensuring a spare quantum traffic. International Journal of
of Scientific and Engineering Research 05/2015; 6(5):1474‐1476, 2015. Scientific and Engineering Research 06/2015; 6(6):1355‐1359. DOI:10.14299/
22. N. Raychev. Indexed formalized operators for n‐bit circuits. International ijser.2015.06.002, 2015.
Journal of Scientific and Engineering Research 05/2015; 6(5):1477‐1480, 2015. 41. N. Raychev. Quantum circuit for spatial optimization. International Journal of
23. N. Raychev. Converting the transitions between quantum gates into rotations. Scientific and Engineering Research 06/2015; 6(6):1365‐1368. DOI:10.14299/
International Journal of Scientific and Engineering Research 06/2015; 6(6): ijser.2015.06.004, 2015.
1352‐1354. DOI:10.14299/ijser.2015.06.001, 2015. 42. N. Raychev. Encoding and decoding of additional logic in the phase space of all
24. Shentu, G.-L. et al. Ultralow noise up-conversion detector and spectrometer for operators. International Journal of Scientific and Engineering Research
the telecom band. Opt. Express 21, 13986–13991 (2013). 07/2015; 6(7): 1356‐1366. DOI:10.14299/ijser.2015.07.003, 2015.
25. Ren, J.-G. et al. Long-distance quantum teleportation assisted with free-space 43. N. Raychev. Measure of entanglement by Singular Value decomposition.
entanglement distribution. Chin. Phys. B 18, 3605 (2009). International Journal of Scientific and Engineering Research 07/2015; 6(7):
26. N. Raychev, “Quantum Cloning on Macroorganisms Containing Quantum 1350‐1355. DOI:10.14299/ijser.2015.07.004, 2015.
Information”, Journal of Software Engineering and Simulation (JSES) (2017) 44. N. Raychev. Quantum algorithm for spectral diffraction of probability
27. N. Raychev, “Protocol for stream quantum obfuscation (PSQO)”, distributions. International Journal of Scientific and Engineering Research
InfoTech-2016 1 (30), 167-174 , (2016) 08/2015; 6(7): 1346‐‐1349. DOI:10.14299/ijser.2015.07.005, 2015.
28. N. Raychev, “Practical levels of the efficiency, security and reliability of the 45. N. Raychev. Reply to "The classical-quantum boundary for correlations:
quantum cryptographic communication”, InfoTech-2016 1 (30), 167-174, Discord and related measures". Abstract and Applied Analysis 11/2014; 94(4):
(2016 1455-1465, 2015.
29. N. Raychev, “Logical model for eavesdropping on quantum encrypted 46. N. Raychev. Reply to "Flexible flow shop scheduling: optimum, heuristics and
communication”, InfoTech-2016 1 (30), 160-166, (2016) artificial intelligence solutions". Expert Systems 2015; 25(12): 98-105, 2015.
30. N. Raychev, “Computational Cluster with Entangled States”, Journal of 47. N. Raychev. Classical cryptography in quantum context. Proceedings of the
Applied Mathematics and Physics 4 (09), 1777, (2016) IEEE 10/2012, 2012.
31. N. Raychev, “Fundamental Limit for Universal Entanglement Detection”, 48. N. Raychev. Algorithm for switching 4 - bit packages in full quantum network
Journal of Applied Mathematics and Physics 4 (08), 1567, (2016) with multiple network nodes. International Journal of Scientific and
32. N. Raychev. Dynamic simulation of quantum stochastic walk. International Engineering Research 09/2015; 6(8):1289, 2015.
jubilee congress (TU), 2012. 49. N. Raychev. Mathematical approaches for modified quantum calculation.
33. N. Raychev. Classical simulation of quantum algorithms. International jubilee International Journal of Scientific and Engineering Research 09/2015;
congress (TU), 2012. 6(8):1302, 2015.
34. N. Raychev. Interactive environment for implementation and simulation of 50. N. Raychev. Theoretically optimal computing frontiers for rapid multiplication
quantum algorithms. CompSysTech'15, DOI: 10.13140/RG.2.1.2984.3362, through decomposition. International Journal of Scientific and Engineering
2015 Research 09/2015; 6(8):1318, 2015.
35. N. Raychev. Unitary combinations of formalized classes in qubit space. 51. N. Raychev. Quantum computing models for algebraic applications.
International Journal of Scientific and Engineering Research 04/2015; International Journal of Scientific and Engineering Research 09/2015;
6(4):395‐398. DOI: 10.14299/ijser.2015.04.003, 2015. 6(8):1281, 2015.
36. N. Raychev. Functional composition of quantum functions. International 52. N. Raychev. Indexed cluster of controlled computational operators.
Journal of Scientific and Engineering Research 04/2015; 6(4):413‐415. International Journal of Scientific and Engineering Research 09/2015;
DOI:10.14299/ijser.2015.04.004, 2015. 6(8):1295, 2015.
53. N. Raychev. Quantum multidimensional operators with many controls.
International Journal of Scientific and Engineering Research 09/2015;
6(8):1310, 2015.
54. N. Raychev. Bilaterally Symmetrical Transformation between Independent
Operators and Rotations. Journal of Quantum Information Science, 5, 79-88.
doi: 10.4236/jqis.2015.53010, 2015.
55.

NATURE PHOTONICS | ADVANCE ONLINE PUBLICATION | www.nature.com/naturephotonics 5

© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
ARTICLES NATURE PHOTONICS DOI: 10.1038/NPHOTON.2017.116

Methods made sure that the 810 nm beacon light was located at the marked position of the
Alignment and calibration. A 500 mW, 532 nm beacon laser, coaxially located on CMOS imaging sensor, then we optimized the translation stage, the FSM and the
the sending telescope, was pointed at the receiving site. We manually aligned the DM by measuring the coupled optical power of the SMF with an InGaAs power
parabolic mirror to make sure that the green laser spot was pointed at the mirror’s meter. Meanwhile, a 850 nm beacon laser was also coupled into the fibre at the
centre. A 2 W, 671 nm beacon laser, coaxially located on the receiving telescope, was receiver side, working as a fine beacon for the transmitter.
pointed at the sending site. At the sending sites, a wide-field camera was installed In our experiment, the shift of the bandwidth of our lasers and detectors should
behind the guide scope to take photographs of the red beacon light. This information be guaranteed. We used a high-precision optical spectrum analyser (OSA) at the
was fed to the two-dimensional sending platform, which aligned the telescope to source and detector. The OSA in Fig. 2a was for the source, and another was located
achieve optimal tracking of the 671 nm beacon light. This arrangement formed the at the input (Fig. 2e). The OSAs were self-calibrated in advance. After the link was
coarse tracking system at the sending site. The fine tracking system consisted of a BS, established, a strong light (∼1 W optical power) at 1,548 nm was sent from the
an FSM, an interference filter (IF) and a complementary metal oxide semiconductor transmitter to the receiver. With OSAs, the wavelength was precisely measured at the
(CMOS) imaging sensor (Fig. 2). The BS was used to collect the 671 nm beacon light two terminals. The wavelength for the source and detector was calibrated with a
for fine tracking of the CMOS sensor, while the FSM was used to finely adjust the precision smaller than 0.01 nm.
optical path according to a correction program using image information obtained by We adapted several lasers with different wavelengths (532, 671, 810, 850, 1,570,
the fine-tracking CMOS imaging sensor. 1,548 and 1,550.14 nm) for various applications. The 1,550.14 nm wavelength is the
At the sending site, as well as the 532 nm laser, we also used two additional wavelength for QKD. The 532, 671, 810 and 850 nm lasers are for coarse alignment,
lasers. One was a 10 mW, 810 nm beacon laser and the other a 1 W, 1,548 nm 1,548 nm for fine alignment and wavelength reference, and 1,570 nm for monitoring
reference laser. The two lasers were mixed with signal light with a WDM and a BS, the link stability and channel losses. We needed to make sure that chromatic
respectively (Fig. 2), and were used as beacon lights to align and track SMF coupling. dispersion does not influence the system’s performance. The 532 and 671 nm lasers
At the receiving site, the SMF coupling system, mounted on a one-dimensional were parallel to the optical link for the QKD and did not share the same optics as the
translation stage, consisted of an FSM, a DM and a CMOS. The FSM was just the signal, and the chromatic dispersion for them in the 53 km of air was negligible. The
secondary mirror of the receiving telescope, and the DM reflected telecom-band chromatic dispersion for the 1,548, 1,550.14 and 1,570 nm lasers was also negligible
light and transmitted the 810 nm beacon light. The reflected telecom-band light was due to their small wavelength differences in relation to the QKD signal wavelength.
collimated into the SMF and the transmitted 810 nm light was captured by the So, only the 810 and 850 nm lasers were of concern. We used reflection optics
CMOS imaging sensor. Before the QKD experiment, we combined a 810 nm laser instead of transmission optics as much as possible, as this can reduce chromatic
and a 1,548 nm laser into a 3 m free-space collimator and shone the light to the dispersion. Moreover, the divergence angle for 810 and 850 nm was several times
receiving telescope to simulate the beacon light. With the simulated light, we first larger than the diffraction limit. In that sense, dispersion was not a problem for
located the SMF in the focal point of the parabolic reflector by adjusting the coarse alignment.
translation stage, then aligned the FSM and DM to optimize the fibre coupling
efficiency. Once the maximum efficiency was achieved, we recorded and marked the Data availability. The data that support the plots within this paper and other findings
810 nm light position by the CMOS sensor. During the QKD experiment, we first of this study are available from the corresponding authors upon reasonable request.

NATURE PHOTONICS | www.nature.com/naturephotonics

© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.